USOO887.4477B2

(12) United States Patent (10) Patent No.: US 8,874.477 B2 Hoffberg (45) Date of Patent: Oct. 28, 2014

(54) MULTIFACTORIAL OPTIMIZATION (56) References Cited SYSTEMAND METHOD U.S. PATENT DOCUMENTS (76) Inventor: Steven Mark Hoffberg, West Harrison, NY (US) 3,573,747 A 4, 1971 Adams et al. 3,581,072 A 5/1971 Nymeyer (*) Notice: Subject to any disclaimer, the term of this (Continued) patent is extended or adjusted under 35 U.S.C. 154(b) by 2050 days. FOREIGN PATENT DOCUMENTS AU 1997.036815 10, 1997 (21) Appl. No.: 11/467,931 AU 1997.036816 10, 1997 (22) Filed: Aug. 29, 2006 (Continued) (65) Prior Publication Data OTHER PUBLICATIONS US 2007/0O87756A1 Apr. 19, 2007 Buttyan, Levente and Hubaux, Jean-Pierre, "Nuglets: a Virtual Cur rency to Stimulate Cooperation in Self-Organized Mobile Ad Hoc Networks', Jan. 18, 2001.* Related U.S. Application Data (Continued) (60) Provisional application No. 60/723,339, filed on Oct. 4, 2005. Primary Examiner — Eric T Wong (51) Int. Cl. (74) Attorney, Agent, or Firm — Steven M. Hoffberg;

G06F 9/50 (2006.01) G06O 50/18 (2012.01) (57) ABSTRACT G06O40/04G06O 10/10 (2012.01) A method for providing- 0 unequal allocation of rights among G06O 30/02 (2012.01) agents while operating according to fair principles, compris G06O 10/06 (2012.01) ing assigning a hierarchal rank to each agent; providing a G06O 30/08 (2012.01) synthetic economic value to a first set of agents at the a high G06O20/22 (2012.01) level of the hierarchy; allocating portions of the synthetic HO4W 84/18 (2009.01) economic value by the first set of agents to a second set of (52) U.S. Cl agents at respectively different hierarchal rank than the first CPC G06O40/00 (2013.01); G06Q50/188 set of agents; and conducting an auction amongst agents ------(201 3.01); G06O 10/1 03 (30 3.01); G06O using the synthetic economic value as the currency. A method 40/04 (2013 01); G06O 3002.01 (2013 01); for allocation among agents, comprising assigning a wealth G06O 10/0635 (2013.01); H04W 84/18 generation function for generating future wealth to each of a (2013.01); G06Q30/08 (2013.01). G06O20/22 plurality of agents, communicating Subjective market infor (2013 6 1); G06O 10/06375 (201 3.01); G06O mation between agents, and transferring wealth generated by • u. fs IO/0631 (301 3.01) the secure wealth generation function between agents in con USPC T05/37: 718f O4 sideration of a market transaction. The method may further (58) Field fo - - - - - ificati------s ------h------s comprise the step of transferring at least a portion of the s OSSO ear 6Q 40/06; G06F 9/46 wealth generation function between agents. USPC ...... 705/35; 455/450, 451; 718/104 See application file for complete search history. 64 Claims, 6 Drawing Sheets US 8,874.477 B2 Page 2

(56) References Cited 5,911, 136 6, 1999 Atkins 5,915,019 6, 1999 Ginter et al. U.S. PATENT DOCUMENTS 5,915,209 6, 1999 Lawrence 5,920,861 7, 1999 Hall et al. 4.486,853 12, 1984 Parsons 5,924,082 7, 1999 Silverman et al. 4,672.572 6, 1987 Alsberg 5,924,083 7, 1999 Silverman et al. 4,789.928 12, 1988 Fujisaki 5,926,796 7, 1999 Walker et al. 4,789,929 12, 1988 Nishimura et al. 5,940,504 8, 1999 Griswold 4,799,156 1, 1989 Shavit et al. 5,941,947 8, 1999 Brown et al. 4,823,264 4, 1989 Deming 5,943,422 8, 1999 VanWie et al. 4,827,508 5, 1989 Shear 5.948,040 9, 1999 DeLorime et al. 4,881,178 11, 1989 Holland et al. 5,953,319 9, 1999 Dutta et al. 4,887,818 12, 1989 Escott 5,970.479 10, 1999 Shepherd 4,903,201 2, 1990 Wagner 5,983,205 11, 1999 Brams et al. 4,930,073 5, 1990 Cina, Jr. 6,012,045 1, 2000 Barzilai et al. 5,016, 170 5, 1991 Pollalis et al. 6,014,627 1, 2000 Togher et al. 5,070,453 12, 1991 Duffany 6,014,643 1, 2000 Minton 5,077,665 12, 1991 Silverman et al. 6,021,397 2, 2000 Jones et al. 5,103,476 4, 1992 Waite et al. 6,021,398 2, 2000 Ausubel 5,111,390 5, 1992 Ketcham 6,023,685 2, 2000 Brett et al. 5,136,501 8, 1992 Silverman et al. 6,023,686 2, 2000 Brown 5,224,163 6, 1993 Gasser et al. 6,026,383 2, 2000 Ausubel 5,235,680 8, 1993 Bijnagte 6,044,363 3, 2000 Mori et al. 5,253,165 10, 1993 Leiseca et al. 6,047,264 4, 2000 Fisher et al. 5,270,921 12, 1993 Hornick 6,047,274 4, 2000 Johnson et al. 5,275.400 1, 1994 Weingardt et al. 6,055,518 4, 2000 Franklin et al. 5,283,731 2, 1994 Lalonde et al. 6,058,379 5/2000 Odom et al. 5,319,705 6, 1994 Halter et al. 6,085, 175 T/2000 Gugel et al. 5,343,527 8, 1994 Moore 6,098,051 8, 2000 Lupien et al. 5,347,452 9, 1994 Bay, Jr. 6,112, 181 8, 2000 Shear et al. 5,377,095 12, 1994 Maeda et al. 6,131,087 10, 2000 Luke et al. 5,390,330 2, 1995 Talati 6,134,536 10, 2000 Shepherd 5,394,324 2, 1995 Clearwater 6,141,423 10, 2000 Fischer 5,459,761 10, 1995 Monica et al. 6,151,309 11, 2000 Busuioc et al. 5,491,800 2, 1996 Goldsmith et al. 6,151,589 11, 2000 Aggarwal et al. 5,530,235 6, 1996 Stefik et al. 6,157,721 12, 2000 Shear et al. 5,534,975 7, 1996 Stefket al. 6,161,099 12, 2000 Harrington et al. 5,573,244 11, 1996 Mindes 6,163,701 12, 2000 Saleh et al. 5,615, 109 3, 1997 Eder 6,167,386 12, 2000 Brown 5,629,980 5, 1997 Stefik et al. 6,169,789 1, 2001 Rao et al. 5,634,012 5, 1997 Stefik et al. 6,175,831 1, 2001 Weinreich et al. 5,634,055 5, 1997 Barnewall et al. 6,185,683 2, 2001 Ginter et al. 5,638.443 6, 1997 Stefik et al. 6,199,050 3, 2001 Alaia et al. 5,640,546 6, 1997 Gopinath et al. 6,226,618 5/2001 Downs et al. 5,640,569 6, 1997 Miller et al. 6,230,146 5/2001 Alaia et al. 5,664,115 9, 1997 Fraser 6,236,977 5/2001 Verba et al. 5,672, 106 9, 1997 Orford et al. 6,240,411 5/2001 Thearling 5,689,652 11, 1997 Lupien et al. 6,243,691 6, 2001 Fisher et al. 5,692,047 11, 1997 McManis 6,249,873 6, 2001 Richard et al. 5,715.403 2, 1998 Stefk 6,260,024 T/2001 Shkedy 5,724.488 3, 1998 Prezioso 6,263,334 T/2001 Fayyad et al. 5,727,165 3, 1998 Ordish et al. 6,266,652 T/2001 Godin et al. 5,745,678 4, 1998 Herzberg et al. 6,272.473 8, 2001 Sandholm 5,748,960 5, 1998 Fischer 6,272.474 8, 2001 Garcia 5,749,785 5, 1998 Rossides 6,321,212 11, 2001 Lange 5,757,914 5, 1998 McManis 6,324,519 11, 2001 Eldering 5,758,328 5, 1998 Giovannoli 6,330,551 12, 2001 Burchetta et al. 5,778,367 7, 1998 Wesinger, Jr. et al. 6,331,986 12, 2001 Mitra et al. 5,794,207 8, 1998 Walker et al. 6,360,222 3, 2002 Quinn 5,797,127 8, 1998 Walker et al. 6,366,907 4, 2002 Fanning et al. 5,806,048 9, 1998 Kiron et al. 6,377,809 4, 2002 Rezaiifar et al. 5,809,144 9, 1998 Sirbu et al. 6,393,276 5/2002 Vanghi 5,819,237 10, 1998 Garman 6,398,245 6, 2002 Gruse et al. 5,826,244 10, 1998 Huberman 6,415, 151 T/2002 Kreppel 5,831,545 11, 1998 Murray et al. 6,430,537 8, 2002 Tedesco et al. 5,835,896 11, 1998 Fisher et al. 6,443,841 9, 2002 Rossides 5,839,114 11, 1998 Lynch et al. 6,449,612 9, 2002 Bradley et al. 5,845,211 12, 1998 Roach, Jr. 6.459,881 10, 2002 Hoder et al. 5,845,266 12, 1998 Lupien et al. 6,466,977 10, 2002 Sitaraman et al. 5,873,068 2, 1999 Beaumont et al. 6,480,587 11, 2002 Rao et al. 5,873,071 2, 1999 Ferstenberg et al. 6,487,541 11, 2002 Aggarwal et al. 5,873,782 2, 1999 Hall 6,499,018 12, 2002 Alaia et al. 5,884,272 3, 1999 Walker et al. 6,501,765 12, 2002 Lu et al. 5,889,799 3, 1999 Grossman et al. 6,522,946 2, 2003 Weis 5,890,138 3, 1999 Godin et al. 6,530,537 3, 2003 Hanlon 5,892,900 4, 1999 Ginter et al. 6,542,742 4/2003 Schramm et al. 5,897,620 4, 1999 Walker et al. 6,556.951 4/2003 Deleo et al. 5,905,975 5, 1999 Ausubel 6,560,580 5/2003 Fraser et al. 5,910,987 6, 1999 Ginter et al. 6,564,192 5/2003 Kinney, Jr. et al. US 8,874.477 B2 Page 3

(56) References Cited 7,698,335 B1 4/2010 Vronay 7,707,118 B2 4/2010 James U.S. PATENT DOCUMENTS 7,711,808 B2 5/2010 Parry 7,725,414 B2 5/2010 Nigam et al. 6,587,837 B1 7/2003 Spagna et al. 7,729,975 B2 6/2010 Ausubel et al. 6,591.232 B1 7/2003 Kassapoglou 7,730,120 B2 6/2010 Singh et al. 6,601,036 B1 7/2003 Walker et al. 7,739,335 B2 6/2010 Siegel et al. 6,606,607 B1 8/2003 Martin et al. 7,742,972 B2 6/2010 Lange et al. 6,609,112 B1 8/2003 Boarman et al. 7,743,163 B2 6/2010 Ruppert 6,611,812 B2 8, 2003 Hurtado et al. 7,751,423 B2 7/2010 Hottinen et al. 6,629,082 B1 9/2003 Hambrecht et al. 7,752,064 B2 7/2010 Kauffman 6,639,898 B1 10/2003 Dutta et al. 7,774.264 B1 8/2010 Ausubel 6,647,373 B1 1 1/2003 Carlton-Foss 7,797,732 B2 9/2010 Tam et al. 6,658.467 B1 12/2003 Rice et al. 7,801,802 B2 9/2010 Walker et al. 6,661,379 B2 12/2003 Stilp et al. 7,808.922 B2 10/2010 Dekorsy 6,662,141 B2 12/2003 Kaub 7,827,128 B1 1 1/2010 Karlsson et al. 6,678,245 B1 1/2004 Cooper et al. 7,843,822 B1 1 1/2010 Paul et al. 6,684,250 B2 1/2004 Anderson et al. 7,844,535 B2 11/2010 Guler et al. 6,704,713 B1 3, 2004 Brett 7,890,581 B2 2/2011 Rao et al. 6,712,701 B1 3/2004 Boylan, III et al. 7,933,829 B2 4/2011 Goldberg et al. 6,718,312 B1 4/2004 McAfee et al. 7,962.346 B2 6/2011 Faltings 6,754,169 B2 6, 2004 Baum et al. 2001/0027431 A1 10/2001 Rupp et al. 6,766,307 B1 7/2004 Israel et al. 2001/0032164 A1 10, 2001 Kim 6,785,671 B1 8/2004 Bailey et al. 2001/0039528 A1 11/2001 Atkinson et al. 6,799,165 B1 9/2004 Boeses 2001/0042785 A1 11, 2001 Walker et al. 6,834,110 B1 12/2004 Marconcini et al. 2001/0047291 A1 11/2001 Garahi et al. 6.834.272 B1 12/2004 Naor et al. 2001/0049650 A1 12/2001 Moshal et al. 6,842,463 B1 1/2005 Drwiega et al. 2001/0051540 A1 12/2001 Hindman et al. 6,847939 B1 1/2005 Shemesh 2002fOO 10669 A1 1/2002 Street 6,865,559 B2 3, 2005 Dutta 2002fOO 10673 A1 1/2002 Muller et al. 6,876,309 B1 4/2005 Lawrence 2002/0013631 A1 1/2002 Parunak et al. 6,926,796 B1 8, 2005 Nishida et al. 2002fOO 13757 A1 1/2002 Bykowsky et al. 6,934,249 B1 8, 2005 Bertin et al. 2002.0035534 A1 3/2002 Buist et al. 6,954,731 B1 10/2005 Montague 2002/004031.0 A1 4/2002 Lieben et al. 6.956,835 B2 10/2005 Tong et al. 2002fOO42769 A1 4/2002 Gujral et al. 6.959,288 B1 10/2005 Medina et al. 2002fOO5281.6 A1 5/2002 Clenaghan et al. 6.968.323 B1 1 1/2005 Bansal et al. 2002O052819 A1 5, 2002 Burton 6,974,412 B2 i2/2005 Dobrovolny 2002/0052873 A1 5/2002 Delgado et al. 6,978,126 B1 12/2005 Blaker et al. 2002fOO55899 A1 5, 2002 Williams 7,003,485 B1 2/2006 Young 2002/00593.79 A1 5/2002 Harvey et al. T.O39,598 B2 5, 2006 Tobin et al. 2002fOO73018 A1 6/2002 Mulinder et al. 7.043,245 B2 5/2006 Dokko 2002fOO73 049 A1 6, 2002 Dutta 7,047,242 B1 5, 2006 Ponte 2002fOO77954 A1 6/2002 Slaight et al. 7.062.461 B1 6, 2006 Ausubel 2002fOO953.27 A1 7/2002 Zumel et al. 706930s B2 6/2006 Abrams 2002/O123954 A1 9, 2002 Hito 7076553 B2 7/2006 Chan et al. 2002fO161671 A1 10, 2002 Matsui et al. 7,085,682 B1 8, 2006 Heller et al. 2002fO165756 A1 11, 2002 Tobin et al. 7,096.197 B2 & 2006 Messmeret al. 2002fO165817 A1 11/2002 Rackson et al. T.O99,839 B2 8, 2006 Madoffet al. 2002/0174052 A1 11/2002 Guler et al. 703.565 B1 9/2006 Vaid 2002/0183066 A1 12/2002 Pankaj 7.130.579 B1 10/2006 Raelet al. 2002/019425.6 A1 12/2002 Needham et al. 7,133,841 B1 1 1/2006 Wurman et al. 2002/0194334 A1 12/2002 Focant et al. 7,142,523 B1 1 1/2006 Chekuri et al. 2003, OO14373 A1 1/2003 Perge et al. 7,165.046 B2 1/2007 Ausubel 2003, OO18561 A1 1/2003 Kitchen et al. 78477 B2 2/2007 Diener et al. 2003.0035468 A1 2/2003 Corbaton et al. 7,200,571 B1 4/2007 Jenniges et al. 2003/0041002 A1 2/2003 Hao et al. 7,246,164 B2 7/2007 Lehmann et al. 2003/0055787 A1 3/2003 Fujii 7,249,027 B1 7/2007 Ausubel 2003/0055898 A1 3/2003 Yeager et al. 7.249,085 B1 7/2007 Kinney, Jr. et al. 2003/0065608 Al 42003 Cutler T.251,589 B1 7/2007 Crowe et al. 2003/0O87652 A1 5/2003 Simon et al. 7.287.007 B1 10/2007 Detering 2003/0088488 A1 5/2003 Solomon et al. 7,319,973 B1 1/2008 Tobin et al. 2003/00973.25 A1 5/2003 Friesen et al. 7,330,826 B1 2/2008 Porat et al. 2003/0101.124 A1 5/2003 Semret et al...... 705/37 T.337,139 B1 2/2008 Ausubel 2003/0101274 A1 5, 2003 Yi et al. 7343.342 B2 3/2008 Ausubel 2003/0115114 A1 6/2003 Tateishi et al. 7,349,827 B1 3/2008 Heller et al. 2003.01.15251 A1 6/2003 Fredrickson et al. 7,415,436 B1 8/2008 Evelyn et al. 2003/O135437 A1 7/2003 Jacobsen 7,457.823 B2 11/2008 Shraim et al. 2003/0139995 A1 7/2003 Farley 7.460,065 B2 12/2008 Ogawa et al. 2003. O195780 A1 10, 2003 Arora et al. 7.464,057 B2 12/2008 Cole et al. 2003/0217106 A1 11/2003 Adar et al. 7,516,089 B1 4/2009 Walker et al. 2003/0233307 A1 12/2003 Salvadori et al. 7,523,085 B2 4/2009 Nigam et al. 2004.0006528 A1 1/2004 Kevin Fung 7,559,070 B2 7/2009 Nakamura et al. 2004/OOO6529 A1 1/2004 Fung 7,596,552 B2 9/2009 Levy et al. 2004/OOO6534 A1 1/2004 Fung 7,600,017 B2 10/2009 Holtzman et al. 2004.0024684 A1 2/2004 Montepeque 7,627,510 B2 12/2009 Jain et al. 2004.0024687 A1 2/2004 Delenda ...... 705/37 7,660,783 B2 2/2010 Reed 2004/0039670 A1 2/2004 Fung 7,676.423 B2 3/2010 Avery 2004/0039685 A1 2/2004 Hambrecht et al. 7,693,939 B2 4/2010 Wu et al. 2004.0054551 A1 3/2004 Ausubel et al. US 8,874.477 B2 Page 4

(56) References Cited WO WO9845768 10, 1998 WO WO990.1815 1, 1999 U.S. PATENT DOCUMENTS WO WO9924928 5, 1999 WO WO9927476 6, 1999 2004/0054617 A1 3/2004 Fung WO WO994.8296 9, 1999 2004/0059665 A1 3, 2004 Suri et al. WO WOO 108063 2, 2001 2004/0068447 A1 4/2004 Mao et al. WO WOO 108072 2, 2001 2004/0093278 A1 5/2004 Burchetta et al. WO WOO11.0076 2, 2001 2004/011 1308 A1* 6/2004 Yakov ...... 70.5/8 WO WOO154335 T 2001 2004/01 17302 A1 6/2004 Weichert et al. WO WOO2O7031 1, 2002 2004/O132405 A1 7/2004 Kitazawa et al. WO WOO291 100 11, 2002 2004/0153375 A1 8/2004 Mukunya et al. WO WO2006O47879 5, 2006 2004/O184478 A1 9/2004 Donescu et al. 2004/0215793 A1 10/2004 Ryan et al. OTHER PUBLICATIONS 2004/0259558 A1 12/2004 Skafidas et al. 2005, OO44032 A1 2/2005 Lee et al. Messer, Alan et al. "A Market Model for Resource Allocation in 2005, 0080710 A1 4/2005 Malato et al. Distributed Operating Systems”, 1995.* 2005/O137939 A1 6/2005 Calabria et al. Messer, Alan, and Tim Wilkinson. Dynamic Locality for Resource 20050144064 Al 6/2005 Calabria et al. Information Dissemination. Technical report, City University, Lon 3988: A. $398 Sabia et al. don, 1997. (Self-citation of 2 A. Messer, T. Wilkinson. A Market 2005, 0198031 A1 9, 2005 Pezaris et al. Model for Resource Allocation in Distributed Operating Systems. 2005/0246420 A1 1 1/2005 Little, II Technical Report, Department of Computer Science, City University. 2005/0286426 A1 12/2005 Padhye et al. Feb. 1995. -- - -- 2005/0289043 A1 12, 2005 Maudlin Messer, Alan, and Tim Wilkinson. A Scalable Proportion-share 2006, 0046658 A1 3, 2006 Cruz et al. Resource Allocator for Distributed Systems, tech. rep., SARC, 2006/01484.14 A1 7/2006 Tee et al. Department of Computer Science, City University, England, 1996. 2006.0167787 A1 7/2006 Ausubel Kara, Mourad. A global plan policy for coherent co-operation in 2006, O1681 19 A1 7/2006 Inoue et al. distributed dynamic load balancing algorithms.” Distributed Systems 2006, O16814.0 A1 7/2006 Inoue et al. Engineering 2.4 (1995): 212. 2006, O168147 A1 7/2006 Inoue et al. Messer, Alan, and Tim Wilkinson. “Power to the process.” (1996). 2006, O259957 A1 11, 2006 Tam et al. Anastasiadi, A., et al. “A computational economy for dynamic load 3829 18463 A1 39. Avery balancing and data replication.” Proceedings of the first international 38.'8: A. 29. NE conference on Information and computation economies. ACM, 1998. 2007/0174179 A1 7/2007 Avery Kuwabara, Kazuhiro, and Toru Ishida. “Equilibratory approach to 2007/O195048 A1 8, 2007 Nam distributed resource allocation: Toward coordinated balancing.” Arti 2007,0297.328 A1 12, 2007 Semret et al. ficial Social Systems. Springer Berlin Heidelberg, 1994. 133-146. 2008/009.5121 A1 4/2008 Shattil Rasmusson, Lars. "Evaluating resource bundle derivatives for multi 2008.0109343 A1 5/2008 Robinson et al...... 705/37 agent negotiation of resource allocation.” E-Commerce Agents. 2008. O162331 A1 7/2008 Ephrati et al. Springer Berlin Heidelberg, 2001. 154-165. 2008. O162666 A1 7/2008 Ebihara et al. Huberman, Bernardo A., and Tad Hogg. “The emergence of compu 2008, 0207149 A1 8, 2008 Unkefer et al. tational ecologies.” Lectures in complex systems (1992). 2008/02274.04 A1 9, 2008 Harel et al. Bandhopadhyay, Gautam, Shalini Gupta, and Pramod Mane. 2008, 0299923 A1 12, 2008 O’Brien et al. “Recovery of two process deadlock using payoff matrix approach.” 2009 OO54018 A1 2/2009 Waheed et al. Computer Science and Information Technology (ICCSIT), 2010 3rd 2009/017O607 A1 7/2009 Chiao et al. IEEE International Conference on. vol. 9. IEEE, 2010. 2009,0180392 A1 7/2009 Greiner et al. Anastasiadi, A., et al. “Economic Models for Resource Allocation and Services in Distributed Information Systems.” (1997). FOREIGN PATENT DOCUMENTS Deng, Xiaotie, Li-Sha Huang, and Minming Li. "On Walrasian price of cpu time.” Computing and Combinatorics. Springer Berlin AU 1997.036840 10, 1997 Heidelberg, 2005. 586-595. EP O128672 12, 1984 Ferguson, Donald F., et al. “Economic models for allocating EP O399822 11, 1990 resources in computer systems.” Market-based control: a paradigm EP O421409 4f1991 for distributed resource allocation (1996): 156-183. EP O565314 10, 1993 Chun, Brent N., and David E. Culler. “User-centric performance EP O715247 6, 1996 analysis of market-based cluster batch schedulers.” Cluster Comput E. g 3.i 1 o' g ing and the Grid, 2002. 2nd IEEE/ACM International Symposium on. EP 1054336 11, 2000 E. 2002. - - - 66 GB 2264,796 9, 1993 ow, Steven, and Pravin Varaiya. An algorithm for optimal service JP 2001019496 1, 2001 provisioning using resource pricing.” INFOCOM'94. Networking JP 2002342684 11, 2002 for Global Communications., 13th Proceedings IEEE IEEE, 1994. RU 2161819 1, 2001 Shum, Kam Hong, and Muslim Bozyigit. "A load distribution RU OO17642 4/2001 through competition for workstation clusters.” Proc. of the Ninth WO WO9002382 3, 1990 International Symposium on Computer and Information Sciences. WO WO922.2870 12/1992 1994. WO WO93O1550 1, 1993 Hogg, Tad, and Bernardo A. Huberman. "Dynamics of large autono WO WO9403859 2, 1994 mous computational systems.” Collectives and the design of complex WO WO9406,103 3, 1994 systems. Springer New York, 2004. 295-315. W. W85. 88: Ferguson, Donald. “An Economy for Managing Replicated Data in WO WO973.7315 10, 1997 Autonomous Decentralized Systems (Troceedings of International WO WO9743761 11, 1997 Symposium on Autonomous and Decentralized Systems, prrs) WO WO9809144 3, 1998 Donald Ferguson Christos Nikolaou Yechiam Yemini.” WO WO9809209 3, 1998 Ferguson, Donald, Christos Nikolaou, and Yechiam Yemini. WO WO9810381 3, 1998 “Microeconomic Algorithms for Flow Control in Virtual Circuit Net WO WO9837481 8, 1998 works (Subset in Infocom 1989).” (1995). US 8,874.477 B2 Page 5

(56) References Cited cation network control.” Market-based control: A paradigm for dis tributed resource allocation (1996): 53-73. OTHER PUBLICATIONS Ishida, Toru, Yoshiyasu Nishibe, and Tatsuya Suda, “Kazuhiro Kuwabara NTT Communication Science Laboratories 2-2 Hikaridai, Byde, Andrew. A comparison between mechanisms for sequential Seika-cho, Soraku-gun, Kyoto 619-02, Japan E-mail: kuwabara(a) compute resource auctions. Springer Berlin Heidelberg, 2008. cslab. kecl. ntt.jp.” Market-based Control: A Paradigm for Distrib Ekbom, Kristofer, and Eric Astor. “Multi-Agent Systems in Compu uted Resource Allocation (1996): 53. tational Markets and Ecosystems.” Organization 1103 (1996): 1581. Kropp, Sebastian, and V. I. C. Caulfield. “Distributed Operating Yamaki, Hirofumi. Yutaka Yamauchiy, and Toru Ishiday. “Market Systems.” (2004). (Cites 3 A. Messer and T. Wilkinson. A market Based Application QoS Control: Implementation Tradeoffs.” (1999). model for resource allocation in distributed operating systems, Youssefimir, Michael, and Bernardo A. Huberman. “Resource Con 1995). tention in Multiagent Systems.” ICMAS. 1995. The references herein cited include bibloiographical information Guenther, Oliver, Tad Hogg, and Bernardo A. Huberman. “Market from Google Scholar, and Applicant does not warrant the accuracy of organizations for controlling Smart matter.” Simulating Social Phe that information, including but not limited to alleged publication nomena. Springer Berlin Heidelberg, 1997. 241-257. date. Kuwabara, Kazuhiro, et al. An equilibratory market-based approach for distributed resource allocation and its applications to communi * cited by examiner U.S. Patent Oct. 28, 2014 Sheet 1 of 6 US 8,874.477 B2

Fig. 1

U.S. Patent Oct. 28, 2014 Sheet 2 of 6 US 8,874.477 B2

HMM with mixture of Gaussians output Fig. 3

Input-output HMM FaCtOria HMM Coupled HMM Fig. 4A Fig. 4B Fig. 4C U.S. Patent Oct. 28, 2014 Sheet 3 of 6 US 8,874.477 B2

u- N. Time Update Measurement Update ("Predict") ("Correct")

Fig. 5

Time Update ("Predict") Measurement Update ("Correct") (1) Project the state ahead (1) Compute the Kalman Gain * = Ax1 + Buk K = P. H. (HP, H+R) (2) Project the error covariance ahead (2) Update estimate with P = AP A+ Q measurement I X Sk H K (Ik- HS) (3) Update the error Covariance P = (1 - KH) P.

Initial estimates for X-1 and P Fig. 6 U.S. Patent Oct. 28, 2014 Sheet 4 of 6 US 8,874.477 B2

Time Update ("Predict”) Measurement Update ("Correct") (1) Project the state ahead (1) Compute the Kalman Gain kM = f(x-1,M u, 0) Kk = P. H.H' (H.P. H.H' + V.R.V.) (2) Update estimate with (2) Project the error covariance ahead measurement I P k = AP.KI k-1 AM k + W.Q.,kV|(-1 W.WW k X = x + K (Ik - h(x, 0))

(3) Update the error Covariance P (1 - KH) P

Initial estimates for X-1 and P-1

Fig. 7

U.S. Patent Oct. 28, 2014 Sheet 6 of 6 US 8,874.477 B2

Proximity 101

W1

Prospective Conjunction 102 W2

Type of event 103

W3

Type of event and Sensed COndition 104 W4 Fig. 9

High Memory dump Unit ID LOCation COCes priority Itinerary string 201 2O2 203 messages 205 2O6 204

Event 301 LOCation 302 Time 303 SOUrCe 304 Expiration 305 Reliability 306 Message 307 Fig. 10 US 8,874,477 B2 1. 2 MULTIFACTORIAL OPTIMIZATION circuit switched technologies, such as ISDN, especially SYSTEMAND METHOD where the high quality needs are intermittent. Internet usage typically involves an Internet server, an PRIORITY CLAIM automated system capable of responding to communications received through the Internet, and often communicating with The present application is a Non-Provisional application of other systems not directly connected to the Internet. The 60/723,339 filed Oct. 4, 2005. server typically has relatively large bandwidth to the Internet, allowing multiple simultaneous communications sessions, FIELD OF THE INVENTION and usually supports the hypertext transport protocol 10 (HTTP), which provides, in conjunction with a so-called web The present invention relates to the field of multifactorial browser on a remote client system, a human readable inter economic optimization, and more generally to optimization face which facilitates navigation of various resources avail of communities of elements having conflicting requirements able in the Internet. The client systems are typically human and overlapping resources. user interfaces, which employ a browser to display HTTP 15 “web pages”. The browser typically does not provide intelli gence, although so-called applets or local servers may pro BACKGROUND OF THE INVENTION vide programmable intelligence or processing capability proximate to the user. Bandwidth between the client and A number of fields of endeavor are relevant to the present Internet is typically relatively small, and various communi invention, and exemplary prior art, each of which is expressly cations and display rendering considered normal. Typically, incorporated herein by reference, are disclosed below. The both client and server are connected to the Internet through references disclosed provide a skilled artisan with disclosure Internet Service providers, each having its own router. It is of embodiments of various elements of the present invention, also known to provide so-called proxy servers and firewalls, and the teachings therein may be combined and Subcombined which are automated systems that insulate the client system in various manners in accordance with the present teachings. 25 from the remote server or Internet in general. These local Therefore, the identified prior art provides a point of refer servers, applications and applets may be non-standard, and ence and set of tools which are expressly available, both as a thus require special software to be available locally for execu part of the invention, and to implement the invention. tion. The topical headings are advisory only, and are not Thus, the Internet poses a number of advantages for com intended to limit the applicability of any reference. While 30 mercial use, including low cost and ubiquitous connectivity. Some embodiments are discussed as being preferred, it should Therefore, it is desirable to employ standard Internet tech be understood that all embodiments discussed, in any portion nologies while achieving sufficient quality communications of this documents, whether stated as having advantages or to effect an efficient transaction. not, form a part of the invention and may be combined and/or A widely dispersed network of access points may imple Subcombined in a consistent manner in accordance with the 35 ment a mobile telecommunications protocol, such as IETF teachings hereof. Likewise, the disclosure herein is intended RFC 3344 (Mobile IP, IPv4), or various mobile ad hoc net to disclose permissive combinations, Subcombinations, and work (MANET) protocols, 2.5G, 3G, 4G cellular, WiMax attributes, and any language which appears to limit the scope (802.16), or other types of protocols. Preferably, the protocol of applicants invention is intended to apply to the particular allows the client to maintain communications between a embodiment referenced, or as a permissive suggestion for 40 Source and destination while the network topology or routing implementation of other embodiments which with it may be path changes, such as in a mobile multihopadhoc network or consistently applied. The present disclosure includes details mobile node within a cellular or hotspot network. See, U.S. of a number of aspects, which may find independent utility, Pub. App. No. 20040073642, expressly incorporated herein and therefore the present specifications are not intended to be by reference. construed as being limited to the conjunction of the elements 45 Mobile Internet Protocol (Mobile IP or MIP, in this case, of the disclosure. V4) is an Internet Engineering Task Force (IETF) network Internet layer protocol, specified in RFC-3344. It is designed to allow The Internet is structured such various networks are inter seamless connectivity session maintenance under TCP connected, with communications effected by addressed pack (Transmission Control Protocol) or other connection oriented ets conforming to a common protocol. Based on the packet 50 transport protocols when a mobile node moves from one IP addressing, information is routed from source to destination, subnet to another. MIPv4 uses two network infrastructure often through a set of networks having multiple potential entities, a Home Agent (HA) and an optional Foreign Agent pathways. The communications medium is shared between (FA), to deliver packets to the mobile node when it has left its all users. Statistically, some proportion of the packets are home network. MIPv4 also supports point-of-attachment extraordinarily delayed, or simply lost. Therefore, protocols 55 Care-of Addresses (CoA) if a FA is unavailable. Mobile IP is involving communications using these packets include error increasingly being deployed for 2.5/3G (2.5 or third genera detection schemes that request a retransmit of required data tion wireless) provider networks and may be deployed in not received within a time window. In the event that the medium and large Enterprise IEEE 802.11-based LANs (Lo network nears capacity or is otherwise Subject to limiting cal Area Networks) with multiple subnets. MIPv4 relies on constraint, the incidence of delayed or lost packets increases, 60 the use of permanently assigned “home’ IP addresses to help thereby increasing requests for retransmission and retrans maintain connectivity when a mobile device connects to a mission. Therefore, as the network approaches available foreign network. On the other hand, IPsec-based (Internet bandwidth, the load increases, ultimately leading to failure. In Protocol Security, a security protocol from IETF) VPNs (Vir instances where a minimum quality of service must be guar tual Private Networks) use a tunneling scheme in which the anteed, special Internet technologies are required, to reserve 65 outer source IP address is based on a CoA at the point-of bandwidth or to specify network pathways. End-to-end qual attachment and an inner Source IP address assigned for the ity of service guarantees, however, may exceed the cost of “home' domain. In general if either address is changed. Such US 8,874,477 B2 3 4 as when the mobile node switches IP subnets, then a new tion, on the other hand, the buyer need not disclose the maxi tunnel is negotiated with new keys and several round trip mum private valuation, and those bidders with lower risk message exchanges. The renegotiation of the tunnel interferes tolerance will bid higher prices. See, www.isoc.org/inet98/ with seamless mobility across wired and wireless IP networks proceedings/3b/3b 3.html: www.ibm.com/iac/reports-tech spanning multiple IP subnets. 5 nical/reports-bus-neg-internet.html. Clearly, newer technologies, such as IP V6, and derivatives Two common types of auction are the English auction, thereof, are known, and may be applied in conjunction with which sells a single good to the highest bidder in an ascending the present invention. price auction, and the Dutch auction, in which multiple units Market Economy Systems are available for sale, and in which a starting price is selected In modern retail transactions, predetermined price transac- 10 by the auctioneer, which is successively reduced, until the tions are common, with market transactions, i.e., commerce supply is exhausted by bidders (or the minimum price/final conducted in a setting which allows the transaction price to time is reached), with the buyer(s) paying the lowest Success float based on the respective valuation allocated by the ful bid. The term Dutch auction is also applied to a type of buyer(s) and seller(s), often left to specialized fields. While sealed bid auction. In a multi-unit live Dutch auction, each interpersonal negotiation is often used to set a transfer price, 15 participant is provided with the current price, the quantity on this price is often different from a transfer price that might hand and the time remaining in the auction. This type of result from a best-efforts attempt at establishing a market auction, typically takes place over a very short period of time price. Assuming that the market price is optimal, it is there and there is a flurry of activity in the last portion of the auction fore assumed that alternatives are sub-optimal. Therefore, the process. The actual auction terminates when there is no more establishment of a market price is desirable over simple nego- 20 product to be sold or the time period expires. tiations. In selecting the optimal type of auction, a number of factors One particular problem with market-based commerce is are considered. In order to sell large quantities of a perishable that both seller optimization and market efficiency depend on commodity in a short period of time, the descending price the fact that representative participants of a preselected class auctions are often preferred. For example, the produce and are invited to participate, and are able to promptly communi- 25 flower markets in Holland routinely use the Dutch auction cate, on a relevant timescale, in order to accurately value the (hence the derivation of the name), while the U.S. Govern goods or services and make an offer. Thus, in traditional ment uses this form to sell its financial instruments. The market-based system, all participants are in the same room, or format of a traditional Dutch auction encourages early bid connected by a high quality (low latency, low error) telecom ders to bid up to their “private value', hoping to pay some munications link. Alternately, the market valuation process is 30 price below the “private value'. In making a bid, the “private prolonged over an extended period, allowing non-real time value” becomes known, helping to establish a published mar communications of market information and bids. Thus, ket value and demand curve for the goods, thus allowing both attempts at ascertaining a market price for non-commodity buyers and sellers to define strategies for future auctions. goods can be subject to Substantial inefficiencies, which In an auction, typically a seller retains an auctioneer to reduce any potential gains by market pricing. Further, while 35 conduct an auction with multiple buyers. (In a reverse auc market pricing might be considered “fair”, it also imposes an tion, a buyer Solicits the lowest price from multiple competing element of risk, reducing the ability of parties to predict future vendors for a desired purchase). Since the seller retains the pricing and revenues. Addressing this risk may also improve auctioneer, the seller essentially defines the rules of the auc efficiency of a market-based system, that is, increase the tion. These rules are typically defined to maximize the rev overall Surplus in the market. 40 enues or profit to the seller, while providing an inviting forum Auction Systems to encourage a maximum number of high valued buyers. If the When a single party seeks to sell goods to the highest rules discourage high valuations of the goods or services, or valued purchaser(s), to establish a market price, the rules of discourage participation by an important set of potential bid conduct typically define an auction. Typically, known auc ders, then the rules are not optimum. Rules may also be tions provide an ascending price or descending price over 45 imposed to discourage bidders who are unlikely to Submit time, with bidders making offers or ceasing to make offers, in winning bids from consuming resources. A rule may also be the descending price or ascending price models, respectively, imposed to account for the valuation of the good or service to define the market price. After determining the winner of the applied by the seller, in the form of a reserve price. It is noted auction, typically abidder who establishes a largest economic that these rules typically seek to allocate to the seller a portion Surplus, the pricing rules define the payment, which may be in 50 of the economic benefit that would normally inure to the accordance with a uniform price auction, wherein all Success buyer (in a perfectly efficient auction), creating an economic ful bidders pay the lowest successful bid, a second price inefficiency. However, since the auction is to benefit the seller, auction wherein the winning bidder pays the amount bid by not Society as a whole, this potential inefficiency is tolerated. the next highest bidder, and pay-what-you-bid (first price) An optimum auction thus seeks to produce a maximum profit auctions. The pay-what-you-bid auction is also known as a 55 (or net revenues) for the seller. An efficient auction, on the discriminative auction while the uniform price auction is other hand, maximizes the sum of the utilities for the buyer known as a non-discriminative auction. In a second-price and seller. It remains a subject of academic debate as to which auction, also known as a Vickrey auction, the policy seeks to auction rules are most optimum in given circumstances; how create a disincentive for speculation and to encourage bidders ever, in practice, simplicity of implementation may be a para to submit bids reflecting their true value for the good, rather 60 mount concern, and simple auctions may result in highest than “shaving the bid to achieve a lower cost. In the uniform revenues; complex auctions, while theoretically more opti price and second price schemes, the bidder is encourages to mal, may discourage bidders from participating or from disclose the actual private value to the bidder of the good or applying their true and full private valuation in the auction service, since at any price below this amount, there is an process. excess gain to the buyer, whereas by withholding this amount 65 Typically, the rules of the auction are predefined and invari the bid may be unsuccessful, resulting in a loss of the pre ant. Further, for a number of reasons, auctions typically apply Sumably desirable opportunity. In the pay-what-you-bid auc the same rules to all bidders, even though, with a priori US 8,874,477 B2 5 6 knowledge of the private values assigned by each bidder to the likewise, in an auction market followed by perfect resale, it is goods, or a prediction of the private value, an optimization optimal for the seller to allocate the goods to those with the rule may be applied to extract the full value assigned by each highest value. Therefore, if post-auction trading is permitted, bidder, while selling above the seller's reserve. the seller will not benefit from these later gains, and the seller In a known ascending price auction, each participant must will obtain sub optimal revenues. be made aware of the status of the auction, e.g., open, closed, These studies, however, typically do not consider transac and the contemporaneous price. A bid is indicated by the tion costs and internal inefficiencies of the resellers, as well as identification of the bidder at the contemporaneous price, or the possibility of multiple classes of purchasers, or even mul occasionally at any price above the minimum bid increment tiple channels of distribution, which may be subject to vary plus the previous price. The bids are asynchronous, and there 10 ing controls or restrictions, and thus in a real market, Such fore each bidder must be immediately informed of the par theoretical optimal allocation is unlikely. In fact, in real mar ticulars of each bid by other bidders. kets the transaction costs involved in transfer of ownership In a known descending price auction, the process tradition are often critical in determining a method of sale and distri ally entails a common clock, which corresponds to a decre bution of goods. For example, it is the efficiency of sale that menting price at each decrement interval, with an ending time 15 motivates the auction in the first place. Yet, the auction pro (and price). Therefore, once each participant is made aware of cess itself may consume a substantial margin, for example the auction parameters, e.g., starting price, price decrement, 1-15% of the transaction value. To presume, even without ending price/time, before the start of the auction, the only externally imposed restrictions on resale, that all of the effi information that must be transmitted is auction status (e.g., ciencies of the market may be extracted by free reallocation, inventory remaining). ignores that the motivation of the buyer is a profitable trans As stated above, an auction is traditionally considered an action, and the buyer may have fixed and variable costs on the efficient manner of liquidating goods at a market price. The order of magnitude of the margin. Thus, there are substantial theory of an auction is that either the buyer will not resell, and opportunities for the seller to gain enhanced revenues by thus has an internal or private valuation of the goods regard defining rules of the auction, strategically allocating inven less of other's perceived values, or that the winner will resell, 25 tory amount and setting reserve pricing. either to gain economic efficiency or as a part of the buyer's Therefore, perfect resale is but a fiction created in auction regular business. In the later case, it is a general presumption (game) theory. Given this deviation from the ideal presump that the resale buyers are notinattendance at the auction or are tions, auction theory may be interpreted to provide the seller otherwise precluded from bidding, and therefore that, after with a motivation to misallocate or withhold based on the the auction, there will remain demand for the goods at a price 30 deviation of practice from theory, likely based on the respec in excess of the price paid during the auction. Extinction of tive transaction costs, seller's utility of the goods, and other this residual demand results in the so-called “winner's curse'. factors not considered by the simple analyses. in which the buyer can make no profit from the transaction A number of proposals have been made for effecting auc during the auction. Since this detracts from the value of the tion systems using the Internet. These systems include con auction as a means of conducting profitable commerce, it is of 35 Sumer-to-consumer, business-to-consumer, and business-to concern to both buyer and seller. In fact, experience with business types. Generally, these auctions, of various types and initial public offerings (IPOs) of stock through various means implementations discussed further below, are conducted has demonstrated that by making stock available directly to through Internet browsers using hypertext markup language all classes of potential purchasers, latent demand for a new (HTML) “web pages, using HTTP. In some instances, such issue is extinguished, and the stock price is likely to decline 40 as BIDWATCH, discussed further below, an application with after issuance, resulting in an IPO which is characterized as associated applets is provided to define a user interface “unsuccessful. This potential for post IPO decline tempers instead of HTML. even initial interest in the issue, resulting in a paradoxical As stated above, the information packets from the transac decline in revenues from the vehicle. In other words, the tion server to client systems associated with respective bid “money on the table' resulting from immediate retrading of 45 ders communicate various information regarding the status of IPO shares is deemed a required aspect of the IPO process. an interactive auction during the progress thereof. The net Thus, methods that retain latent demand after IPO shares work traffic from the client systems to the transaction server result in post IPO increases, and therefore a “successful IPO. is often limited to the placement of bids; however, the amount Therefore, where the transaction scheme anticipates demand of information required to be transmitted can vary greatly, and for resale after the initial distribution, it is often important to 50 may involve a complex dialogue of communications to com assure a reasonable margin for resellers and limitations on plete the auction offer. Typically, Internet based auction sys direct sale to ultimate consumers. tems have scalability issues, wherein economies of scale are Research into auction theory (game theory) shows that in not completely apparent, leading to implementation of rela an auction, the goal of the seller is to optimize the auction by tively large transaction server Systems to handle peak loads. allocating the goods inefficiently, if possible, and thus to 55 When the processing power of the transaction server system appropriate to himself an excess gain. This inefficiency mani is exceeded, entire system outages may occur, resulting in lost fests itself by either withholding goods from the market or sales or diminished profits, and diminished goodwill. placing the goods in the wrong hands. In order to assure for In most Internet auction system implementations, there are the seller a maximum gain from a misallocation of the goods, large quantities of simultaneous auctions, with each auction restrictions on resale are imposed; otherwise, post auction 60 accepting tens or hundreds of bids over a timescale of hours to trading will tend to undue the misallocation, and the antici days. In systems where the transaction Volume exceeds these pation of this trading will tend to control the auction pricing. scales, for example in Stock and commodity exchanges, The misallocation of goods imposed by the seller through which can accommodate large numbers of transactions per restrictions allows the seller to achieve greater revenues than second involving the same issue, a private network, or even a if free resale were permitted. It is believed that in an auction 65 local area network, is employed, and the public Internet is not followed by perfect resale, that any mis-assignment of the used as a direct communications system with the transaction goods lowers the seller's revenues below the optimum and server. Thus, while infrastructures are available to allow suc US 8,874,477 B2 7 8 cessful handling of massive transaction per second Volumes, having a stated refresh period of 10 seconds, with a long these systems typically avoid direct public Internet commu latency for confirmation of a bid, due to constraints on Soft nications or use of Some of its limiting technologies. The ware execution, quality of service in communications transaction processing limitations are often due to the finite streams, and bid confirmation dialogue. Thus, it is possible to time required to handle, e.g., open, update, and close, data 5 lose a bid even if an attempt was made prior to another bidder. base records. The need to quickly enter the bid, at risk of being too late, In business-to-business auctions, buyers seek to ensure that makes the process potentially error prone. the population of ultimate consumers for the good or services Proxy bidding, as discussed above, is a known technique are not present at the auction, in order to avoid the “winner's for overcoming the constraints of Internet communications curse', where the highest bidder in the auction cannot liqui 10 and client processing limitations, since it bypasses the client date or work the asset at a profit. Thus, business-to-business and telecommunications links and may execute solely on the auctions are distinct from business-to-consumer auctions. In host system or local thereto. However, proxy bidding under the former, the optimization by the seller must account for the mines some of the efficiencies gained by a live market. desire or directive of the seller to avoid direct retail distribu U.S. Pat. No. 5,890,138 to Godin, et al. (Mar. 30, 1999), tion, and instead to rely on a distribution tier represented in 15 expressly incorporated herein by reference in its entirety, the auction. In the latter, the seller seeks maximum revenues relates to an Internet auction system. The system implements and to exhaust the possibilities for downstream trade in the a declining price auction process, removing a user from the goods or services. In fact, these types of auctions may be auction process once an indication to purchase has been distinguished by various implementing rules, such as requir received. See, Rockoff, T. E. Groves, M.: “Design of an ing sales tax resale certificates, minimum lot size quantities, Internet-based System for Remote Dutch Auctions”, Internet preregistration or qualification, Support or associated Ser Research, V 5, n 4, pp. 10-16, MCB University Press, Jan. 1, vices, or limitations on the title to the goods themselves. The 1995. conduct of these auctions may also differ, in that consumer A known computer site for auctioning a product on-line involvement typically is permissive of mistake or indecision, comprises at least one web server computer designed for while in a pure business environment professionalism and 25 serving a host of computer browsers and providing the brows decisiveness are mandated. ers with the capability to participate in various auctions, In many instances, psychology plays an important role in where each auction is of a single product, at a specified time, the conduct of the auction. In a live auction, bidders can see with a specified number of the product available for sale. The each other, and judge the tempo of the auction. In addition, web server cooperates with a separate database computer, multiple auctions are often conducted sequentially, so that 30 separated from the web server computer by a firewall. The each bidder can begin to understand the other bidder's pat database computer is accessible to the web computer server terns, including hesitation, bluffing, facial gestures or man computer to allow selective retrieval of product information, nerisms. Thus, bidders often prefer live auctions to remote or which includes a product description, the quantity of the automated auctions if the bidding is to be conducted Strate product to be auctioned, a start price of the product, and an gically. 35 image of the product. The web server computer displays, Internet auctions are quite different from live auctions with updated during an auction, the current price of the product, respect to psychological factors. Live auctions are often the quantity of the product remaining available for purchase monitored closely by bidders, who strategically make bids, and the measure of the time remaining in the auction. The based not only on the “value' of the goods, but also on an current price is decreased in a predetermined manner during assessment of the competition, timing, psychology, and 40 the auction. Each user is provided with an input instructing progress of the auction. It is for this reason that so-called the system to purchase the product at a displayed current proxy bidding, wherein the bidder creates a preprogrammed price, transmitting an identification and required financial “strategy”, usually limited to a maximum price, are disfa authorization for the purchase of the product, which must be Vored as a means to minimize purchase price, and offered as confirmed within a predetermined time. In the known system, a service by auctioneers who stand to make a profit based on 45 a certain fall-out rate in the actual purchase confirmation may the transaction price. A maximum price proxy bidding system be assumed, and therefore some overselling allowed. Further, is somewhat inefficient, in that other bidders may test the after a purchase is indicated, the user's screen is not updated, proxy, seeking to increase the bid price, without actually obscuring the ultimate lowest selling price from the user. intending to purchase, or contrarily, after testing the proxy, a However, if the user maintains a second browser, he can bidder might give up, even below a price he might have been 50 continue to monitor the auction to determine whether the willing to pay. Thus, the proxy imposes inefficiency in the product could have been purchased at a lower price, and if so, system that effectively increases the transaction cost. fail to confirm the committed purchase and purchase the same In order to address a flurry of activity that often occurs at goods at a lower price while reserving the goods to avoid risk the end of an auction, an auction may be held open until no of loss. Thus, the system is flawed, and may fail to produce an further bids are cleared for a period of time, even if advertised 55 efficient transaction or optimal price. to end at a certain time. This is common to both live and An Internet declining price auction system may provide the automated auctions. However, this lack of determinism may ability to track the price demand curve, providing valuable upset coordinated Schedules, thus impairing efficient busi marketing information. For example, in trying to determine ness use of the auction system. the response at different prices, companies normally have to In order to facilitate management of bids and bidding, 60 conduct market Surveys. In contrast, with a declining price some of the Internet auction sites have provided non-Hyper auction, Substantial information regarding price and demand text Markup Language (HTML) browser based software is immediately known. The relationship between participat “applet” to track auctions. For example, ONSALE.COM has ing bidders and average purchasers can then be applied to made available a Marimba Castanet(R) applet called Bidwatch provide a conventional price demand curve for the particular to track auction progress for particular items or classes of 65 product. items, and to facilitate bidding thereon. This system, how U.S. Pat. No. 5,835,896, Fisher, et al., issued Nov. 10, ever, lacks real-time performance under many circumstances, 1998, expressly incorporated herein by reference in its US 8,874,477 B2 9 10 entirety, provides method and system for processing and system's database. Participants may revise their executory transmitting electronic auction information over the Internet, bids, by entering updated bids. Thus, at one extreme, a bidder between a central transaction server system and remote bid who wishes to economize on his time may choose to enter his der terminals. Those bids are recorded by the system and the entire set of bidding rules into the computerized system at the bidders are updated with the current auction status informa start of the auction, effectively treating this as a sealed-bid tion. When appropriate, the system closes the auction from auction. At the opposite extreme, a bidder who wishes to further bidding and notifies the winning bidders and losers as closely participate in the auction may choose to constantly to the auction outcome. The transaction server posts informa monitor the auction’s progress and to Submit all of his bids in tion from a database describing a lot available for purchase, real time. See also, U.S. patent application Ser. No. 08/582, receives a plurality of bids, stored in a bid database, in 10 901 filed Jan. 4, 1996, which provides a method for auction response to the information, and automatically categorizes ing multiple, identical objects and close Substitutes. the bids as successful or unsuccessful. Each bid is validated, Cryptographic Technology and an electronic mail message is sent informing the bidder of U.S. Pat. No. 5,956,408 (Arnold, Sep. 21, 1999), expressly the bid status. This system employs HTTP, and thus does not incorporated herein by reference, relates to an apparatus and automatically update remote terminal Screens, requiring the 15 method for secure distribution of data. Data, including pro e-mail notification feature. gram and software updates, is encrypted by a public key The auction rules may be flexible, for example including encryption system using the private key of the data sender. Dutch-type auctions, for example by implementing a price The sender also digitally signs the data. The receiver decrypts markdown feature with Scheduled price adjustments, and the encrypted data, using the public key of the sender, and English-type (progressive) auctions, with price increases cor Verifies the digital signature on the transmitted data. The responding to Successively higher bids. In the Dutch type program interacts with basic information stored within the auction, the price markdown feature may be responsive to confines of the receiver. As result of the interaction, the soft bidding activity overtime, amount of bids received, and num ware updates are installed within the confines of the user, and ber of items bid for. Likewise, in the progressive auction, the the basic information stored within the confines of the user is award price may be dependent on the quantity desired, and 25 changed. typically implements a lowest successful bid price rule. Bids U.S. Pat. No. 5,982,891 (Ginter, et al., Nov. 9, 1999); U.S. that are below a preset maximum posted selling price are Pat. No. 5,949,876 (Ginter, et al., Sep. 7, 1999); and U.S. Pat. maintained in reserve by the system. If a certain sales Volume No. 5,892,900 (Ginter, et al., Apr. 6, 1999), expressly incor is not achieved in a specified period of time, the price is porated herein by reference, relate to systems and methods for reduced to liquidate demand above the price point, with the 30 secure transaction management and electronic rights protec new price becoming the posted price. On the other hand, if a tion. Electronic appliances. Such as computers, help to ensure certain sales volume is exceeded in a specified period of time, that information is accessed and used only in authorized the system may automatically increase the price. These auto ways, and maintain the integrity, availability, and/or confi matic price changes allow the seller to respond quickly to dentiality of the information. Such electronic appliances pro market conditions while keeping the price of the merchandise 35 vide a distributed virtual distribution environment (VDE) that as high as possible, to the seller's benefit. A “Proxy Bidding may enforce a secure chain of handling and control, for feature allows a bidder to place a bid for the maximum example, to control and/or meter or otherwise monitor use of amount they are willing to pay, keeping this value a secret, electronically stored or disseminated information. Such a displaying only the amount necessary to win the item up to the virtual distribution environment may be used to protect rights amount of the currently high bids or proxy bids of other 40 of various participants in electronic commerce and other elec bidders. This feature allows bidders to participate in the elec tronic or electronic-facilitated transactions. Distributed and tronic auction without revealing to the other bidders the other operating systems, environments and architectures, extent to which they are willing to increase their bids, while Such as, for example, those using tamper-resistant hardware maintaining control of their maximum bid without closely based processors, may establish security at each node. These monitoring the bidding. The feature assures proxy bidders the 45 techniques may be used to Support an all-electronic informa lowest possible price up to a specified maximum without tion distribution, for example, utilizing the “electronic high requiring frequent inquiries as to the state of the bidding. way.” A "Floating Closing Time’ feature may also be imple U.S. Pat. No. 6,009,177 (Sudia, Dec. 28, 1999), expressly mented whereby the auction for a particular item is automati incorporated herein by reference, relates to a cryptographic cally closed if no new bids are received within a predeter 50 system and method with a key escrow feature that uses a mined time interval, assuming an increasing price auction. method for verifiably splitting users’ private encryption keys Bidders thus have an incentive to place bids expeditiously, into components and for sending those components to trusted rather than waiting until near the anticipated close of the agents chosen by the particular users, and provides a system auction. that uses modern public key certificate management, U.S. Pat. No. 5,905,975, Ausubel, issued May 18, 1999, 55 enforced by a chip device that also self-certifies. The methods expressly incorporated herein by reference in its entirety, for key escrow and receiving an escrow certificate are also relates to computer implemented methods and apparatus for applied herein to a more generalized case of registering a auctions. The proposed system provides intelligent systems trusted device with a trusted third party and receiving autho for the auctioneer and for the user. The auctioneer's system rization from that party enabling the device to communicate contains information from a user system based on bid infor 60 with other trusted devices. Further preferred embodiments mation entered by the user. With this information, the auc provide for rekeying and upgrading of device firmware using tioneer's system determines whether the auction can be con a certificate system, and encryption of stream-oriented data. cluded or not and appropriate messages are transmitted. At U.S. Pat. No. 6,052,467 (Brands, Apr. 18, 2000), expressly any point in the auction, bidders are provided the opportunity incorporated herein by reference, relates to a system for to submit not only their current bids, but also to enter future 65 ensuring that the blinding of secret-key certificates is bids, or bidding rules which may have the opportunity to restricted, even if the issuing protocol is performed in parallel become relevant at future times or prices, into the auction mode. A cryptographic method is disclosed that enables the US 8,874,477 B2 11 12 issuer in a secret-key certificate issuing protocol to issue the place of the trusted individual or organization in protect triples consisting of a secret key, a corresponding public key, ing the party whose record contains information that has legal and a secret-key certificate of the issuer on the public key, in mandates to rights of privacy. Such a way that receiving parties can blind the public key and E-Commerce Systems the certificate, but cannot blind a predetermined non-trivial U.S. Pat. No. 5,946,669 (Polk, Aug. 31, 1999), expressly predicate of the secret key even when executions of the issu incorporated herein by reference, relates to a method and ing protocol are performed in parallel. apparatus for payment processing using debit-based elec U.S. Pat. No. 6,052,780 (Glover, Apr. 18, 2000), expressly tronic funds transfer and disbursement processing using incorporated herein by reference, relates to a computer sys addendum-based electronic data interchange. This disclosure tem and process for accessing an encrypted and self-decrypt 10 describes a payment and disbursement system, wherein an initiator authorizes a payment and disbursement to a collector ing digital information product while restricting access to and the collector processes the payment and disbursement decrypted digital information. Some of these problems with through an accumulator agency. The accumulator agency digital information protection systems may be overcome by processes the payment as a debit-based transaction and pro providing a mechanism that allows a content provider to 15 cesses the disbursement as an addendum-based transaction. encrypt digital information without requiring either a hard The processing of a debit-based transaction generally occurs ware or platform manufacturer or a content consumer to by electronic funds transfer (EFT) or by financial electronic provide Support for the specific form of corresponding data interchange (FEDI). The processing of an addendum decryption. This mechanism can be provided in a manner that based transaction generally occurs by electronic data inter allows the digital information to be copied easily for back-up change (EDI). purposes and to be transferred easily for distribution, but U.S. Pat. No. 6,005,939 (Fortenberry, et al., Dec. 21, 1999), which should not permit copying of the digital information in expressly incorporated herein by reference, relates to a decrypted form. In particular, the encrypted digital informa method and apparatus for storing an Internet user's identity tion is stored as an executable computer program that and access rights to World Wide Web resources. A method includes a decryption program that decrypts the encrypted 25 and apparatus for obtaining user information to conduct information to provide the desired digital information, upon secure transactions on the Internet without having to re-enter Successful completion of an authorization procedure by the the information multiple times is described. The method and user. In combination with other mechanisms that track distri apparatus can also provide a technique by which secured bution, enforce royalty payments and control access to access to the data can be achieved over the Internet. A pass decryption keys, an improved method is provided for identi 30 port containing user-defined information at various security fying and detecting sources of unauthorized copies. Suitable levels is stored in a secure server apparatus, or passportagent, authorization procedures also enable the digital information connected to computer network. A user process instructs the to be distributed for a limited number of uses and/or users, passport agent to release all or portions of the passport to a thus enabling per-use fees to be charged for the digital infor recipient node and forwards a key to the recipient node to mation. 35 unlock the passport information. A patent application entitled “Three Party Authorization U.S. Pat. No. 6,016,484 (Williams, et al., Jan. 18, 2000), System', (Robert Nagel, David Felsher, and Steven Hoffberg, expressly incorporated herein by reference, relates to a sys inventors, 2001) provides a three party authorization encryp tem, method and apparatus for network electronic payment tion technique. This technique has the significant advantage instrument and certification of payment and credit collection of being more secure than a Public Key system because it 40 utilizing a payment. An electronic monetary system provides requires the agreement of all three parties—the creator of the for transactions utilizing an electronic-monetary system that secure record, the party that the secure record is about, and the emulates a wallet or a purse that is customarily used for database repository—on a case by case basis, in order to keeping money, credit cards and other forms of payment release secure records. Then and only then can a released organized. Access to the instruments in the wallet or purse is secure record be decrypted and accessed by the requesting 45 restricted by a password to avoid unauthorized payments. A party alone. Each access generates an entry into a log file with certificate form must be completed in order to obtain an automatic security alerts for any unusual activity. A compo instrument. The certificate form obtains the information nec nent of this system is that each party wishing to secure records essary for creating a certificate granting authority to utilize an enters into a contract with a “virtual trust agency to represent instrument, a payment holder and a complete electronic wal that party in all matters where privacy is an issue. The virtual 50 let. Electronic approval results in the generation of an elec trust agency never has access to the data contained in the tronic transaction to complete the order. If a user selects a secured records and yet acts on behalf of the party whose particular certificate, a particular payment instrument holder information is contained in the secure records to control data will be generated based on the selected certificate. In addi access to authorized requesting parties. To enable this pri tion, the issuing agent for the certificate defines a default vacy, the virtual trust agency issues a public-private key pair 55 bitmap for the instrument associated with a particular certifi and maintains the party's private key. The private key is only cate, and the default bitmap will be displayed when the cer used in calculations to generate an intermediate key that is tificate definition is completed. Finally, the number associ passed on to the data repository and used to re-encrypt the ated with a particular certificate will be utilized to determine data for the requesting party's view. A unique aspect of this if a particular party can issue a certificate. patent pending technique is the fact that the party's records 60 U.S. Pat. No. 6,029,150 (Kravitz, Feb. 22, 2000), expressly are actually protected from unauthorized use at all times incorporated herein by reference, relates to a system and inside the organization that holds the database repository or method of payment in an electronic payment system wherein by the original records creator, not simply in transmission a plurality of customers have accounts with an agent. A cus from the database repository or the individual or organization tomer obtains an authenticated quote from a specific mer that created the record in the first place, to outside requesting 65 chant, the quote including a specification of goods and a parties. This system requires consent of all three parties to payment amount for those goods. The customer sends to the decrypt secured information. Its virtual trust component takes agent a single communication including a request for pay US 8,874,477 B2 13 14 ment of the payment amount to the specific merchant and a munication between creditor and debtor is not required; unique identification of the customer. The agent issues to the rather, the transaction produces a result which eventually customeran authenticated payment advice based only on the results in an economic transfer, but which may remain out single communication and secret shared between the cus standing Subsequent to transfer of the underlying goods or tomer and the agent and status information, which the agent services. The theory underlying this micropayment scheme is knows about the merchant, and/or the customer. The cus that the monetary units are Small enough Such that risks of tomerforwards a portion of the payment advice to the specific failure in transaction closure is relatively insignificant for merchant. The specific merchant provides the goods to the both parties, but that a user gets few chances to default before customer in response to receiving the portion of the payment credit is withdrawn. On the other hand, the transaction costs advice. 10 of a non-real time transactions of Small monetary units are U.S. Pat. No. 6,047.269 (Biffar, Apr. 4, 2000), expressly Substantially less than those of secure, unlimited or poten incorporated herein by reference, relates to a self-contained tially high value, real time verified transactions, allowing and payment system with creating and facilitating transfer of facilitating Such types of commerce. Thus, the rights manage circulating digital Vouchers representing value. A digital ment system may employ applets local to the client system, Voucher has an identifying element and a dynamic log. The 15 which communicate with other applets and/or the server and/ identifying element includes information Such as the trans or a vendor/rights-holder to validate a transaction, at low ferable value, a serial number and a digital signature. The transactional costs. dynamic log records the movement of the Voucher through Neural Networks the system and accordingly grows over time. This allows the The resources relating to Neural Networks, listed in the system operator not only to reconcile the vouchers before Neural Networks References Appendix, each of which is redeeming them, but also to recreate the history of movement expressly incorporated herein by reference, provides a Sound of a voucher should an irregularity like a duplicate voucher be basis for understanding the field of neural networks (and the detected. These vouchers are used within a self-contained subset called artificial neural networks, which distinguish system including a large number of remote devices that are biological systems) and how these might be used to Solve linked to a central system. The central system can be linked to 25 problems. A review of these references will provide a state of an external system. The external system, as well as the remote knowledge appropriate for an understanding of aspects of the devices, is connected to the central system by any one or a invention which rely on Neural Networks, and to avoid a combination of networks. The networks must be able to trans prolix discussion of no benefit to those already possessing an port digital information, for example the Internet, cellular appropriate state of knowledge. networks, telecommunication networks, cable networks or 30 Telematics proprietary networks. Vouchers can also be transferred from The resources relating to telematics listed in the Telematics one remote device to another remote device. These remote Appendix, each of which is expressly incorporated herein by devices can communicate through a number of methods with reference, provides a background in the theory and practice of each other. For example, for a non-face-to-face transaction telematics, as well as some of the underlying technologies. A the Internet is a choice, for a face-to-face or close proximity 35 review of these references is therefore useful in understand transactions tone signals or light signals are likely methods. ing practical issues and the context of functions and technolo In addition, at the time of a transaction a digital receipt can be gies which may be used in conjunction with the advances set created which will facilitate a fast replacement of vouchers forth herein. stored in a lost remote device. Micropayments 40 BRIEF DESCRIPTION OF THE DRAWINGS U.S. Pat. No. 5,999,919 (Jarecki, et al., Dec. 7, 1999), expressly incorporated herein by reference, relates to an effi The drawings show: cient micropayment system. Existing software proposals for FIG. 1 shows a Bayesian Network; electronic payments can be divided into “on-line' schemes FIG. 2 shows a Markov chain; which require participation of a trusted party (the bank) in 45 FIG.3 shows a model of the output of a Markov chain as a every transaction and are secure against overspending, and mixture of Gaussians; “off-line' schemes which do not require a third party and FIGS. 4A-4C show an input-output, a factorial, and a guarantee only that overspending is detected when vendors coupled Hidden Markov Model (HMM), respectively; submit their transaction records to the bank (usually at the end FIG.5 shows a predictor correctoralgorithm of the discrete of the day). A new “hybrid’ scheme is proposed which com 50 Kalman filter cycle; bines the advantages of both “on-line” and “off-line' elec FIG. 6 shows aspects of the discrete Kalman filter cycle tronic payment schemes. It allows for control of overspending algorithm; at a cost of only a modest increase in communication com FIG. 7 shows aspects of the extended Kalman filter cycle: pared to the off-line schemes. The protocol is based on proba FIG. 8 shows a block diagram of a preferred embodiment bilistic polling. During each transaction, with some Small 55 of a communications system according to the present inven probability, the vendor forwards information about this trans tion; action to the bank. This enables the bank to maintain an FIG. 9 is a schematic diagram showing the prioritization accurate approximation of a customer's spending. The fre scheme; and quency of polling messages is related to the monetary value of FIG. 10 is a block diagram representing a message format. transactions and the amount of overspending the bank is 60 willing to risk. For transactions of high monetary value, the DESCRIPTION OF THE INVENTION cost of polling approaches that of the on-line schemes, but for micropayments, the cost of polling is a small increase over the The present invention seeks, among other aspects, to apply traffic incurred by the off-line schemes. aspects of optimization theory to the control and arbitration of Micropayments are often preferred where the amount of 65 communities of resources, that is, elements or agents which the transaction does not justify the costs of complete financial operate independently and therefore cannot be directly con security. In the micropayment scheme, typically a direct com trolled without compromise. Rather, the system is controlled US 8,874,477 B2 15 16 by providing incentives and disincentives for various behav is, the presence and action of the arbitrage agent increases the iors and actions seeking to promote an efficient outcome for economic Surplus of the transaction and the marketingeneral. the system as a whole, given the external constraints. Each An object of the present invention therefore seeks to over agent then maximizes its own state based on its own value come the inefficiency of seeking to solve a complex NP function, in light of the incentives and disincentives, resulting complete optimization problem by providing arbitrage in an optimal network. opportunities that allow a market solution to the optimization This optimization employs elements of game theory, and which balances optimization cost with intermediary profits. the present invention therefore invokes all those elements Accordingly, while the net result may deviate from an abstract encompassed within its scope as applied to the problems and optimum condition, when one considers the cost of achieving 10 this abstract optimum, the arbitrage mediated result is Supe systems presented. The ad hoc network of elements typically rior in terms of market surplus. The availability of arbitrage reside within “communities, that is, the present invention and intermediaries therefore allows a particular agent to bal does not particularly seek to apply principles to trivial net ance its own optimization costs against overall gains. works which can be optimized without compromise or arbi The Subject of complexity theory, including combinatorial tration, although its principles may be applicable. The present 15 optimization, and solution of approximation of NP-complete invention therefore applies to the enhancement or optimiza problems, has been extensively studied. See, e.g., the refer tion of communities. These communities may themselves ences set forth in the combinatorial optimization and auction have various rules, reputation hierarchies, arrangements or appendix, which are expressly incorporated herein by refer cultures, which can be respected or programmed as a part of CCC. the system operation, or merely operate as constraints on Assuming a set of rational agents, each agent will seek to optimization. Thus, in accordance with a game theoretic locally optimize its state to achieve the greatest gains and analysis, various rules and perceived benefits may be applied incur the lowest net costs. Thus, in a system which seeks to to appropriately model the real system, or may be imposed to optimize a network of Such agents, by permitting each agent control behavior. to optimize its local environment state, the network may then These communities may be formed or employed for vari 25 be approximated by a network of local, environments, each ous purposes, and typically interoperate in a "commons' or typically comprising a plurality of agents. Thus, in the same economy, in which all relevant actions of each member of the way as the complexity of an NP-complete problem grows in community have an effect on others, and this effect can be polynomial space, the simplification of an NP complete prob quantified or normalized into “economic' terms. For lem will also be polynomial. While this simplification incurs example, a typical application of this technology would be to 30 an inefficiency, each agent models a proximate region in the arbitrate access to a common or mutually interfering medium, space of interest, which tends to be linear (i.e., Superposable) such as a communications network. By optimizing commu in a preferred embodiment. Agents compete with each other, nications, the greatest aggregate value of communications and therefore the incentive to distort is limited. Likewise, will generally be achieved, which may or may not correspond since the preferred simplification of the problem does not to a greatest aggregate communications bandwidth. For 35 impose a heuristic (i.e., a Substitution of a first relatively example, both quantity and quality of service may be inde simpler or more readily analytic algorithm for a second more pendent (or semi-independent) parameters. Thus, the system intractable algorithm), it does not accordingly distort the tends to promote a high quality of service (or at least as high incentives from those inherent in the basic optimization. a quality as is required) overa bulk volume of service. This, in In a simple auction, a role is imposed on an agent, while in turn, permits new applications which depend on reliable com 40 a market an agent can choose its own role dynamically, in a munications. series of transactions, dependent on its own value function. In A general type of economic optimization is a market a market, a set of agents having a resource or requiring a economy, in which independent agents each act to maximize resource seek to interact to reallocate the resource, such that their respective interests. A Subset of a market is an auction, in the needs are generally satisfied, up to a “market clearing which a resource is allocated to a highest valued user or 45 price'. That is, individual agents transact to transfer the consumer, or a user or consumer acquires a resource at a resource from those with Supply to those with demand, at a lowest cost, in a single process. In a market economy, agents price between the private values of the two agents, which may act as peers, and each agent may act as a source of Supply reallocation occurs until the demand ask price is higher than or assert a demand. That is, at Some market price, a consumer the Supply bid price. An auction is designed to maximize the may relinquish assets to others and become a Supplier. Gen 50 economic Surplus, which is then typically allocated to the erally, a consumer has a higher private value for a resource more restrictive of the Source of Supply or consumer or the than a Supplier. The Suppler, for example, may have a lower sponsor. A market, on the other hand, generally operates to cost to obtain the resource, or a lower value for consumption minimize the gap between bids and asks over an extended of the resource, or both. Peers that both buy and sella resource period, and thus the economic Surplus tends to proportionate may seek to arbitrage, that is, seek to establish a committed 55 based on the balance of Supply and demand at the clearing Source of Supply at a lower cost than a committed purchaser, price. The particular reallocation also depends on the state of thus assuring a profit. In order to be effective, arbitrage information of each agent, inefficiencies in transfer, and the requires that the intermediary have an advantage, or else the like. ultimate buyer would transact directly with the ultimate Where a need or Supply is not unitary, one possible means seller. The advantage(s) may be, for example, information, 60 for achieving an optimal solution is a combinatorial auction, proprietary elements or methods, location, lower transac in which multiple Suppliers, or multiple consumers, or both, tional costs, available capital, risk tolerance, storage facility, reallocate the resource orportions thereof. Thus, a single need or the like. is not met by a single Supplier, but rather there are at least So long as the advantage does not derive from an economi three parties to a transaction. The net result is a competition cally inefficient monopoly or other structure that artificially 65 between parties that raises the potential for a holdout. In fact, and/or “illegally limits the efficiency of other agents, the one way to circumvent this issue (a “holdout' problem) is to arbitrage agent increases net efficiency of the network. That have director indirect (bypass) competition for each element. US 8,874,477 B2 17 18 In Sucha circumstance, no agent can effectively demand more delay. Indeed, these premises in a narrow sense might be than the least cost alternate(s). false, i.e., a rational auctioneer might indeed make greater A combinatorial auction (multifactorial optimization, also investment, undertake higher risk, or incur greater delay. known as a Vickrey Clarke Grove IVCG Auction) seeks to However, this possible inefficiency merely shifts the alloca match, for the entire network of possibilities, the highest tion of the Surplus, and to the extent there is a Substantial gain valued users with the lowest cost suppliers. This leads, how to be made, encourages arbitrage, which in turn encourages ever, to a surplus between the two, which must be allocated. In competition at Subsequent auctions, leading to at least a par a one-to-many auction, the Surplus is generally allocated to tial remediation of the allocation"error” in the longterm, over the restricted agent, i.e., the agent(s) with “market power. On a series of auctions. the other hand, in an optimal market, the Surplus will tend 10 toward Zero. That is, the profit to each party will tend toward Therefore, the market system, with derivative and arbitrage a competitive mean, with higher profits only gained by under possibilities, and deviations from optimal performance is at taking higher risk. In imperfect markets, arbitrage opportu least partially self-correcting over the long term. nities exist, where profits can be made by trading the same Likewise, because the system has mechanisms for reduc SOUC. 15 ing the effects of imperfections in the presumptions and/or the In a multihopadhoc network, a path between Source and conformance of a real system to the stated mechanisms and destination consists of a number of legs, with alternate paths applicable rules, particular aspects of the system which available for selection of an optimum. If each node has its own impose administrative or overhead burdens may be circum distinct destination, there will likely be competing demands vented by imposing less restrictive criteria and allowing a for each intermediate communication node. 'selfcorrecting mechanism to remediate. Thus, for example, One way to promote revealing a private value is if the end ifa theoretically ideal mechanism imposes a 15% burden due result of the process does not penalize those with aberrantly to overhead, thus achieving an 85% overall efficiency (100 high or low values. One known method is to compute the 15–85), while a simplifying presumption achieves a result result of the process as if the bidder or askor was not involved, which imposes a 20% efficiency impairment but only a 2% leading to a so-called second price. Thus, the highest bidder 25 overhead factor (100-20-2–78), and an arbitrage mechanism wins, at a price established by a second highest bid. A lowest is available to correct the simplified model to gain 12% effi askor wins, at a price established by the second lowest askor. ciency with another 2% overhead (78+12-2), the net result is In a market, the highest bidder and lowest askor win, with a 88% efficiency, above that of the theoretically ideal mecha second price dependant on a more restrictive of Supply and nism. demand. In a combinatorial auction, this may be extended to 30 An arbitrage mechanism seeks to identify inefficiency price each component as if the highest bidder was uninvolved. based on Superior information or mechanism, and a pricing or In one embodiment of the invention, the second price applies value disparity, and conduct a countertrade seeking to exploit to both buyer and seller, respectively, with the economic the disparity while undertaking relatively low risk, to increase Surplus allocated to other purposes. Thus, in this case, neither overall market efficiency. (That is, to ultimately reallocate party gains the particular benefit of an imbalance of Supply 35 resources from a lower valued holder to a higher valued and demand. In fact, this perturbs the traditional market holder). Somewhat, in that an imbalance in Supply and demand does An ad hoc network generally presents a case where indi not particularly recruit new market entrants in the same man vidual nodes have imperfect information, and efforts to gain ner as an allocation of the Surplus. better information invariably lead to increased overhead. Arbritage 40 Therefore, by intentionally truncating the information gath The present invention seeks to model, within a micro ering and discovery aspect of the ad hoc network, a residual economy, the elements of the real economy which tend to arbitrage opportunity will remain, but the inherent ineffi improve efficiency toward “perfection', that is, a perfect uni ciency of the arbitrage may be less than the corresponding Versal balance of Supply and demand, for which no strategy overhead involved in providing more perfect information to (other than bidding a true private value) will produce a Supe 45 the individual nodes (i.e., overall arbitrage cost is less than rior result. efficiency gain). These efficiency producing elements, paradoxically, are AS Such, a design feature of an embodiment of the inven the parasitic elements which thrive off of predictable ineffi tion is to provide or even encourage arbitrage mechanisms ciencies. That is, by promoting competition among the para and arbitrage opportunities, in an effort to improve overall sitic elements, an efficient balance of optimization of the 50 system efficiency. In fact, an embodiment of the system is direct market and optimization of the derivative markets will preferably constructed to regularly provide arbitrage oppor produce an overall Superior result to an optimization of the tunities which can be conducted with low risk and with sub direct market alone. stantial market efficiency gains, and these arbitrage opportu While the use of derivative markets in real economies is nities may be an important part of the operation of the well known, the implementation of these as aspects of micro 55 embodiment. economies and isolated markets is not well known, and a part A second opportunity provides risk transference, such as of an embodiment of the present invention. For example, in a debt transactions, insurance, and market-making, and/or the corporate bankruptcy auction, there are often resellers present like. In such transactions, a market risk is apparent. Each who seek to purchase assets cheaply at a “wholesale price, node, on the other and, has its own Subjective risk tolerance. and to redistribute them on a less urgent basis or in Smaller 60 Likewise, the market risk provides an opportunity for nodes quantities, or at a different location, or otherwise trans having a high risk tolerance to improve yield, by trading risk formed, and to sell themata higher “retail price. The auction for return. Those nodes which have generally greater liquid price is determined by the number and constitution of bidders, resources, which inherently have no return while uninvested, as well as the possibility of proxy or absentee bidding. In fact, and may permit other nodes having lesser resources to bor we presume that the auctioneers themselves are efficient, and 65 row, at interest. Because there is a risk of non-payment, nodes that significantly higher bid prices are not available in a modi may have different credit ratings, and this creates an oppor fied process without incurring Substantial investment, risk, or tunity for credit rating "agencies' and/or guarantors. In an ad US 8,874,477 B2 19 20 hoc network, there is also a possibility for delivery failure, the network from external wealth redistribution, real effi which, in turn, provides an opportunity for insurance. ciency gains may be unavailable. Therefore, both types of Fair Allocation Perturbed by Hierarchal Considerations economies, as well as hybrids, are available. Thus, as dis In a typical auction, each player is treated fairly; that is, the cussed in more detail below, a “fair' initial (or recurring) same rules apply to each player, and therefore a single wealth distribution may be applied, which may be supple economy describes the process. The fair auction therefore mented with, and/or provide an output of external wealth. poses challenges for an inherently hierarchal set of users, The rules or proportion of external influence may be prede Such as a military organization, where rank is accompanied termined, adaptive, or otherwise. by privilege. The net result, however, is a decided disadvan In accordance with the proposed artificial economy, each tage to lower ranking agents, at least when viewed in light of 10 node has a generator function for generating economic units, constricted self-interest. The issues that arise are similar to which are then used in an auction with other nodes to create a the relating to “altruism', although not identical, and thus the market economy, that is, each node has a Supply and demand game theoretic analysis of altruistic behavior may be function, and acts as a source or sink for a limited resource. In imported for consideration as appropriate. Some cases, nodes may have only supply or demand func In a mobile ad hoc communications network, a real issue is 15 tions, or a degree of asymmetry, but in this case, these are user defection or non-compliance. For example, where a cost typically subject to an external economic consideration, and is imposed on a user for participating in the ad hoc network, the artificial economy will be less effective in providing e.g., battery power consumption, if the anticipated benefit appropriate incentives. According to this embodiment, the does not exceed the cost, the user will simply turn off the artificial economic units have a temporally and spatially device until actually needed. The result of mass defection declining value, so that wealth does not accumulate over long will, of course, be the instability and failure of the ad hoc periods and cannot be transferred over large distances. The network itself, leading to decreased utility, even for those who decline may be linear, exponential, or based on Some other gain an unfair or undue advantage under the system. Thus, function. This creates a set of microeconomies insulated from perceived fairness and net benefit is required to network Suc each other. Where distant microeconomies must deal with cess, assuming that defection and/or non-compliance are pos 25 each other, there is a discount. This architecture provides a sible. number of advantages, for example, by decreasing the influ On the otherhand, in military systems, the assertion of rank ence of more spatially and temporally distant effects, the as a basis for priority is not itself arbitrary and capricious. Scope of an optimization analysis may be relatively con Orders and communications from a central command are strained, while reducing the amount of information which critical for the organization itself, and thus the lower ranking 30 must be stored overtime and/or carried over distance in order agents gain at least a peripheral benefit as their own chain of to permit an optimization. Likewise, since the economy is command employs their resources. Therefore, the difficulty artificial, the discount need not be recouped within the scope in analyzing the application of a fair game to a hierarchal of the system. In the same manner, a somewhat different organization is principally a result of conceptualizing and incentive structure may be provided; that is, economic units aligning the individual incentives with those of the organiza 35 generated at one location and at one time may have a higher tion as a whole and the relationship between branches. Thus, value at a different location and time; this may encourage in contradistinction to typical self-organizing peer-to-peer reduced immediate use of the system, and relocation to higher networks, a hierarchal network is not seen as self-organizing, valued locations. As discussed below, one embodiment of the at least in terms of the hierarchy, which is extrinsic to the invention permits trading of credits, and thus, for example, a formation of the communications network under consider 40 user may establish a repeater site at an underserved location to ation. gain credits for use elsewhere. Preferably, beyond a “near As discussed below, the “distortions” of the network field’ effect, the value does not continue to increase, since this imposed by the external hierarchy can be analyzed and may result in inflationary pressures, and undermine the utility accounted forby, for example, the concepts of inheritance and of the system in optimally balancing immediate Supply and delegation. Thus, each branch of a hierarchy tree may be 45 demand at a particular location. considered an object, which receives a set of characteristics As can be seen, through modifications of the governing from its root, and from which each sub-branch inherits the rules and formulae, the system can be incentivized to behave characteristics and adds subcharacteristics of, for example, in certain ways, but care should be exercised since a too specialization. It is noted that the hierarchy need not follow narrow analysis of the incentive might result in unintended non-ambiguous or perfect rules, and thus there is no particular 50 effects. To the extent that human behavior is involved, care limit imposed that the hierarchy necessarily follow these for should also be exercised in applying a rationality assumption, malisms. Rather, by analyzing those aspects of the hierarchy since this is not always true. Rather, there may be applicable which comply with these formalisms in accordance there models for human irrational behavior that are better Suited to with, efficiency is facilitated. an understanding of the network behavior in response to a In establishing an economic system, a preliminary ques 55 perturbation. tion is whether the system is microeconomic or macroeco The typical peer-to-peer ad hoc network may be extended nomic; that is, whether the economy is linked to a real to the hierarchal case by treating each branch (including economy or insulated from it. One disadvantage of a real Sub-branches) within the chain of command as an economic economy with respect to a peer relationship is that external unit with respect to the generator function. At any level of the wealth can override internal dynamics, thus diminishing the 60 hierarchy, the commander retains a portion of the wealth advantages to be gained by optimization, and potentially cre generation capacity, and delegates the remainder to its Sub ating a perception of unfairness for externally less wealthy ordinates. Therefore, the rank and hierarchal considerations agents, at least unless and until the system accomplishes a are translated to an economic wealth (or wealth generation) wealth redistribution. An artificial economy provides a solu distribution. One aspect of this system allows wealth transfer tion for a peer network in which each node has an equal 65 or redistribution, although in a real system, a time delay is opportunity to gain control over the adhoc network, indepen imposed, and in the event of a temporally and/or spatially dent of outside influences. On the other hand, by insulating declining value, the transfer will impose a cost. Thus, an US 8,874,477 B2 21 22 initial misallocation is undesired, and there will be an incen encourages bidders and/or sellers to seek to be in the limited tive to optimally distribute the wealth initially. Of course, if group splitting the pool, and thus incentivizes higher bids and centralized control with low penalty is desired, it is possible to lower asks. Of course, a party will have a much stronger limit the penalty, of any, for wealth redistribution through incentive to avoid bidding outside its valuation bounds, so the appropriate rules, although the time for propagation through 5 risk of this type of inefficiency is small. the network remains an issue, and blind nodes (i.e., those As discussed above, one embodiment of the invention pro which do not have an efficient communication path, or have vides a possible redistribution or wealth among nodes within insufficient resources to utilize otherwise available paths a hierarchal chain. This redistribution may be of accumulated through the hierarchy) may also lead to limitations on system wealth, or of the generation function portion. Trading among performance. 10 hierarchally related parties is preferred, since the perceived In this system, there may be an economic competitive cost is low, and the wealth can be repeatedly redistributed. In distortion, under which a node's subjective value of a fact, it is because of the possibility of wealth oscillation and resource is influenced by its then subjective wealth. If a node teaming that the declining wealth function is preferred, since is supplied with wealth beyond its needs, the wealth is wasted, this will tend to defeat closely related party control over the since it declines in value and cannot be hoarded indefinitely. 15 network for extended periods. (In a network wealth model in which wealth could be hoarded It is noted that, in a multihop mobile ad hoc network, if a indefinitely, Small deviations from optimality and arbitrage communication path fail, no further transfers are possible, opportunities may be exploited to create a perception of potentially resulting in stalled or corrupt system configura unfairness, thus, this is not preferred.) If a node is Supplied tion. It is possible to transfer an expiring or declining portion with insufficient wealth, economic Surplus through transac of the generating function; however, this might lead a node tional gains are lost. Thus, each node must analyze its which is out of range to have no ability to rejoin the network expected circumstances to retain or delegate the generator upon return, and thus act as an impediment to efficient net function, and to optimally allocate wealth between compet work operation. Therefore, it is preferred that, in an artificial ing Subordinates. Likewise, there may be a plurality of quasi economy, each node has some intrinsic wealth generator optimal states. 25 function, so an extended period of inactivity, a node gains In any economic transaction, there is an amount that a wealth likely sufficient to rejoin the network as a full partici seller requires to part with the resource, a price a buyer is pant. willing to pay, and a surplus between them. Typically, in a two In practice, in a typical military-type hierarchy, the bulk of party transaction, the Surplus is allocated to the party initiat the wealth generating function will be distributed to the low ing the transaction, that is, the party initiating the transaction 30 est ranks with the highest numbers. Thus, under normal cir uses some discovery mechanism to find the minimum price cumstances, the network will appear to operate according to acceptable by the buyer. In brokered or agent-mediated trans a non-hierarchal (i.e., peer-to-peer) model, with the distortion actions, a portion of the Surplus is allocated to a facilitator. In that not all nodes have a common generator function. On the accordance with this aspect of the present invention, compli other hand, hierarchically Superior nodes either retain, or ance with the community rules, as well as an incentive to bid 35 more likely, can quickly recruit Surrounding Subordinates to or ask a true private value is encouraged by distributing a allocate their wealth generating function and accumulated portion of the transaction Surplus to competitive bidders in wealth to pass urgent or valuable messages. Thus, if 85% of accordance with their reported valuations. In particular, the the wealth and network resources are distributed to the low competitive bidders seeking to allocate a scarce resource for est-ranking members, then the maximum distortion due to themselves receive compensation for deferring to the winning 40 hierarchal modifications is 15%. bidder in an amount commensurate with their reported value. One way that this allocation of wealth may be apparent is Thus, sellers receive their minimum acceptable value, buyers with respect to the use of expensive assets. Thus, a high level pay their maximum valuation, the Surplus is distributed to the node might have access to a high power broadcast system or community in a manner tending to promote the highest bids, licensed spectrum, while low level nodes might ordinarily be that is, the true bidder value (or even possibly slightly higher). 45 limited to lower power transmission and/or unlicenced spec In a corresponding manner, the auction rules can be estab trum or cellular wireless communications. For a low level lished to incentivized sellers to ask the minimum possible node to generate a broadcast using an expensive asset (or to amount. For example, a portion of the Surplus may be allo allocate a massive amount of space bandwidth product), it cated to bidders in accordance with how close they come to must pass the request up through the chain of command, until the winning ask. Therefore, both incentives may be applied, 50 sufficient wealth (i.e., authority) is available to implement the for example with the surplus split in two, and half allocated to broadcast. the bidder pool and half allocated to the seller pool. Clearly, In fact, such communications and authorizations are quite other allocations are possible. consistent with the expectations within a hierarchal organi The winning bidder and/or seller may be included within Zation, and this construct is likely to be accepted within a the rebate pool. This is particularly advantageous where for 55 military-type hierarchal organization. various reasons, the winning bidder is not selected. Thus, this Under normal circumstances, a Superior would have an process potentially decouples the bidding (auction) process incentive to assure that each Subordinate node possesses Suf and the resulting commercial transaction. It may also be ficient wealth to carry out its function and be incentivized to useful to apply Vickrey (second price) rules to the auction, participate in the network. If a subordinate has insufficient that is, the winning bidder pays the second bid price, and/or 60 initial wealth (or wealth generating function) allocation, it the winning seller pays the second ask price. may still participate, but it must expend its internal resources Because of transactional inefficiencies, human behavioral to obtain wealth for participation in its own benefit. This, in aspects, and a desire to avoid increased network overhead by turn, leads to a potential exhaustion of resources, and the “false' bidders seeking a share of the allocation pool without unavailability of the node for ad hoc intermediary use, even intending to win the auction, it may be useful to limit the 65 for the benefit of the hierarchy. An initial surplus allocation allocation of the surplus pool to a subset of the bidders and/or will lead to overbidding for resources, and thus inefficient sellers, for example the top three of one or both. This therefore resource allocation, potential waste of allocation, and a dis US 8,874,477 B2 23 24 incentive to act as an intermediary in the ad hoc network. auctions, each having their limitations and attributes, as a While in a traditional military hierarchy, cooperation can be class these are well accepted as a means for transfer of goods mandated, in Systems where cooperation is perceived as con or rights at an optimum price. Where multiple goods or rights trary to the net personal interests of the actor, network stabil are required in a Sufficient combination to achieve a require ity may be poor, and defection in spite of mandate. ment, a so-called Vickrey–Clarke-Groves (VCG) auction may In a military system, it is thus possible to formulate an be employed. In Such an auction, each Supplier asserts a “engineered solution which forces participation and elimi desired price for his component. The various combinations nates defection; however, it is clear that such solutions forfeit which meet the requirement are then compared, and the low the potential gains of optimality, and incentivizes circumven est selected. In a combinatorial Supply auction, a plurality of tion and non-compliance. Further, because Such a system is 10 not “cost sensitive' (however the appropriate cost function buyers each seek a divisible commodity, and each bids its best might be expressed), it fails to respond to “market' forces. price. The bidders with the combination of prices which is Accordingly, a peer to peer mobile adhoc network Suitable maximum are selected. In a commodity market, there are a for respecting hierarchal organization structures is delegation plurality of buyers and sellers, so the auction is more com is provided. In this hierarchal system, the hierarchy is repre 15 plex. In a market economy, the redistribution of goods or sented by an initial wealth or wealth generation function services is typically transferred between those who value distribution, and the hierarchally higher nodes can reallocate them least to those who value them most. The transaction wealth of nodes beneath themselves, exercising their higher price depends on the balance between Supply and demand; authority. This wealth redistribution can be overt or covert, with the Surplus being allocated to the limiting factor. and if overt, the hierarchal orders can be imposed without Derivatives, Hedges, Futures and Insurance nodal assent. In a covert redistribution, trust may be required In a market economy, the liquidity of the commodity is to assure redistribution by a node to a grandchild node. The typically such that the gap between bid and ask is Small wealth and its distribution can be implemented using modi enough that the gap between them is Small enough that it is fied micropayment techniques and other verifiable crypto insignificant in terms of preventing a transaction. In a tradi graphic techniques. This wealth can be applied to auctions 25 tional market, the allocation of the Surplus oscillates indepen and markets, to allocate resources. Various aspects of this dence on whether it is a buyer's or seller's market. Of course, system are discussed in more detail elsewhere in this speci the quantum of liquidity necessary to assure an acceptably fication. low gap is Subjective, but typically, if the size of the market is Manet System sufficient, there will be low opportunity for arbitrage, or at Multihop Ad Hoc Networks require cooperation of nodes 30 least a competitive market for arbitrage. The arbitrage may be which are relatively disinterested in the content being con either in the commodity, or options, derivatives, futures, or veyed. Typically, such disinterested intermediaries incur a the like. cost for participation, for example, power consumption or In a market for communications resources, derivatives may opportunity cost. Economic incentives may be used to pro provide significant advantages over a simple unitary market mote cooperation of disinterested intermediaries, also known 35 for direct transactions. For example, a node may wish to as recruitment. An economic optimization may be achieved procure a reliable communications pathway (high quality of using a market-finding process. Such as an auction. In many service or QoS) for an extended period. Thus, it may seek to scenarios, the desire for the fairness of an auction is tempered commit resources into the future, and not be subject to future by other concerns, i.e., there are constraints on the optimiza competition for or price fluctuation of those resources, espe tion which influence price and parties of a transaction. For 40 cially being Subject to a prior broadcast of its own private example, in military communication systems, rank may be valuation and a potential understanding by competitors of the deemed an important factor in access to, and control over, the presumed need for continued allocation of the resources. communications medium. A simple process of rank-based Thus, for similar reasons for the existence of derivative, preemption, without regard for Subjective or objective impor options, futures, etc. markets in the real economy, their anal tance, will result in an inefficient economic distortion. In 45 ogy may be provided within a communications resource mar order to normalize the application of rank, one is presented ket. with two options: imposing a normalization scheme with In a futures market analogy, an agent seeks to procure its respect to rank to create a unified economy, or providing long-term or bulk requirements, or seeks to dispose of its considering rank using a set of rules outside of the economy. assets in advance of their availability. In this way, there is One way to normalize rank, and the implicit hierarchy under 50 increased predictability, and less possibility of self-competi lying the rank, is by treating the economy as an object-ori tion. It also allows transfer of assets in bulk to meet an entire ented hierarchy, in which each individual inherits or is allo requirement or production lot capability, thus increasing effi cated a subset of the rights of a parent, with peers within the ciency and avoiding partial availability or disposal. hierarchy operating in a purely economic manner. The extrin One issue in mobile ad hoc networks is accounting for sic consideration of rank, outside of an economy, can be 55 mobility of nodes and unreliability of communications. In denominated “respect', which corresponds to the societal commodities markets, one option is insurance of the under treatment of the issue, rather than normalizing this factor lying commodity and its production. The analogy in commu within the economy, in order to avoid unintended secondary nications resource markets focuses on communications reli economic distortion. Each system has its merits and limita ability, since one aspect of reliability, nodal mobility is tions. 60 “voluntary and not typically associated with an insurable An economic optimization is one involving a transaction in risk. On the other hand, the mobility risk may be mitigated by which all benefits and detriments can be expressed in normal an indemnification. In combination, these, and other risk ized terms, and therefore by balancing all factors, including transfer techniques, may provide means for a party engaged Supply and demand, at a price, an optimum is achieved. Auc in a communications market transaction to monetarily com tions are well known means to achieve an economic optimi 65 pensate for risk tolerance factors. An agent in the market Zation between distinct interests, to transfer a good or right in having a low risk tolerance can undertake risk transference, at exchange for a market price. While there are different types of Some additional but predetermined transaction costs, while US 8,874,477 B2 25 26 one with a high risk tolerance can 'go bare’ and obtain a sellers is optimized to achieve the maximum economic Sur lower transaction cost, or undertake third party risk for profit. plus. Winning bidders pay the maximum bid price or second Insurance may be provided in various manners. For price, while winning sellers receive their winning ask or example, Some potential market participants may reserve second price. The remaining Surplus is redistributed among wealth, capacity or demand for a fee, Subject to claim in the the winners and losing bidders, whose cooperation and non event of a risk event. In other cases, a separate system may be interference with the winning bidders is required for network employed, such as a cellular carrier, to step in, in the event that operation. The allocation of the portion to losing bidders is, a lower cost resource is unavailable (typically for bandwidth for example, in accordance with their proportionate bid for Supply only). A service provider may provide risk-related contested resources, and for example, limited to the few (e.g., allocations to network members in an effort to increase per 10 ceived network stability; likewise, if the network is externally 3) highest bidders or lowest offerors. The winning bids are controlled, each node can be subject to a reserve requirements determined by a VCG combinatorial process. The result is an which is centrally (or hierarchally) allocated. optimum network topology with a reasonable, but by no If an agent promises to deliver a resource, and ultimately means the only, fairness criterion, while promoting network fails to deliver, it may undertake an indemnification, paying 15 stability and utility. the buyer an amount representing 'damages', the transaction The purpose of rewarding losers is to encourage recruit cost of buyer, e.g., the cost of re-procurement plus lost pro ment, and therefore market liquidity. In order to discourage ductivity. Likewise, if an agent fails to consume resources strategic losing bids, one possible option is to impose a sta committed to it, it owes the promised payment, less the resale tistical noise on the process to increase the risk that a strate value of the remaining resources. An indemnification insurer/ gically losing bid will be a winning bid. Another way is to guarantor can undertake to pay the gap on behalf of the allocate the available Surplus corresponding to the closeness defaulting party. Typically, the insurer is not a normal agent of the losing bid to the winning bid, not merely on its mag peer, but can be. nitude. Alternately, a “historical value for the resource may Hedge strategies may also be employed in known manner. be established, and an allocation made only if the bid is at or In order for markets to be efficient, there must be a possi 25 above the trailing mean value. Further, the loser's allocation bility of beneficial use or resale of future assets. This imposes may be dependent on a future bid with respect to a corre Some complexity, since the assets are neither physical nor sponding resource at or above the prior value. In similar possessed by the intermediary. However, cryptographic manner, various algorithms for Surplus allocation may be authentication of transactions may provide Some remedy. On designed to encourage recruitment of agents seeking to win, the other hand, by increasing liquidity and providing market 30 while possibly discouraging bidders who have little realistic makers, the transaction Surplus may be minimized, and thus possibility of winning. Bidders who do not seek to win the reallocation of the surplus as discussed above minimized. impose an inefficiency on the network, for example requiring Likewise, in a market generally composed of agents within other agents to communicate, evaluate, acknowledge, and close proximity, the interposition of intermediaries may defer to these bids. Therefore, a relatively small bidding fee result in inefficiencies rather than efficiencies, and the utility 35 may be imposed in order to assert a bid, which may be used to of such complexity may better come from the facilitation of increase the available surplus to be allocated between the distant transactions. Thus, if one presumes slow, random winning and top losing bidders. nodal mobility, little advantage is seen from liquid resource As discussed above, risk may be a factor in valuing a and demand reallocation. On the other hand, if an agent has a resource. The auction optimization may therefore be normal predefined itinerary for rapidly relocating, it can efficiently 40 ized or perturbed in dependence on an economic assessment conduct transactions over its path, prearranging communica of a risk tolerance, either based on a personal valuation, or tion paths, and thus providing trunk services. Thus, over a based on a third party valuation (insurance/indemnification). short term, direct multihop communications provide long Likewise, the optimization may also be modified to account distance communications of both administrative and content for other factors. data. On the other hand, over a longer term, relocation of 45 Thus, one issue with such a traditional scheme for fair agents may provide greater efficiency for transport of admin allocation of resources is that it does not readily permit inten istrative information, increasing the efficiency of content data tional distortions, that is, the system is “fair. However, in communications over the limited communications resources, Some instances, a relatively extrinsic consideration to Supply especially if a store-and-forward paradigm is acceptable. and Subjective demand may be a core requirement of a sys It is noted that in an economy having persistent and deep 50 tem. For example, in military systems, it is traditional and use of financial derivatives, a stable currency is preferred, and expected that higher military rank will provide access to and the declining value credit discussed above would provide a control over resources on a favored basis. (Note that, in con disincentive to agents who might otherwise take risks over a trast to an example discussed elsewhere herein, this favorit long time-frame. It is possible, however, to distinguish ism is not enforced by a hierarchal wealth generation distri between credits held by “consumers' and those held by “arbi 55 bution). In civilian systems, emergency and police use may trageurs’ or institutions, with the former having a declining also be considered privileged. However, by seeking to apply value but can be spent, and those which have a stable value but economic rules to this access, a number of issues arise. Most must be first converted (at some possible administrative cost) significantly, as a privileged user disburses currency, this is for consumer use. distributed to unprivileged users, leading to an inflationary Bandwidth Auction 60 effect and comparative dilution of the intended privilege. If A previous scheme proposes the application of game the economy is real, that is the currency is linked to a real theory in the control of multihop mobile ad hoc networks economy, this grant of privilege will incurreal costs, which is according to “fair principles. In this prior scheme, nodes also not always an intended effect. If the economy is syn seeking to control the network (i.e., are “buyers' of band thetic, that is, it is unlinked to external economies, then the width), conduct an auction for the resources desired. Like 65 redistribution of wealth within the system can grant dramatic wise, potential intermediate nodes conduct an auction to Sup and potentially undesired control to a few nodes, potentially ply the resources. The set of winning bidders and winning conveying the privilege to those undeserving, except perhaps US 8,874,477 B2 27 28 due to fortuitous circumstances such as being in a critical sufficient in the case that respect is not afforded, for example, location or being capable of interfering with a crucial com from those of equal rank or above, or those who for various munication. reasons have other factors that override the assertion of Two different schemes may be used to address this desire respect. Therefore, one way to determine the amount of for both economic optimality and hierarchal considerations. respect to be afforded is the self-worth advertised for the One Scheme maintains optimality and fairness within the resources requested. This process therefore may minimize the economic structure, but applies a generally orthogonal con deviation from optimal and therefore promotes stability of the sideration of “respect' as a separate factor within the opera network. It is further noted that those who assert respect based tion of the protocol. Respect is a subjective factor, and thus on hierarchy typically have available Substantial economic permits each bidder to weight its own considerations. It is 10 further noted that Buttyan et al. have discussed this factor as resources, and therefore it is largely a desire to avoid eco a part of an automated means for ensuring compliance with nomic redistribution rather than an inability to effect such a network rules, in the absence of a hierarchy. Levente Buttyan redistribution, that compels a consideration of respect. and Jean-Pierre Hubaux, Nuglets: a Virtual Currency to In a combinatorial auction, each leg of a multihop link is Stimulate Cooperation in Self-Organized Mobile Ad Hoc 15 separately acquired and accounted. Therefore, administration Networks, Technical Report DSC/2001/004, EPFL-DI-ICA, of the process is quite involved. That is, each bidder broad January 2001, incorporated herein by reference. See, P. casts a set of bids for the resources required, and an optimal Michiardi and R. Molva, CORE: A collaborative reputation network with maximum Surplus is defined. Each leg of each mechanism to enforce node cooperation in mobile ad hoc path is therefore allocated a value. In this case, it is the networks. In B. Jerman-Blazic and T. Klobucar, editors, Com winning bidder who defers based on respect, since the other munications and Multimedia Security, IFIPTC6/TC11 Sixth resources are compensated equally and therefore agnostic. Joint Working Conference on Communications and Multime Thus, if pricing is defined by the economic optimization, dia Security, Sep. 26-27, 2002, Portoroz, Slovenia, volume then the respect consideration requires that a Subsidy be 228 of IFIP Conference Proceedings, pages 107-121. Kluwer applied, either as an excess payment up to the amount of the Academic, 2002: Sonja Buchegger and Jean-Yves Le Bou 25 winning bid, or as a discount provided by the sellers, down to dec, A Robust Reputation System for P2P and Mobile Ad-hoc the actually bid value. Networks, Second Workshop on the Economics of Peer-to Since the pricing is dependent on the network absent the Peer Systems, June 2004: Po-WahYau and Chris J. Mitchell, respect consideration, there is an economic deficit or required Reputation Methods for Routing Security for Mobile Ad Hoc Subsidy. In some cases, the respected bidder simply pays the Networks; Frank Kargil, Andreas Klenk, Stefan Schlott, and 30 amount required, in excess of its actual bid. If we presume Micheal Weber. Advanced Detection of Selfish or Malicious that the respected bidder could have or would have outbid the Nodes in Ad Hoc Network. The 1st European Workshop on winning bidder, it then pays the third price, rather than the Security in Ad-Hoc and Sensor Networks (ESAS 2004); He, second price. If the respected bidder does not have, or will not Qi, et al., SORI: A Secure and Objective Reputation-based allocate the resources, then the Subsidy must come from the Incentive Scheme for Ad-Hoc Networks, IEEE Wireless 35 others involved. On one hand, since the respect in this case Communications and Networking Conference 2004, each of may be defined by the otherwise winning bidder, this bidder, which is expressly incorporated herein by reference. as an element of its respect, may pay the difference. However, The bias introduced in the system operation is created by an this cost (both the lost economic gains of the transaction and assertion by one claiming privilege, and deference by one the Subsidy) will quickly disincentivize any sort of grant of respecting privilege. One way to avoid Substantial economic 40 respect. The recipients could also provide a discount, how distortions is to require that the payment made be based on a ever this would require consent of both the winning bidder purely economic optimization, while selecting the winner and the recipients, making concluding the transaction more based on other factors. In this way, the perturbations of the difficult. One other possibility is to request“donations” from auction process itself is Subtle, that is, since bidders realize nearby nodes to meet the subsidy, a failure of which under that the winning bid may not result in the corresponding 45 mines the assertion of respect. benefit, but incurs the publication of private values and poten Another alternate is to assume that there is a Surplus tial bidding costs, there may be perturbation of the bidding between the aggregate winning bid and the aggregate cost, strategy from optimal. Likewise, since the privilege is itself and so long as the bidder claiming respect pays the minimum unfair and predictable, those with lower privilege ratings will cost, then the system remains operable, although the benefits have greater incentive to defect from, or act against, the 50 of Surplus allocation are lost, and all affected nodes must network. Therefore, it is important that either the assertion of defer to this respect mechanism. In this case, it is more diffi privilege be subjectively reasonable to those who must defer cult to arbitrate between competing demands for respect, to it, or the incidence or impact of the assertions be uncom unless a common value function is available, which in this mon or have low anticipated impact on the whole. On the case we presume is not available. other hand, the perturbation is only one-sided, since the pay 55 The node demanding respect may have an impact on path ment is defined by the network absent the assertion of privi segments outside its required route and the otherwise optimal lege. interfering routes; and thus the required payment to meet the In the extreme case, the assertion of privilege will com differential between the optimum network and the resulting pletely undermine the auction optimization, and the system network may thus be significant. will be prioritized on purely hierarchal grounds, and the pric 60 It is noted that, in the real economy, where the U.S. Gov ing non-optimal or unpredictable. This condition may be ernment allocates private resources, it is required to pay their acceptable or even efficient in military systems, but may be full value. This model appears rational, and therefore a pre unacceptable where the deference is voluntary and choice of ferred system requires a node claiming privilege and gaining network protocol is available, i.e., defection from the network a resulting benefit to pay the winning bid value (as an expres policies is an available choice. 65 sion of market value), and perhaps in addition pay the win It is noted that those seeking access based on respect, must ning bidder who is usurped its anticipated benefit, that is, the still make an economic bid. This bid, for example, should be difference in value between the second price and its published US 8,874,477 B2 29 30 private valuation, this having an economically neutral affect, agents, is to provide the greatest anticipated benefit, at the but also requiring a respected node to potentially possess least anticipated cost, compared to the alternates. substantial wealth. Game Theory provides a basis for understanding the A further possible resolution of this issue provides for an actions of Ad hoc network nodes. A multihopadhoc network assessment of an assertion of respect by each involved node. requires a communication to be passed through a disinter Since the allocation of respect is subjective, each bidder Sup ested node. The disinterested node incurs some cost, thus plies a bid, as well as an assertion of respect. Each other node leading to a disincentive to cooperate. Meanwhile, bystander receives the bids and assertions, and applies a weighting or nodes must defer their own communications in order to avoid discount based on its subjective analysis of the respect asser interference, especially in highly loaded networks. By under 10 standing the decision analysis of the various nodes in a net tion. In this case, the same bid is interpreted differently by work, it is possible to optimize a system which, in accordance each Supplier, and the Subjective analysis must be performed with game theory, provides benefits or incentives, to promote by or for each Supplier. By converting the respect assertion network reliability and stability. The incentive, in economic into a Subjective weighting or discount, a pure economic form, may be charged to those benefiting from the commu optimization may then be performed, with the subjectively 15 nication, and is preferably related to the value of the benefit perturbed result by each node reported and used to compute received. The proposed network optimization scheme the global optimization. employs a modified combinatorial (VCG) auction, which An alternate scheme for hierarchal deference is to organize optimally compensates those involved in the communication, the economy itself into a hierarchy, as discussed in the first with the benefiting party paying the second highest bid price example. In a hierarchy, a node has one parent and possibly (second price). The Surplus between the second price and multiple children. At each level, a node receives an allocation VCG price is distributed among those who defer to the win of wealth from its parent, and distributes all or a portion of its ning bidder according to respective bid value. Equilibrium wealth to children. A parent is presumed to control its chil usage and headroom may be influenced by deviating from a dren, and therefore can allocate their wealth or subjective Zero-sum condition. The mechanism seeks to define fairness valuations to its own ends. When nodes representing different 25 in terms of market value, providing probable participation lineages must be reconciled, one may refer to the common benefit for all nodes, leading to network stability. ancestor for arbitration, or a set of inherited rules to define the Ad Hoc Networks hierarchal relationships. An ad hoc network is a wireless network which does not In this system, the resources available for reallocation require fixed infrastructure or centralized control. The termi between branches of the hierarchy depend on the allocation 30 nals in the network cooperate and communicate with each by the common grandparent, as well as competing allocations other, in a self organizing network. In a multihop network, within the branch. This system presumes that children com communications can extend beyond the scope of a single municate with their parents and are obedient. In fact, if the node, employing neighboring nodes to forward messages to communication presumption is violated, one must then rely their destination. In a mobile ad hoc network, constraints are on a priori instructions, which may not be sufficiently adap 35 not placed on the mobility of nodes, that is, they can relocate tive to achieve an optimal result. If the obedience presump within a time scale which is short with respect to the commu tion is violated, then the hierarchal deference requires an nications, thus requiring consideration of dynamic changes in enforcement mechanism within the hierarchy. If both pre network architecture. Sumptions are simultaneously violated, then the system will Ad hoc networks pose control issues with respect to con likely fail, except on a voluntary basis, with results similar to 40 tention, routing and information conveyance. There are typi the “reputation' scheme described herein. cally tradeoffs involving equipment size, cost and complex Thus, it is possible to include hierarchal deference as a ity, protocol complexity, throughput efficiency, energy factor in optimization of a multihop mobile ad hoc network, consumption, and “fairness” of access arbitration. Other fac leading to compatibility with tiered organizations, as well as tors may also come into play. L. Buttyan and J.-P. Hubaux. with shared resources. 45 Rational exchange—a formal model based on game theory. In Game Theory Proceedings of the 2nd International Workshop on Electronic Use of Game Theory to control arbitration of ad hoc net Commerce (WELCOM), November 2001. citeseerist.ps works is well known. F. P. Kelly, A. Maulloo, and D. Tan. Rate u.edu/anO1 rational.html; P. Michiardi and R. Molva. Game control in communication networks: shadow prices, propor theoretic analysis of security in mobile ad hoc networks. tional fairness and stability. Journal of the Operational 50 Technical Report RR-02-070, Institut Eurécom, 2002: P. Research Society, 49, 1998. citeseerist.psu.edu/ Michiardi and R. Molva. A game theoretical approach to kelly98rate.html; J. MacKie-Mason and H. Varian. Pricing evaluate cooperation enforcement mechanisms in mobile ad congestible network resources. IEEE Journal on Selected hoc networks. In Proceedings of WiOpt 03, March 2003: Areas in Communications, 13(7): 1141-1149, 1995. Some Michiardi, P., Molva, R.: Making greed work in mobile ad hoc prior studies have focused on the incremental cost to each 55 networks. Technical report, Institut Eurecom (2002) citese node for participation in the network, without addressing the erist.psu.edu/michiardiO2making.html; S. Shenker. Making opportunity cost of a node foregoing control over the com greed work in networks: A game-theoretic analysis of Switch munication medium. Courcoubetis, C., Siris, V. A. and Sta service disciplines. IEEE/ACM Transactions on Networking, moulis, G. D. Integration of pricing and flow control for 3(6):819-831, December 1995; A. B. MacKenzie and S. B. available bit rate services in ATM networks. In Proceedings 60 Wicker. Selfish users in aloha: A game-theoretic approach. In IEEE Globecom '96, pp. 644-648. London, UK. citese Vehicular Technology Conference, 2001. VTC 2001 Fall. erist.psu.edu/courcoubetis96integration.html. IEEE VTS 54th, volume 3, October 2001; J. Crowcroft, R. A game theoretic approach addresses the situation where Gibbens, F. Kelly, and S. Ostring. Modelling incentives for the operation of an agent which has freedom of choice, allow collaboration in mobile ad hoc networks. In Proceedings of ing optimization on a high level, considering the possibility of 65 WiOptO3, 2003. alternatives to a well designed system. According to game Game theory studies the interactions of multiple indepen theory, the best way to ensure that a system retains compliant dent decision makers, each seeking to fulfill their own objec US 8,874,477 B2 31 32 tives. Game theory encompasses, for example, auction theory efficiency, that is, honesty in reporting the private value and strategic decision-making. By providing appropriate results in the maximum likelihood of prospective gain. incentives, a group of independent actors may be persuaded, Application of Game Theory to Ad Hoc Networks according to self-interest, to act toward the benefit of the There are a number of aspects of ad hoc network control group. That is, the selfish individual interests are aligned with 5 which may be adjusted in accordance with game theoretic the community interests. In this way, the community will be approaches. An example of the application of game theory to both efficient and the networkofactors stable and predictable. influence system architecture arises when communications Of course, any systems wherein the “incentives' impose too latency is an issue. A significant factor in latency is the node high a cost, themselves encourage circumvention. In this hop count. Therefore, a system may seek to reduce node hop 10 count by using an algorithm other than a nearest neighbor case, game theory also addresses this issue. algorithm, bypassing some nodes with longer distance com In computer networks, issues arise as the demand for com munications. In analyzing this possibility, one must not only munications bandwidth approaches the theoretical limit. look at the cost to the nodes involved in the communication, Under such circumstances, the behavior of nodes will affect but also the cost to nodes which are prevented from simulta how close to the theoretical limit the system comes, and also 15 neously accessing the network dude to interfering uses of which communications are permitted. The well known colli network resources. As a general proposition, the analysis of sion sense, multiple access (CSMA) protocol allows each the network must include the impact of each action, or net node to request access to the network, essentially without cost work State, on every node in the system, although simplifying or penalty, and regardless of the importance of the commu presumptions may be appropriate where information is nication. While the protocol incurs relatively low overhead unavailable, or the anticipated impact is trivial. and may provide fully decentralized control, under congested Game theory is readily applied in the optimization of com network conditions, the system may exhibit instability, that is, munications routes through a defined network, to achieve the a decline in throughput as demand increases, resulting in ever best economic Surplus allocation. In addition, the problem of increasing demand on the system resources and decreasing determining the network topology, and the communications throughput. Durga P. Satapathy and Jon M. Peha, “Perfor 25 themselves, are ancillary, though real, applications of game mance of Unlicensed Devices With a Spectrum Etiquette.” theory. Since the communications incidental to the arbitration Proceedings of IEEE Globecom, November 1997, pp. 414 require consideration of some of the same issues as the under 418. citeseerist.psu.edu/satapathy97performance.html. lying communications, corresponding elements of game According to game theory, the deficit of the CSMA protocol theory may apply at both levels of analysis. Due to various is that it is a dominant strategy to be selfish and hog resources, 30 uncertainties, the operation of the system is stochastic. This regardless of the cost to Society, resulting in “the tragedy of presumption, in turn, allows estimation of optimality within a the commons.” Garrett Hardin. The Tragedy of the Com margin of error, simplifying implementation as compared to a mons. Science, 162:1243-1248, 1968. Alternate Location: rigorous analysis without regard to statistical significance. diedff.com/page95.htm. There are a number of known and proven routing models In an ad hoc network used for conveying real-time infor 35 proposed for forwarding of packets inadhoc networks. These mation, as might be the case in a telematics system, there are include Ad Hoc On-Demand Distance Vector (AODV) Rout potentially unlimited data communication requirements (e.g., ing, Optimized Link State Routing Protocol (OLSR), Video data), and network congestion is almost guaranteed. Dynamic Source Routing Protocol (DSR), and Topology Dis Therefore, using a CSMA protocol as the paradigm for basic semination Based on Reverse-Path Forwarding (TBRPF). M. information conveyance is destined for failure, unless there is 40 Mauve, J. Widmer, and H. Hartenstein. A survey on position a disincentive to network use. (In power constrained circum based routing in mobile ad hoc networks. IEEE Network stances, this cost may itself provide Such a disincentive). On Magazine, 15(6):30-39, November 2001. citeseerist.ps the other hand, a system which provides more graceful deg u.edu/article/mauveO1 survey.html; Z. Haas. A new routing radation under high load, sensitivity to the importance of protocol for reconfigurable wireless networks. In IEEE 6th information to be communicated, and efficient utilization of 45 International Conference on Universal Communications the communications medium would appear more optimal. Record, volume 2, pages 562-566, October 1997: X. Hong, K. One way to impose a cost which varies in dependence on Xu, and M. Gerla. Scalable routing protocols for mobile ad the Societal value of the good or service, is to conduct an hoc networks. IEEE Networks, 16(4): 11-21, July 2002: D. auction, which is a mechanism to determine the market value Johnson, D. Maltz, andY.-C. Hu. The dynamic source routing of the good or service, at least between the auction partici 50 protocol for mobile ad hoc networks, April 2003. www.iet pants. forg/internet-drafts/draft-ietf-manet-dsr-09.txt; S.-J. Lee, W. Walsh, W. and M. Wellman (1998). A market protocol for Su, J. Hsu, M. Gerla, and R. Bagrodia. A performance com decentralized task allocation, in “Proceedings of the Third parison study of ad hoc wireless multicast protocols. In Pro International Conference on Multi-Agent Systems. pp. 325 ceedings of IEEE INFOCOM 2000, pages 565-574, March 332, IEEE Computer Society Press, Los Alamitos. In an 55 2000; K. Mase, Y. Wada, N. Mori, K. Nakano, M. Sengoku, auction, the bidder seeks to bid the lowest value, up to a value and S. Shinoda. Flooding schemes for a universal ad hoc less than or equal to his own private value (the actual value network. In Industrial Electronics Society, 2000. IECON which the bidder appraises the good or service, and above 2000, v.2, pp. 1129-1134, 2000; R. Ogier, F. Templin, and M. which there is no surplus), that will win the auction. Since Lewis. Topology dissemination based on reversepath for competitive bidders can minimize the gains of another bidder 60 warding, October 2003. Vesuvio.ipv6.cselt.it/internet-drafts/ by exploiting knowledge of the private value attached to the draft-ietf-manet-tbrpf-11.txt; A. Orda, R. Rom, and N. good or service by the bidder, it is generally a dominant Shimkin. Competitive routing in multi-user communication strategy for the bidder to attempt to keep its private value a networks. IEEE/ACM Transactions on Networking, secret, at least until the auction is concluded, thus yielding 1(5):510-521, October 1993: C. Perkins, E. Belding-Royer, strategies that result in the largest potential gain. On the other 65 and S. Das. Ad hoc on-demand distance vector (AODV) rout hand, in certain situations, release or publication of the pri ing. Request for comments 3561, Internet Engineering Task vate value is a dominant strategy, and can result in Substantial Force, 2003: C. E. Perkins, editor. Ad Hoc Networking. Addi US 8,874,477 B2 33 34 son-Wesley, Boston, 2001; E. Royer and C.-K. Toh. A review While the basic system may be a Zero sum system, and over of current routing protocols for ad hoc mobile wireless net time, the economic effects will likely average out (assuming works. IEEE Personal Communications, 6(2):46-55, April symmetric nodes), in any particular instance, the incentive for 1999; Holger Füfsler, Hannes Hartenstein, Dieter Vollmer, selfish behavior by a node will be diminished. Martin Mauve, Michael Kasemann, Location-Based Routing for Vehicular Ad-Hoc Networks, Reihe Informatik March One way to remedy selfish behavior is to increase the cost 2002, citeseerist.psu.edu/560036.html; J. Broch, D. A. of acting this way, that is, to impose a cost or tax for access to Maltz, D. B. Johnson, Y. C. Hu, and J. Jetcheva. A Perfor the network. In a practical implementation, however, this is mance Comparison of Multi-Hop Wireless Ad Hoc Network problematic, since under lightly loaded conditions, the Routing Protocols. In Proc. of the ACM/IEEE MobiCom, “value of the communications may not justify a fixed cost October 1998, citeseer.ist.psu.edu/broch98 performance 10 which might be reasonable under other conditions, and like .html. wise, under heavier loads, critical communications may still In most systems analyzed to date, the performance metrics be delayed or impeded. A variable cost, dependent on relative studied were power consumption, end-to-end data through “importance', may be imposed, and indeed, as alluded to put and delay, route acquisition time, percentage out-of-order delivery, and efficiency. A critical variable considered in 15 above, this cost may be market based, in the manner of an many prior studies is power cost, presuming a battery oper auction. In a multihop network, such an auction is compli ated transceiver with limited power availability. Juha Leino, cated by the requirement for a distribution of payments within Applications of Game Theory in Ad Hoc Network’, Mas the chain of nodes, with each node having potential alternate ter's Thesis, Helsinki University Of Technology (2003); J. demands for its cooperation. The market-based price-finding Shneidman and D. Parkes, “Rationality and Self-Interest in mechanism excludes nodes which ask a price not supported Peer to Peer Networks”. In Proc. 2nd Int. Workshop on Peer by its market position, and the auction itself may comprise a to-Peer Systems (IPTPS ’03), 2003, citeseer.nj.nec.com/ value function encompassing reliability, latency, quality of shneidman03rationality.html: V. Rodoplu and H.-Y. Meng. service, or other non-economic parameters, expressed in eco Minimum energy mobile wireless networks. IEEE Journal on nomic terms. The network may further require compensation Selected Areas in Communications, 17(8): 1333-1344, 25 to nodes which must defer communications because of incon August 1999; S. Singh, M. Woo, and C. S. Raghavendra. sistent states. Such as in order to avoid interference or dupli Power-aware routing in mobile ad hoc networks. In Proceed cative use of an intermediary node, and which take no direct ing of MOBICOM 1998, pages 181-190, 1998; A. Urpi, M. part in the communication. It is noted that the concept of the Bonuccelli, and S. Giordano. Modelling cooperation in winner of an auction paying the losers is not generally known, mobile ad hoc networks: a formal description of selfishness. 30 and indeed somewhat counterintuitive. Indeed, the effect of In Proceedings of WiOpt 03, March 2003: A. van den Nou this rule perturbs the traditional analysis framework, since the weland, P. Borm, W. van Golstein Brouwers, R. Groot Bru possibility of a payment from the winner to the loser alters the inderink, and S. Tijs. A game theoretic approach to problems allocation of economic surplus between the bidder, seller, and in telecommunication. Management Science, 42(2):294-303, others. Likewise, while the cost to the involved nodes may be February 1996. 35 real, the cost to the uninvolved nodes may be subjective. There can be significant differences in optimum routing While it would appear that involved nodes would generally be depending on whether a node can modulate its transmit better compensated than uninvolved nodes, the actual alloca power, which in turn controls range, and provides a further tion or reallocation of wealth according to the optimization control over network topology. Likewise, steerable antennas, may result in a different outcome. antenna arrays, and other forms of multiplexing provide fur 40 The network provides competitive access to the physical ther degrees of control over network topology. Note that the transport medium, and cooperation with the protocol pro protocol-level communications are preferably broadcasts, vides significant advantages over competition with it. Under while information conveyance communications are typically normal circumstances, a well developed ad hoc network sys point-to-point. Prior studies typically presume a single trans tem can present as a formidable coordinated competitor for ceiver, with a single omnidirectional antenna, operating 45 access to contested bandwidth by other systems, while within according to in-band protocol data, for all communications. the network, economic Surplus is optimized. Thus, a node The tradeoff made in limiting system designs according to presented with a communications requirement is presented these presumptions should be clear. not with the simple choice to participate or abstain, but rather It is the general self-interest of a node to conserve its own whether to participate in an ad hoc network with predicted resources, maintain an opportunity to access network 50 stability and mutual benefit, or one with the possibility of resources, while consuming whatever resource of other nodes failure due to selfish behavior, and non-cooperation. Even in as it desires. Clearly, this presents a significant risk of the the absence of a present communication requirement, a net “tragedy of the commons', in which selfish individuals fail to work which rewards cooperative behavior may be preferable respect the very basis for the community they enjoy, and a to one which simply expects altruism without rewarding it. network of rational nodes operating without significant 55 The protocol may also encompass the concept of node incentives to cooperate would likely fail. On the other hand, if reputation, that is, a positive or negative statement by others donating a node's resources generated a sufficient associated regarding the node in question. P. Michiardi and R. Molva. benefit to that node, while consuming network resources Core: A collaborative reputation mechanism to enforce node imposed a sufficient cost, stability and reliability can be cooperation in mobile ad hoc networks. In Communication achieved. So long as the functionality is sufficient to meet the 60 and Multimedia Security 2002 Conference, 2002. This repu need, and the economic Surplus is “fairly allocated, that is, tation may be evaluated as a parameter in an economic analy the cost incurred is less than the private value of the benefit, sis, or applied separately, and may be anecdotal or statistical. and that cost is transferred as compensation to those burdened In any case, if access to resources and payments are made in an amount in excess of their incremental cost, adoption of dependent on reputation, nodes will be incentivized to main the system should increase stability. In fact, even outside of 65 tain a good reputation, and avoid generating a bad reputation. these bounds, the system may be more stable than one which Therefore, by maintaining and applying the reputation in a neither taxes system use nor rewards altruistic behavior. manner consistent with the community goals, the nodes are US 8,874,477 B2 35 36 compelled to advance those goals in order to benefit from the Feigenbaum, Joan, Christos Papadimitriou, Rahul Sami, and community. Game theory distinguishes between good repu Scott Shenker (2002). “A BGP-based Mechanism for Lowest tation and bad reputation. Nodes may have a selfish motiva Cost Routing.” In Proc. 21st Symposium on Principles of tion to assert that another node has a bad reputation, while it Distributed Computing, ACM Press, 173-182; J. Feigenbaum would have little selfish motivation, absent collusion, for and S. Shenker. Distributed algorithmic mechanism design: undeservedly asserting a good reputation. On the other hand, Recent results and future directions. In Proc. 6th Int’l Work a node may have a selfish motivation in failing to reward shop on Discrete Algorithms and Methods for Mobile Com behavior with a good reputation. puting and Communications, pages 1-13, Atlanta, Ga., Sep Economics and reputation may be maintained as orthogo tember 2002: Nisan, N. and A. Ronen (2000). nal considerations, since the status of a node's currency 10 Computationally Feasible VCG Mechanisms, to be presented account provides no information about the status of its repu at “Games 2000.’ http:/www.cs.huji.ac.il/-noam/vcg tation. based.ps; Tuomas Sandholm. Limitations of the Vickrey Auc This reputation parameter may be extended to encompass tion in Computational Multiagent Systems. In Proceedings of respect, that is, a Subjective deference to another based on an the 2nd International Conference on Multi-Agent Systems asserted or imputed entitlement. While the prior system uses 15 (ICMAS). AAAI Press, 1996. Menlo Park, Calif.; C. Jason reputation as a factor to ensure compliance with system rules, Woodard and David C. Parkes, 1st Workshop on the Econom this can be extended to provided deferential preferences ics of P2P systems, Strategy proof Mechanisms for Ad Hoc either within or extrinsic to an economy. Thus, in a military Network Formation, 2003, www.sims.berkeley.edu/research/ hierarchy, a relatively higher ranking official can assert rank, conferences/p2pecon/papers/so-woodard.pdf: D. C. Parkes. and if accepted, override a relatively lower ranking bidder at Iterative Combinatorial Auctions: Achieving Economic and the same economic bid. For each node, an algorithm is pro Computational Efficiency (Chapter 2). PhD thesis, University vided to translate a particular assertion of respect (i.e., rank of Pennsylvania, May 2001. www.eecs.harvard.edu/parkes/ and chain of command) into an economic perturbation. For pubs/ch2.ps. example, in the same chain of command, each difference in In the classic case, each bidder bids a value vector for each rank might be associated with a 25% compounded discount, 25 available combination of goods or services. The various com when compared with other bids, i.e. ponents and associated ask price are evaluated combinatori ally to achieve the minimum Sum to meet the requirement. The winning bid set is that which produces the maximum Wherein B1 is the attributed bid, B0 is the actual bid, and value of the accepted bids, although the second (Vickrey) AR is the difference in rank, positive or negative. 30 price is paid. In theory, the Vickrey price represents the maxi Outside the chain of command, a different, generally mum state of the network absent the highest bidder, so that lower, discount (dNCOC) may be applied, possibly with a each bidder is incentivized to bit its private value, knowing base discount as compared to all bids within the chain of that its pricing will be dependent not on its own value, but the command (dCOC), i.e., subjective value applied by others. In the present context, 35 each offeror submits an ask price (reserve) or evaluatable B=Bo10(1+dCOC+dNCOCAR). value function for a component of the combination. If the The discount is applied so that higher ranking officers pay minimum aggregate to meet the bid requirement is not met, less, while lower ranking officers pay more. Clearly, there is the auction fails. If the auction is successful, then the set of a high incentive for each bid to originate from the highest offerors selected is that with the lowest aggregate bid, and available commander within the chain of command, and 40 they are compensated that amount. given the effect of the perturbation, for ranking officers to The VCG auction is postulated as being optimal for allo “pull rank” judiciously. cation of multiple resources between agents. It is “strat The Modified VCG Auction egyproof and efficient, meaning that it is a dominant strategy A so-called Vickrey–Clarke-Groves, or VCG, auction, is a for agents to report their true valuation for a resource, and the type of auction Suitable forbidding, in a single auction, for the 45 result of the optimization is a network which maximizes the goods or services of a plurality of offerors, as a unit. Vickrey, value of the system to the agents. Game theory also allows an W. (1961). Counterspeculation, auctions, and competitive allocation of cost between various recipients of a broadcast or sealed tenders, Journal of Finance 16, 8-37; Clarke, E. H. multicast. That is, the communication is of value to a plurality (1971). Multipart pricing of public goods, Public Choice 11, of nodes, and a large set of recipient nodes may efficiently 17-33; Felix Brandt and Gerhard Weifs. Antisocial Agents 50 receive the same information. This allocation from multiple and Vickrey Auctions. In Pre-proceedings of the Eighth Inter bidders to multiple sellers is a direct extension of VCG theory, national Workshop on Agent Theories, Architectures, and and a similar algorithm may be used to optimize allocation of Languages (ATAL-2001), pages 120-132, 2001; Tuomas costs and benefit. Sandholm. Limitations of the Vickrey Auction in Computa The principal issue involved in VCG auctions is that the tional Multiagent Systems. In Proceedings of the 2nd Inter 55 computational complexity of the optimization grows with the national Conference on Multi-Agent Systems (ICMAS). number of buyers and their different value functions and AAAI Press, 1996. Menlo Park, Calif.; Ron Lavi, Ahuva allocations. While various simplifying presumptions may be Mu'alem, and Noam Nisan, “Towards a Characterization of applied, studies reveal that these simplifications may under Truthful Combinatorial Auctions, citeseerist.psu.edu/ mine the VCG premise, and therefore do not promote honesty laviO3towards.html; Moulin, H. (1999). Incremental cost 60 in reporting the buyer's valuation, and thus are not “strat sharing; characterization by Strategyproofness, Social Choice egyproof, which is a principal advantage of the VCG pro and Welfare 16, 279-320; Moulin, H. and S. Shenker (1997). CCSS, Strategyproof Sharing of Submodular Costs: Budget Balance The Surplus, i.e., gap between bid and ask, is then available Versus Efficiency, to appear in Economic Theory.www.aciri to compensate the deferred bidders. This surplus may be, for .org/shenker/cost.ps; Moulin, Hervé, and Scott Shenker 65 example, distributed proportionately to the original bid value (2001). “Strategy proof Sharing of Submodular Costs: Budget of the bidder, thus further encouraging an honest valuation of Balance versus Efficiency.” Economic Theory 18, 5.11-533; control over the resource. Thus, if we presume that a bidder US 8,874,477 B2 37 38 may have an incentive to adopt a strategy in which it shaves its communications can occur in a lightly loaded network even if bid to lower values, an additional payoff dependent on a the content communications are Saturated. An initial node higher value bid will promote higher bides and disincentivize transmits using an adaptive power protocol, to achieve an shaving. On the other hand, it would be inefficient to promote effective transmit range, for example, of greater than an aver bidding above a bidder's private value, and therefore care age internodal distance, but not encompassing the entire net must be exercised to generally avoid this circumstance. In work. This distance therefore promotes propagation to a set of similar manner, potential offerors may be compensated for nearby nodes, without unnecessarily interfering with com low bids, to promote availability of supply. It is noted that, by munications of distance nodes and therefore allowing this broadcasting Supply and demand, fault tolerance of the net task to be performed in parallel in different regions. Neigh work is improved, since in the event that an involved node 10 boring nodes also transmit in Succession, providing sequen becomes unavailable, a competing node or set of nodes for tial and complete protocol information propagation over a that role may be quickly enlisted. relevance range, for example 3-10 maximum range hops. The optimization is such that, if any offeror asks an amount If we presume that there is a spatial limit to relevance, for that is too high, it will be bypassed in favor of more “reason example, 5 miles or 10 hops, then the network State propaga able' offerors. Since the bidderpays the second highest price, 15 tion may be so limited. Extending the network to encompass honesty in bidding the full private value is encouraged. The a large number of nodes will necessarily reduce the tractabil distribution of the surplus to losing bidders, which exercise ity of the optimization, and incur an overhead which may be deference to the winner, is proportional to the amount bid, inefficient. Each node preferably maintains a local estimate of that is, the reported value. relevance. This consideration is accommodated, along with a In a scenario involving a request for information meeting desire to prevent exponential growth in protocol-related data specified criteria, the auction is complicated by the fact that traffic, by receiving an update from all nodes within a node's the information resource content is unknown to the recipient, network relevance boundary, and a state variable which rep and therefore the bid is blind, that is, the value of the infor resents an estimate of relevant status beyond the arbitrarily mation to the recipient is indeterminate. However, game defined boundary. The propagation of network State may thus theory Supports the communication of a value function or 25 conveniently occur over a finite number of hops, for example utility function, which can then be evaluated at each node 3-10. In a dense population of nodes, such as in a city, even a possessing information to be communicated, to normalize its single maximum range communication may result in a large value to the requestor. Fortunately, it is a dominant strategy in number of encompassed nodes. On the other hand, in a a VCG auction to communicate a truthful value, and therefore deserted environment, there may be few or no communica broadcasting the private value function, to be evaluated by a 30 tions partners, at any time. recipient, is not untenable. In a mere request for information Under conditions of relatively high nodal densities, the conveyance, such as the intermediate transport nodes in a System may employ a Zone strategy, that is, proximate groups multihop network, or in a cellular network infrastructure of nodes are is treated as an entity or cluster for purposes of extension model, the bid may be a true (resolved) value, since external state estimation, especially with respect to distant the information content is not the subject of the bidding: 35 nodes or Zones. In fact, a Supernode may be nominated within rather it is the value of the communications per se, and the a cluster to control external communications for that cluster. bidding node can reasonably value its bid. Such a presumption is realistic, since at extended distances, Game theory also allows an allocation of cost between geographically proximate nodes may be modeled as being various recipients of a broadcast or multicast. That is, in many similar or inter-related, while at close distances, and particu instances, information which is of value to a plurality of 40 larly within a Zone in which all nodes are in direct commu nodes, and a large set of recipient nodes may efficiently nication, internode communications may be subject to mutual receive the same information. This allocation is a direct interference, and can occur without Substantial external influ extension of VCG theory. ence. Alternately, it is clear that to limit latencies and com Operation of Protocol munication risks, it may be prudent to bypass nearby and The preferred method for acquiring an estimate of the state 45 neighboring nodes, thus trading latency for power consump of the network is through use of a proactive routing protocol. tion and overall network capacity. Therefore, a hierarchal Thus, in order to determine the network architecture state, scheme may be implemented to geographically organize the each node must broadcast its existence, and, for example, a network at higher analytical levels, and geographic cells may payload of information including its identity, location, itiner cooperate to appear externally as a single coordinated entity. ary (navigation vector) and “information value function'. 50 In order to estimate a network edge condition, a number of Typically, the system operates in a continuous set of states, so presumptions must be made. The effect of an inaccurate esti that it is reasonable to commence the process with an estimate mate of the network edge condition typically leads to ineffi of the state based on prior information. Using an in-band or ciency, while inordinate efforts to accurately estimate the out-of-band propagation mechanism, this information must networkedge condition may also lead to inefficiency. Perhaps propagate to a network edge, which may be physically or 55 the best way to achieve compromise is to have a set of adap artificially defined. If all nodes operate with a substantially tive presumptions or rules, with a reasonable starting point. common estimation of network topology, only deviations For example, in a multihop network, one might arbitrarily set from previously propagated information need be propagated. a network edge the maximum range of five hops of adminis On the other hand, various nodes may have different esti trative data using a 95% reliable transmission capability. mates of the network State, allowing efficiency gains through 60 Beyond this range, a set of state estimators is provided by exploitation of Superior knowledge as compared with seeking each node for its Surroundings, which are then communicated to convey full network state information to each node. up to five hops (or the maximum range represented by five A CSMA scheme may be used for the protocol-related hops). This state estimator is at least one cycle old, and by the communications because it is relatively simple and robust, time it is transferred five hops away, it is at least six cycles old. and well Suited for ad hoc communications in lightly loaded 65 Meanwhile, in a market economy, each node may respond to networks. We presume that the network is willing to tolerate perceived opportunities, leading to a potential for oscillations a protocol related inefficiency, and therefore that protocol if a time-element is not also communicated. Thus, it is pre US 8,874,477 B2 39 40 ferred that the networkedge state estimators represent a time channels, in order to have a better chance to disseminate the prediction of network behavior under various conditions, control information to potentially interfering sources, and to rather than a simple Scalar value or instantaneous function. allow coordination of nodes more globally. For example, each node may estimate a network Supply In order to effectively provide decentralized control, either function and a network demand function, liquidity estimate each node must have a common set of information to allow and bid-ask gap for its environment, and its own Subjective execution of an identical control algorithm, or nodes defer to risk tolerance, if separately reported; the impact of nodes the control signals of other nodes without internal analysis for closer than five hops may then be subtracted from this esti optimality. A model of semi-decentralized control is also mate to compensate for redundant data. Further, if traffic known, in which dispersed Supernodes are nominated as mas routes are identifiable, which would correspond in a physical 10 ter, with other topologically nearby nodes remaining as slave setting of highways, fixed infrastructure access points, etc., a nodes. In the pure peer network, relatively complete informa state estimator for these may be provided as well. As dis tion conveyance to each node is required, imposing a rela cussed above, nodes may bid not only for their own needs or tively high overhead. In a master-slave (or Supernode) archi resources, but also to act as market-makers or merchants, and tecture, increased reliance on a single node trades-off may obtain long term commitments (futures and/or options) 15 reliability and robustness (and other advantages of pure peer and employ risk reduction techniques (insurance and/or to-peer networks) for efficiency. A supernode within a cellu indemnification), and thus may provide not only an estimate lar Zone may be selected for its Superior capability, or perhaps of network conditions, but also 'guaranty this state. is at a central location or is immobile. A node seeking to communicate within the five hop range Once each control node (node or Supernode) has an esti needs to consider the edge State estimate only when calculat mate of network topology, the next step is to optimize net ing its own Supply and demand functions, bearing in mind work channels. According to VCG theory, each agent has an competitive pressures from outside. On the other hand, nodes incentive to broadcast its truthful value or value function for seeking resources outside the five hop range must rely on the the scarce resource, which in this case, is control over com estimate, because a direct measurement or acquisition of munications physical layer, and or access to information. This information would require excess administrative communi 25 communication can be consolidated with the network discov cations, and incur an inefficient administrative transaction. ery transmission. Each control node then performs a combi Thus, a degree of trust and reliance on the estimate may natorial Solution to select the optimum network configuration ensue, wherein a node at the arbitrary network edge is desig from the potentially large number of possibilities, which may nated as an agent for the principal in procuring or selling the include issues of transmit power, data rate, path, timing, reli resource beyond its own sphere of influence, based on the 30 ability and risk criteria, economic and virtual economic costs, provided parameters. The incentive for a node to provide multipath and redundancy, etc., for the set of simultaneous misinformation is limited, since nodes with too high a equations according to VCG theory (or extensions thereof). reported estimate value lose gains from competitive sale This solution should be consistent between all nodes, and the transactions, and indeed may be requested to be buyers, and effects of inconsistent solutions may be resolved by collision vice versa. While this model may compel trading by interme 35 sensing, and possibly an error/inconsistency detection and diary nodes, if the information communicated accurately rep correction algorithm specifically applied to this type of infor resents the network state, an economic advantage will accrue mation. Thus, if each node has relatively complete informa to the intermediary participating, especially in a non-power tion, or accurate estimates for incomplete information, then constrained, unlicensed spectrum node configuration. each node can perform the calculation and derive a closely It should be borne in mind that the intended administration 40 corresponding solution, and Verify that solutions reported by of the communications is an automated process, with little others are reasonably consistent to allow or promote reliance human involvement, other than setting goals, risk tolerance, thereon. cost constraints, etc. In a purely virtual economy with tem As part of the network mapping, communications impair porally declining currency value, the detriment of inaccurate ment and interference sources are also mapped. GPS assis optimizations is limited to reduced nodal efficiency, and with 45 tance may be particularly useful in this aspect. Where net appropriate adaptivity, the system can learn from its “mis work limitations are caused by interfering communications, takes”. (A defined decline in currency value tends to define the issue is a determination of a strategy of deference or the cost constraints for that node, since wealth cannot be competition. If the interfering communication is continuous accumulated nor overspent). or unresponsive, then the only available strategy is competi A Supernode within a Zone may be selected for its Superior 50 tion. On the other hand, when the competing system uses, for capability, or perhaps a central location. The Zone is defined example, a CSMA system, such as 802.11, competition with by a communication range of the basic data interface for Such a communication simply leads to retransmission, and communications, with the control channel preferably having therefore ultimately increased network load, and a deference a longer range, for example at least double the normal data strategy may be more optimal, at least and until it is deter communications range. Communications control channel 55 mined that the competing communication is incessant. Other transmitters operate on a number of channels, for example at communications protocols, however may have a more or less least 7, allowing neighboring Zones in a hexagonal tiled array aggressive strategy. By observation of a system over time, its to communicate simultaneously without interference. In a strategies may be revealed, and game theory permits compo geographic Zone system, alternate Zones which would other sition of an optimal strategy to deal with interference or wise be interfering may use an adaptive multiplexing scheme 60 coexistence. It is noted that this strategy may be adopted to avoid interference. All nodes may listen on all control adaptively by the entire ad hoc network, which may coordi channels, permitting rapid analysis and propagation of con nate deference or competition as determined optimal. trol information. As discussed elsewhere herein, directional The optimization process produces a representation of antennas of various types may be employed, although it is optimal network architecture during the Succeeding period. preferred that out-of-band control channels employ omnidi 65 That is, value functions representing bids are broadcast, with rectional antennas, having a generally longer range (and the system then being permitted to determine an optimal real lower databandwidth) than the normal data communications valuation and distribution of that value. Thus, prior to US 8,874,477 B2 41 42 completion of the optimization, potentially inconsistent allo decrement, or other variable, rather than hops. Likewise, the cations must be prevented, and each node must communicate range may be adaptively determined, rather than predeter its evaluation of other node's value functions, so that the mined, based on Some criteria. optimization is performed on a normalized economic basis. After propagation, each node evaluates the set of value This step may substantially increase the system overhead, and functions for its community, with respect to its own informa is generally required for completion of the auction. This valu tion and ability to forward packets. Each node may then make ation may be inferred, however, for intermediate nodes in a an offer to supply or forward information, based on the pro multihop network path, since there is little subjectivity for vided information. In the case of multihop communications, nodes solely in this role, and the respective value functions the offers are propagated to the remainder of the community, may be persistent. For example, the valuation applied by a 10 for the maximum number of hops, including the originating node to forward information is generally independent of con node. At this point, each node has a representation of the State tent and involved party. of its community, with community edge estimates providing A particular complication of a traffic information system is consistency for nodes with differing community Scopes, the that the nature of the information held by any node is private valuation function each node assigns to control over portions to that node (before transmission), and therefore the valuation 15 of the network, as well as a resolved valuation of each node is not known until after all bids are evaluated. Thus, prior to for Supplying the need. Under these circumstances, each node completion of optimization, each node must communicate its may then evaluate an optimization for the network architec evaluation of other nodes value functions, so that the opti ture, and come to a conclusion consistent with that of other mization is performed on an economic basis. This required members of its community. If supported, node reputation may step Substantially increases the system overhead. This valua be updated based on past performance, and the reputation tion may be inferred, however, for transit nodes in a multihop applied as a factor in the optimization and/or externally to the network path. optimization. As discussed above, a VCG-type auction is As discussed above, may of the strategies for making the employed as a basis for optimization. Since each node economic markets more efficient may be employed either receives bid information from all other nodes within the directly, or analogy, to the virtual economy of the ad hoc 25 maximum node count, the VCG auction produces an opti network. The ability of nodes to act as market maker and mized result. derivative market agents facilitates the optimization, since a As discussed above, by permitting futures, options, deriva node may elect to undertake a responsibility (e.g., transaction tives, insurance/indemnification/guaranties, long and short risk), rather than relay to others, and therefore the control/ sales, etc., the markets may be relatively stabilized as com administrative channel chain may be truncated at that point. If 30 pared to a simple set of independent and sequential auctions, the network is dense, then a node which acts selfishly will be which may show increased Volatility, oscillations, chaotic bypassed, and if the network is sparse, the node may well be behavior, and other features which may be inefficient. entitled to gain transactional profit by acting as a principal and Transmissions are preferably made in frames, with a single trader, subject to the fact that profits will generally be subop bidding process controlling multiple frames, for example a timal if pricing is too high or too low. 35 multiple of the maximum number of hops. Therefore, the bid After the network architecture is defined, compensation is encompasses a frame's-worth of control over the modalities. paid to those nodes providing value or subjected to a burden In the event that the simultaneous use of, or control over, a (including foregoing communication opportunity) by those modality by various nodes is not inconsistent, then the value gaining a benefit. The payment may be a virtual currency, of the respective nodes may be Summed, with the resulting with no specific true value, and the virtual currency system 40 allocation based on, for example, a ratio of the respective provides a convenient method to flexibly tax, subsidize, or value functions. As a part of the optimization, nodes are control the system, and thus steer the virtual currency to a rewarded not only for Supporting the communication, but also normalized extrinsic value. In a real currency system, exter for deferring their own respective communications needs. As nal controls are more difficult, and may have unintended a result, after controlling the resources, a node will be rela consequences. A hybrid economy may be provided, linking 45 tively less wealthy and less able to subsequently control the both the virtual and real currencies, to some degree. This is resources, while other nodes will be more able to control the especially useful if the network itself interfaces with an out resources. The distribution to deferred nodes also serves to side economy, such as the cellular telephony infrastructure prevent pure reciprocal communications, since the proposed (e.g., 2G, 2.5G, 3G, 4G, proposals for 5G, WiFi (802.11x) mechanism distributes and dilutes the wealth to deferring hotspots, WiMax (802.16x), etc.) 50 nodes. Using the protocol communication system, each node Another possible transaction between nodes is a loan, that transmits its value function (or change thereof), passes is, instead of providing bandwidth perse, one node may loan through communications from neighboring nodes, and may, a portion of its generator function or accumulated wealth to for example transmit payment information for the immediate another node. Presumably, there will be an associated interest past bid for incoming communications. 55 payment. Since the currency in the preferred embodiment is Messages are forwarded outward (avoiding redundant itself defined by an algorithm, the loan transaction may also propagation back to the source), with messages appended be defined by an algorithm. While this concept is somewhat from the series of nodes. Propagation continues for a finite inconsistent with a virtual currency which declines in value number of hops, until the entire community has an estimate of over time and/or space, it is not completely inconsistent, and, the state and value function of each node in the community. 60 in fact, the exchange may arbitrage these factors, especially Advantageously, the network beyond a respective community location-based issues. may be modeled in simplified form, to provide a better esti Because each node in the model presented above has com mate of the network as a whole. If the propagation were not plete information, for a range up to the maximum node count, reasonably limited, the information would be stale by the time the wealth of each node can be estimated by its neighbors, and it is employed, and the system latency would be inordinate. 65 payment inferred even if not actually consummated. (Failure Of course, in networks where a large number of hops are of payment can occur for a number of reasons, including both realistic, the limit may be time, distance, a counter or value malicious and accidental). Because each hop adds significant US 8,874,477 B2 43 44 cost, the fact that nodes beyond the maximum hop distance cagal.02minimumenergy.html; N. Ben Salem, L. Buttyan, are essentially incommunicado is typically of little conse J.P. Hubaux, and Jakobsson M. A charging and rewarding quence; since it is very unlikely that a node more than 5 or 10 scheme for packet forwarding. In Proceedings of MobiHOC, hops away will be efficiently directly included in any com June 2003. For example, the virtual currency may have a munication, due to the increasing cost with distance, as well half-life or temporally declining value. On the other hand, the as reduction in reliability and increase in latency. Thus, large value may peak at a time after generation, which would area and scalable networks may exist. encourage deference and short term savings, rather than Communications are generally of unencrypted data. immediate spending, and would allow a recipient node to Assuming the network is highly loaded, this may allow a node benefit from virtual currency transferred before its peak to incidentally fulfill its data requirements as a bystander, and 10 value. This also means that long term hoarding of the cur thus at low cost meet its needs, allowing nodes with more rency is of little value, since it will eventually decay in value, urgent or directed needs to both control and compensate the while the system presupposes a nominal rate of spending, network. While this may reduce compensation to intermedi which is normalized among nodes. The variation function aries and data sources, the improvements in efficiency will may also be adaptive, but this poses a synchronization issue likely benefit the network as a whole in increase stability, 15 for the network. An external estimate of node wealth may be since we assume that peak load conditions will occur fre used to infer counterfeiting, theft and failure to pay debts, and quently. to further effect remediation. Enforcement of responsibility may be provided by a cen The currency is generated and Verified in accordance with tralized system which assures that the transactions for each micropayment theory. Rivest, R. L., A. Shamir, Pay Word and node are properly cleared, and that non-compliant nodes are MicroMint: Two simple micropayment schemes, also pre either excluded from the network or at least labeled. Whilean sented at the RSA 96 conference, http//theory.lcs.mit.edu/ automated clearinghouse which periodically ensures nodal rivest/RivestShamirmpay-ps, citeseerist.psu.edu/ compliance is preferred, a human discretion clearinghouse, rivest96payword.html; Silvio Micali and Ronald Rivest. for example presented as an arbitrator or tribunal, may be Micropayments revisited. In Bart Preneel, editor, Progress in employed. 25 Cryptology—CT-RSA 2002, volume 2271 of Lecture Notes The Synthetic Economy in Computer Science. Springer-Verlag, Feb. 18-22, 2002. Exerting external economic influences on the system may citeseerist.psu.edu/micali02micropayments.html. have various effects on the optimization, and may exacerbate Micropayment theory generally encompasses the transfer differences in subjective valuations. The application of a of secure tokens (e.g., cryptographically endorsed informa monetary value to the virtual currency Substantially also 30 tion) having presumed value, which are intended for verifi increases the possibility of misbehavior and external attacks. cation, ifatall, in a non-real time transaction, after the transfer On the other hand, a virtual currency with no assessed real to the recipient. The currency is circulated (until expiration) value is self-normalizing, while monetization leads to exter as a token, and therefore may not be subject to immediate nal and generally irrelevant influences as well as possible definitive authentication by Source. Since these tokens may external arbitrage (with potential positive and negative 35 be communicated through an insecure network, the issue of effects). External economic influences may also lead to ben forcing allocation of payment to particular nodes may be efits, which are discussed in various publications on non-zero dealt with by cryptographic techniques, in particular public Sum games. key cryptography, in which the currency is placed in a cryp In order to provide fairness, the virtual currency (similar to tographic “envelope” (cryptolope) addressed to the intended the so-called “nuglets” or “nuggets' proposed for use in the 40 recipient, e.g., is encrypted with the recipient’s public key, Terminodes project) is self-generated at each node according which must be broadcast and used as, or in conjunction with, to a schedule, and itself may have a time dependent value. L. a node identifier. This makes the payment unavailable to other Blazevic, L. Buttyan, S. Capkun, S. Giordiano, J.-P. Hubaux, than the intended recipient. The issue of holding the and J.-Y. Le Boudec. Self-organization in mobile ad-hoc net encrypted token hostage and extorting a portion of the value works: the approach of terminodes. IEEE Communications 45 to forward the packet can be dealt with by community pres Magazine, 39(6):166-174, June 2001; M. Jakobsson, J.P. Sure, that is, any node presenting this (or other undesirable) Hubaux, and L. Buttyan. A micro-payment scheme encour behavior might be ostracized. The likelihood of this type of aging collaboration in multi-hop cellular networks. In Pro misbehavior is also diminished by avoiding monetization of ceedings of Financial Crypto 2003, January 2003; J.P. the virtual currency. Further, redundant routing of such infor Hubaux, et al., “Toward Self-Organized Mobile Ad Hoc Net 50 mation may prevent single-node control over Such commu works: The Terminodes Project', IEEE Communications, nications. 39(1), 2001. citeseer.ist.psu.edu/hubaux01 toward.html; But This currency generation and allocation mechanism gen tyan, L., and Hubaux, J.-P. Stimulating Cooperation in Self erally encourages equal consumption by the various nodes Organizing Mobile Ad Hoc Networks. Tech. Rep. DSC/cite over the long term. In order to discourage excess consumption seerist.psu.edu/buttyan01stimulating.html; Levente Buttyan 55 of bandwidth, an external tax may be imposed on the system, and Jean-Pierre Hubaux, “Enforcing Service Availability in that is, withdrawing value from the system based on usage. Mobile Ad-Hoc WANs' 1st IEEE/ACM Workshop onMobile Clearly, the effects of such a tax must be carefully weighed, Ad Hoc Networking and Computing (MobiHOC citese since this will also impose an impediment to adoption as erist.psu.edu/buttyan00enforcing.html; L. Buttyan and J.-P. compared to an untaxed system. On the other hand, a similar Hubaux. Nuglets: a virtual currency to stimulate cooperation 60 effect use-disincentive may be obtained by rewarding low in self-organized ad hoc networks. Technical Report DSC/ consumption, for example by allocating an advertising Sub 2001, citeseerist.psu.edu/article/buttyan01 nuglets.html; sidy between nodes, or in reward of deference. The external Mario Cagal, Jean-Pierre Hubaux, and Christian Enz. Mini tax, if associated with efficiency-promoting regulation, may mum-energy broadcast in all-wireless networks: NP-com have a neutral or even beneficial effect. pleteness and distribution issues. In The Eighth ACM Inter 65 Each node computes a value function, based on its own national Conference on Mobile Computing and Networking knowledge state, risk profile and risk tolerance, and wealth, (MobiCom 2002), citeseerist.psu.edu/ describing the value to it of additional information, as well as US 8,874,477 B2 45 46 its own value for participating in the communications of payment inferred even if not actually consummated. (Failure others. The value function typically includes a past travel of payment can occur for a number of reasons, including both history, future travel itinerary, present location, recent com malicious and accidental). Because each hop adds significant munication partners, and an estimator of information strength cost, the fact that nodes beyond the maximum hop distance and weakness with respect to the future itinerary. It may be are essentially incommunicado is typically of little conse presumed that each node has a standard complement of sen quence; since it is very unlikely that a node more than 5 or 10 sors, and accurately acquired descriptive data for its past hops away will be efficiently included in any communication, travel path. Otherwise, a description of the available informa due to the increasing cost with distance, as well as reduction tion is required. One advantage of a value function is that it in reliability and increase in latency. Thus, large area and changes little over time, unless a need is satisfied or circum 10 Scalable networks may exist. stances change, and therefore may be a persistent attribute. Typically, cryptography is employed for both authentica Using the protocol communication system, each node tion and to preserve privacy. External regulation, in a legal transmits its value function (or change thereof), passes sense at least, is typically imposed by restrictions on hard through communications from neighboring nodes, and may, ware and Software design, as well as Voluntary compliance at for example transmit payment information for the immediate 15 risk of detection and legal sanction. past bid for incoming communications. A synthetic economy affords the opportunity to provide Messages are forwarded outward (avoiding redundant particular control over the generator function, which in turn propagation back to the source), with messages appended Supports a hierarchy. In this scheme, each node controls the from the series of nodes. Propagation continues for a finite generator function at respectively lower nodes, and thus can number of hops, until the entire community has an estimate of allocate wealth among Subordinates. If one assumes real time the state and value function of each node in the community. communications, then it is clear that the Superordinate node Advantageously, the network beyond a respective community can directly place bids on behalf of subordinates, thus effec may be modeled in simplified form, to provide a better esti tively controlling its entire branch. In the absence of real time mate of the network as a whole. communications, the Superordinate node must defer to the After propagation, each node evaluates the set of value 25 discretion of the subordinate, subject to reallocation later if functions for its community, with respect to its own informa the Subordinate defects. If communications are impaired, and tion and ability to forward packets. Each node may then make a set of a priori instructions are insufficient, then it is up to the an offer to supply or forward information, based on the pro Subjective response of a node to provide deference. vided information. In the case of multihop communications, It is noted that when sets of nodes “play favorites”, the the offers are propagated to the remainder of the community, 30 VCG auction will no longer be considered “strategy proof. for the maximum number of hops, including the originating The result is that bidders will assume bidding strategies that node. At this point, each node has a representation of the state do not express their secret valuation, with the result being of its community, with community edge estimates providing likely suboptimal market finding during the auction. This consistency for nodes with differing community Scopes, the factor can be avoided if hierarchal overrides and group bid valuation function each node assigns to control over portions 35 ding play only a small role in the economy, and thus the of the network, as well as a resolved valuation of each node expected benefits from shaded bidding are outweighed by the for Supplying the need. Under these circumstances, each node normal operation of the system. For example, by taxing trans may then evaluate an optimization for the network architec actions, over-valued bidding will be disincentivized, and by ture, and come to a conclusion consistent with that of other redistributing economic Surplus to bystanders, the aggregate members of its community. If supported, node reputation may 40 wealth of the controlling group will be mitigated. be updated based on past performance, and the reputation A synthetic economy affords the opportunity to provide applied as a factor in the optimization and/or externally to the particular control over the generator function, which in turn optimization. As discussed above, a VCG-type auction is provides particular advantages with respect to a hierarchal employed as a basis for optimization. Since each node organization. In this scheme, each node has the ability to receives bid information from all other nodes within the 45 control the generator function at respectively lower nodes, maximum node count, the VCG auction produces an opti and thus can allocate wealth among Subordinates. If one mized result. assumes real time communications, then it is clear that the Transmissions are made in frames, with a single bidding superordinate node can directly place bids on behalf of sub process controlling multiple frames, for example a multiple ordinates, thus effectively controlling its entire branch. In the of the maximum number of hops. Therefore, the bid encom 50 absence of real time communications, the Superordinate node passes a frame's-worth of control over the modalities. In the must defer to the discretion of the subordinate, subject to event that the simultaneous use of, or control over, a modality reallocation later if the subordinate defects. If communica by various nodes is not inconsistent, then the value of the tions are impaired, and a set of a priori instructions are insuf respective nodes may be summed, with the resulting alloca ficient, then it is up to the Subjective response of a node to tion based on, for example, a ratio of the respective value 55 provide deference. Thus, a node may transfer all or a portion functions. As a part of the optimization, nodes are rewarded of its generator function, either for a limited time or perma not only for Supporting the communication, but also for defer nently, using feed-forward or feedback control. In this sense, ring their own respective needs. As a result, after controlling the hierarchal and financial derivatives, options, futures, the resources, a node will be relatively less wealthy and less loans, etc. embodiments of the invention share a common able to subsequently control the resources, while other nodes 60 theme. will be more able to control the resources. The distribution to It is noted that when sets of nodes “play favorites”, the deferred nodes also serves to prevent pure reciprocal commu VCG auction will no longer be considered “strategy proof. nications, since the proposed mechanism distributes and The result is that bidders will assume bidding strategies that dilutes the wealth to deferring nodes. do not express their secret valuation, with the result being Because each node in the model presented above has com 65 likely suboptimal market price finding during the auction. plete information, for a range up to the maximum node count, This factor can be avoided if hierarchal overrides and group the wealth of each node can be estimated by its neighbors, and bidding play only a small role in the economy, and thus the US 8,874,477 B2 47 48 expected benefits from shaded bidding are outweighed by the cated a priori. Likewise, wealth may also be reallocated, normal operation of the system. On the other hand, the present although a penalty is incurred in the event of an initial mis invention potentially promotes competition within branches allocation since the transfer itself incurs a cost, and there will of a hierarchy, to the extent the hierarchy does not prohibit be an economic competitive distortion, under which a node's this. Between different branches of a hierarchy, there will 5 subjective value of a resource is influenced by its subjective generally be full competition, while within commonly con wealth. If a node is supplied with wealth beyond its needs, the trolled branches of a hierarchy, cooperation will be expected. wealth is wasted, since it declines in value and cannot be Since the competitive result is generally more efficient, there hoarded indefinitely. If a node is supplied with insufficient will be incentive for the hierarchal control to permit compe wealth, economic Surplus through transactional gains are lost. tition as a default state, asserting control only where required 10 Thus, each node must analyze its expected circumstances to for the hierarchal purpose. retain or delegate the generator function, and to optimally Military Hierarchy allocate wealth between competing Subordinates. In a typical auction, each player is treated fairly; that is, the In any transaction, there will be a component which repre same rules apply to each player, and therefore a single sents the competitive “cost', and a possible redistribution economy describes the process. The fair auction therefore 15 among nodes within a hierarchal chain. This redistribution poses challenges for an inherently hierarchal set of users, may be of accumulated wealth, or of the generation function Such as a military organization. In the military, there is typi portion. In the former case, if the communication path fails, cally an expectation that “rank has its privileges'. The net no further transfers are possible, while in the later case, the result, however, is a decided subjective unfairness to lower result is persistent until the transfer function allocation is ranking nodes. In a mobile adhoc network, a real issue is user 20 reversed. It is also possible to transferan expiring or declining defection or non-compliance. For example, where a cost is portion of the generating function; however, this might lead a imposed on a user for participating in the adhoc network, e.g., node which is out of range to have no ability to rejoin the battery power consumption, if the anticipated benefit does not network upon return, and thus act as an impediment to effi exceed the cost, the user will simply turn off the device until cient network operation. As discussed above, one possibility actually needed, to conserve battery power outside the control 25 is for nodes to borrow or load currency. In this case, a node of the network. The result of mass defection will of course be deemed credit-worthy may blunt the impact of initially hav the instability and failure of the ad hoc network itself. Thus, ing insufficient wealth by merely incurring a transaction cost perceived fairness and net benefit is required to important for (including interest, if applied). network Success, assuming that defection or non-compliance In practice, the bulk of the wealth generating function will remain possible. 30 be widely distributed, and not concentrated at the top of the On the otherhand, in military systems, the assertion of rank hierarchy. If this is true, under most circumstances, the net as a basis for priority is not necessarily perceived as arbitrary work will appear to operate according to a non-hierarchal or and capricious, and is generally not perceived subjectively as fair VCG model, but in some circumstances, normal opera Such. Orders and communications from a central command tion may be usurped by nodes which have apparent excess are critical for the organization itself. Therefore, the difficulty 35 wealth resulting from a Superior wealth generator function. in analyzing the application of a fair game to a hierarchal Typically, hierarchically superior nodes will use their ability organization is principally a result of conceptualizing and to transfer wealth to themselves, or to recruit subordinates to aligning the individual incentives with those of the organiza cooperate, in order to directly or indirectly control the net tion as a whole. Since the organization exists outside of the ad work resources. It is possible, however, for nodes within one hoc network, it is generally not unrealistic to expect compli- 40 branch of a hierarchy to conspire against nodes outside that ance with the hierarchal attributes both within and outside of branch, resulting in a different type of distortion. Since the ad the network. hoc network typically gains by having a larger number of An artificial economy provides a basis for an economically participating nodes, this type of behavior may naturally be efficient solution. In this economy, each node has a generator discouraged. On the other hand, hierarchically Superior nodes function for generating economic units which are used in a 45 either retain, or more likely, can quickly recruit Surrounding combinatorial auction with other nodes. The economic units Subordinates to allocate their wealth generating function and may have a declining value, so that wealth does not accumu accumulated wealth to pass urgent or valuable messages. late over long periods, and by implication, wealth accumu Where expensive assets are employed, an actual transfer of lated in one region is not available for transfer in a distant wealth or the generator function to a single entity may be region, since the transfer may be subject to latency and/or 50 required. For example, a high level node might have access to cost. Even if a low latency system is employed to transfer the a high power broadcast system, which interferes with other value, an express spatially declining value function may also communications, or simply incurs a high cost to operate. Low be imposed. The geographic decline may also be explicit, for level nodes might ordinarily be limited to cellular (i.e., short example based on a GPS or navigational system. In other range, low power radio) wireless communications. In order cases, nodal motility is valuable, and mobile nodes are to be 55 for a low level node to control an expensive asset, the assent rewarded over those which are stationary. Therefore, the or cooperation of others may be required, for example by value or a portion thereof, or the generator function, may hierarchal Superiors. increase with respect to relocations. Since the network should be stable in the absence of com This scheme may be extended to the hierarchal case by mand and control communications, a hierarchal Superior treating each chain of command as an economic unit with 60 should assure that Subordinate nodes possess Sufficient respect to the generator function. At any level of the hierarchy, wealth and motivation to maintain ad hoc network operation. the commander retains a portion of the wealth generation Insufficient wealth will tend to eliminate the advantage to capacity, and delegates the remainder to its subordinates. In nodal participation (and therefore encourage defection), the case of real-time communications, a commander may unless payments from acting as intermediary are significant. directly control allocation of the generator function at each 65 Thus, a node with insufficient wealth generation function time period. Typically, there is no real-time communications may potentially exhaust its resources, and be unavailable for capability, and the wealth generator function must be allo adhoc intermediary use, even for the benefit of the hierarchy. US 8,874,477 B2 49 50 On the other hand, an initial allocation of too much wealth mensurate with the benefits provided, or else the ad hoc will encourage high spending and less active participation as network will become unreliable. an intermediary. While it is possible in a military system to While the true economics of the costs and value functions formulate an “engineered solution which forces participa for the participants are not defined, some estimates are avail tion and eliminates defection, this solution does not gain the able. The issue is: are users willing to pay for extended cel benefit of economic optimization and may have limited appli lular reach? If so, do they value the benefits commensurate cation outside of mandatory hierarchies. with the overall costs, including service fees, hardware, and Cellular Network Extension adhoc cooperative burdens? As such, care must be exercised Cellular Networks provide efficient coverage of large por to define competitive compensation or the business will be tions of the inhabited landmass. On the other hand, achieving 10 inefficient. Since this extension is driven by the cellular net complete coverage, including relatively uninhabited areas, work operator, a Suitable return on investment is mandated. may be cost inefficient or infeasible. On the other hand, there Many analyses and studies have concluded that Voluntary remains significant unmet demand for coverage of certain adhoc networks are efficient when the incentives to cooperate aaS. with the network goals are aligned and Sufficient to incentiv Generally, it is likely that a need for service arises within a 15 ize users accordingly. If the reward for cooperation is near few miles from the edge of a cellular network. That is, the optimum, then the network will benefit by increased coverage fixed infrastructure is almost in reach. On the other hand, the and reliability, each node will benefit from increased utility, infrastructure costs required to fill in gaps or marginally and intermediary nodes will specifically benefit through com extend the network may be inordinately high, for the direct pensation. Due to the technical possibility for potential inter economic benefits achieved. At present, there is no effective mediaries to fail to either participate in network administra means for remediating these gaps. tion or operation, while taking advantage of the network as a One problem arises in that the present networks generally beneficiary, the promotion of network availability as an have a threshold usage plan. All territory encompassed by a incentive for cooperation is typically itself insufficient incen network is treated as fungible, and incurs the same cost. tive to assure cooperation. The particular cost of the limited Likewise, usage of partner networks is also treated as fun 25 power resource for potential intermediaries makes non-coop gible. Therefore, the incentive to extend network reach for eration a particularly important factor. On the other hand, the any company is limited to the overall incentive for customers presumption of power cost as a critical factor may be accurate to defect to different networks, balanced against the increased only in some circumstances: In many cases, a cheap power cost of extending the network. It is in the context of this Source is available, such as in a home or office, or in a vehicle, economic problem that a solution is proposed. Quite simply, 30 making other factors more important. in the same areas where the cellular infrastructure is insuffi It is noted that, in a cellular telephone system, the reason cient and there is demand for service, it may be possible to able acts of a user which might undermine the network are implement a peer-to-peer network or multihop network to limited. Clearly, the user can choose a different network or extend a cellular network system. In fact, if we presume that provider. The user may turn off his phone or make it unavail the coverage is absent, the network extension function may 35 able. The user may abuse the service contract, taking advan make use of the licensed cellular spectrum, thus making the tage of promotions or “free” access to the detriment of others. ad hoc-cellular transceiver design more efficient, and elimi Notably, the user typically has no reasonable ability to repro nating extrinsic competing uses for the bandwidth. Likewise, gram the phone or alter its operation in accordance with the this may be implemented in or as part of existing cellular protocol, unless granted this option by the network operator. network handsets, using common protocols. On the other 40 The user cannot reasonably compete or interfere with the hand, different spectrum and/or protocols may be employed, licensed spectrum. While older analog cellular phones pro which may be licensed or unlicensed. vided the user with an option to install power amplifiers and Various studies have shown that modeled multihop mobile vehicle mount antennas, few current users employ these ad hoc network architectures tend to have low efficiency in options. excess of three to five hops. This is due to node mobility and 45 If one limits the present system to a five hop distance from the probability of finding an end-to-end connection, mutual fixed cellular infrastructure (or more accurately, permits the interference and competition for bandwidth in shared channel system to deny service to nodes more than five hops away) protocols, and the overhead of maintaining useful routing then the routing requirements and node complexity may be tables. If we take five hops as a reasonable maximum, and substantially simplified. We also presume that each node has each transceiver has a 1000 meter range, then a 5 km maxi 50 geolocation capability, and therefore can provide both its mum range extension is possible. It is believed that by extend location and Velocity vector. This is reasonable, since the ing the fringe of cellular networks by 3-5 km, a significant FCC E911 mandate provides for geolocation of handsets portion of the unmet demand for cellular service will be within range of the cellular infrastructure, and GPS is a one satisfied. significant and readily available option to provide this feature, If we assume that a significant portion of the mobile nodes 55 independent of the cellular infrastructure. are power constrained (e.g., battery operated), that is, retrans The ad hoc communications can occur using a licensed or mission of packets imposes a power cost, then the stability of unlicensed band. For example, since we presume that nodes the mobile ad hoc network and cooperation with its require are beyond range of a fixed cellular tower (except the closest ments will depend on properly incentivizing intermediary node), the ad hoc network may reuse licensed bandwidth in nodes to allocate their resources to the network. Since this 60 the uncovered region. The ad hoc communications may also incentive is provided in a commercial context, that is, the occur in unlicensed spectrum, such as the 2.4 GHz ISM band, cellular service is a commercial enterprise with Substantial 5.8 GHz ISM band, DSRC band, or 900 MHz band. cash flow, a real economy with monetary incentives may be In order to provide optimum compensation, two issues are provided. Under Such circumstances, it is relatively straight confronted. First, the total compensation paid; and second, forward to allocate costs and benefits between the competing 65 the distribution of payments between the intermediaries. The interests to achieve consistent and apparent incentives. On the VCG auction is a known means for optimizing a payment other hand, the cost of this additional process must be com which must be distributed between a number of participants. US 8,874,477 B2 51 52 In this case, each potential intermediary places a “bid. A of the communication. The actual price may be, for example, multi-factorial optimization is performed to determine the the bid price less otherwise attributable discount under the lowest cost set which provides sufficient services. first mode based on the maximum number of hops, etc. The In a cellular system, each Subscriber typically purchases a intermediate nodes set forth their bids in the manner of a VCG number of minute units on a monthly recurring charge basis. auction, with each bid presumably exceeding the first mode Compensation might therefore be based on minutes or compensation. The VCG optimization may be corrected for money. Since there is a substantial disincentive to exceed the quality of service factors and anticipated network Stability. number of committed minutes, providing a surplus of minutes It is noted, as elsewhere herein, that the preferred bidding may not provide a significant incentive, because the user will and optimization is performed automatically as a part of the rarely exceed the committed amount. Monetary incentives, 10 protocol, and not under direct human control and Supervision. on the other hand, must be coupled to a higher monthly Therefore, the automated processes may be defined to pro recurring fee, since the proposal would by unprofitable oth mote stability and cooperation by both the device and its erwise. owner with the network. In other cases, human involvement A more direct Scheme provides an economy for multihop may be used, although thus will typically be quite inefficient networks somewhat independent from the cellular system 15 and impose transactional expenses (opportunity costs) in economy. That is, nodes that participate as intermediary, may excess of the underlying transaction value. The use therefore also participate as a principal to the information communica provides a set of explicit or implicit subjective criteria as a tion, while those who abstain from intermediary activities are basis for the agent to act accordingly. An intermediary denied access to the network extension as a principal. chooses its bid for providing packet forwarding services While, on a theoretical basis, optimization of both price based on a number of factors, such as anticipated power cost, and distribution would be considered useful, in a practical opportunity cost, etc. system, it may be useful to make simplifying presumptions Clearly, the economics of the system are substantially and simplifications. For example, while a VCG auction may under the control of the cellular carrier, who may offer provide an optimal cost and distribution of compensation, in “plans” and “services” for their customers, thus providing an a commercial network, a degree of certainty may actually 25 alternative to the usage-based bidding process, at least for prove advantageous. For example, a fixed compensation per some users. The VCG process, however, remains useful for hop or per milli Watt-second may prove both fair and reason compensating intermediaries. able. The intermediary chooses its bid for providing packet for Likewise, a degree of certainty over cost would be benefi warding services based on a number of factors, such as antici cial over an “optimal cost. On the other hand, fixed cost and 30 pated power cost, opportunity cost, etc. These bids are deter fixed compensation are inconsistent in a revenue neutral sys mined automatically, and therefore do not consume tem. Even if the cellular carrier subsidizes the extension substantial human efforts. operation, there is little rationale for making the usage at the Summary fringe insensitive to cost, other than the relief from uncer Game theory is a useful basis for analyzing ad hoc net tainty, which will tend to increase fringe usage, and the scope 35 works, and understanding the behavior of complex networks of the subsidy cost. of independent nodes. By presuming a degree of choice and As discussed above, there are methods drawn from finan decision-making by nodes, we obtain an analysis that is cial models which may also serve to improve certainty and robust with respect to such considerations. The principal reduce perceived risk. issues impeding deployment are the inherent complexity of Therefore, it is realistic for a node requesting extension 40 the system, as well as the overhead required to continuously service to apply a value function to define a maximum pay optimize the system. Determination of a set of simplifying ment for service. The payment is therefore dependent on presumptions to reduce protocol overhead and reduce com system cost, alleviating the requirement for Subsidy, but also plexity may improve performance. Hierarchal considerations dependent on need. can be imposed to alter the optimization of the system, which In the typical case, the load on the extension network will 45 would be expected to provide only a small perturbation to the be low, since if demand were high, the fixed infrastructure efficient and optimal operation of the system according to a would likely be extended to this region. On the other hand, pure VCG protocol. A marketplace auction with competition there may be cases where demand is high, and therefore there between potential buyers and potential sellers, and with the is some competition for access to the network, leading to a economic surplus distributed between parties which must need to arbitrate access. 50 defer to active participants, provides incentive to all affected In general, where economic demand is high, there is a parties, and therefore may provide a better result thana simple tendency to recruit new Sources of supply. That is, the system transfer between Supply and demand elements only. may operate in two modes. In a first, low demand mode, costs The ad hoc network does not exist in a vacuum. There are are based on a relatively simple algorithm, with a monetary various competing interests seeking to use the same band cap. In a second mode, costs are competitive (and in excess of 55 width, and technological Superiority alone does not assure the algorithmic level), with compensation also being com dominance and commercial Success. Game theory may also petitive. The Surplus, in this case, would generally be allo be used as a tool to analyze the entities which seek to deploy cated to the cellular carrier, since this represents the commer ad hoc networks, especially where they compete. cial profit of the enterprise. The first mode also serves another purpose; under lightly loaded conditions, the market may be 60 First Embodiment thin, and therefore pricing unstable. Therefore, the imposition of fixed pricing leads to reduced pricing risk. In a typical auction, each player is treated fairly; that is, the In the second mode, an intended user specifies his demand same rules apply to each player, and therefore a single as a maximum price and demand function, that is, a bid based economy describes the process. The fair auction therefore on, for example, a value of the communication. Generally, 65 poses challenges for an inherently hierarchal set of users, this would be set by the user in advance as a static value or Such as a military organization, where rank is accompanied relatively simple function representing the contextual value by privilege. The net result, however, is a decided apparent US 8,874,477 B2 53 54 disadvantage to lower ranking agents, at least when viewed in In accordance with the proposed artificial economy, each light of constricted self-interest. The issues that arise are node has a generator function for generating economic units, similar to the relating to “altruism', although not identical, which are then used in a transaction (e.g., an auction) with and thus the game theoretic analysis of altruistic behavior other nodes to create a market economy, that is, each node has may be imported for consideration, as appropriate. 5 a Supply and demand function, and acts as a source or sink for In a mobile ad hoc communications network, a real issue is a limited resource. In some cases, nodes may have only Sup user defection or non-compliance. For example, where a cost ply or demand functions, or a degree of asymmetry, but in this is imposed on a user for participating in the ad hoc network, case, these are typically Subject to an external economic con e.g., battery power consumption in a mesh radio network, if sideration, and the artificial economy will be less effective in 10 providing appropriate incentives. According to one imple the anticipated benefit does not exceed the cost, the user will mentation of this embodiment, the artificial economic units simply turn off or disable the device until actually needed. have a temporally and/or spatially declining value, so that The result of mass defection will, of course, be the instability wealth does not accumulate over long periods and/or cannot and failure of the ad hoc network itself, leading to decreased be transferred over large distances. The decline may be linear, utility, even for those who gain an unfair or undue advantage 15 exponential, or based on Some other function. This creates a under the system. Thus, perceived fairness and net benefit is set of microeconomies insulated from each other. Where dis required for network Success, assuming that defection and/or tant microeconomies must deal with each other, there is a non-compliance are possible. discount. This architecture provides a number of advantages, On the otherhand, in military systems, the assertion of rank for example, by decreasing the influence of more spatially as a basis for priority is not itself necessarily arbitrary or and temporally distant effects, the scope of an optimization capricious. Orders and communications from a central com analysis may be relatively constrained, while reducing the mand are critical for the organization itself, and thus the lower amount of information which must be stored overtime and/or ranking agents gain at least a peripheral, if not direct benefit carried over distance in order to permitan optimization. Like as their own chain of command employs their resources. wise, since the economy is artificial, the discount need not be Therefore, the difficulty in analyzing the application of a fair 25 recouped within the scope of the system; that is, conservation game paradigm to a hierarchal organization is principally a of capital is not required. In the same manner, a somewhat result of conceptualizing and aligning the individual incen different incentive structure may be provided; that is, eco tives with those of the organization as a whole and the rela nomic units generated at one location and at one time may tionship between branches. Thus, in contradistinction to typi have a higher value at a different location and time; this may cal self-organizing peer-to-peer networks, a hierarchal 30 encourage reduced immediate use of the system resources, network is not seen as self-organizing, at least in terms of the and relocation to higher valued locations. As discussed below, hierarchy, which is extrinsic to the formation of the commu one embodiment of the invention permits trading of credits, nications network under consideration. and thus, for example, a user may establisha repeater site at an As discussed below, the “distortions” of the network under-served location to gain credits for use elsewhere. Pref imposed by the external hierarchy can be analyzed and 35 erably, beyond a “near field’ effect, the value does not con accounted forby, for example, the concepts of inheritance and tinue to increase, since this may result in inflationary pres delegation. Thus, each branch of a hierarchy tree may be Sures, and undermine the utility of the system in optimally considered an object, which receives a set of characteristics balancing immediate Supply and demand at a particular loca from its root, and from which each sub-branch inherits the tion. characteristics and adds subcharacteristics of, for example, 40 As can be seen, through modifications of the governing specialization. It is noted that the hierarchy need not follow rules and formulae, the system can be incentivized to behave non-ambiguous or perfect rules, and thus there is no particular in certain ways, but care should be exercised since a too limit imposed that the hierarchy necessarily follow these for narrow analysis of the incentive might result in unintended malisms. Rather, by analyzing those aspects of the hierarchy long term or distant effects. To the extent that human behavior which comply with these formalisms in accordance there 45 and Subjective analysis is involved, care should also be exer with, efficiency is facilitated. cised in applying a rationality assumption, since this is not In establishing an economic system, a preliminary ques always true. Rather, there may be applicable models for tion is whether the system is microeconomic or macroeco human irrational behavior that are better suited to an under nomic; that is, whether the economy is linked to a real standing of the network behavior in response to a perturba economy or insulated from it. One disadvantage of a real 50 tion. economy with respect to a peer relationship is that external The typical peer-to-peer ad hoc network may be extended wealth can override internal dynamics, thus diminishing the to the hierarchal case by treating each branch (including advantages to be gained by optimization, and potentially cre Sub-branches) within the chain of command as an economic ating a perception of unfairness for externally less wealthy unit with respect to the generator function. At any level of the agents, at least unless and until the system accomplishes a 55 hierarchy, the commander optionally retains a portion of the wealth redistribution. An artificial economy provides a solu wealth generation capacity, and delegates the remainder to its tion for a peer network in which each node has an equal subordinates. Therefore, the rank and hierarchal consider opportunity to gain control over the adhoc network, indepen ations are translated to an economic wealth (or wealth gen dent of outside influences and constraints. On the other hand, eration) distribution. One aspect of this system allows wealth by insulating the network from external wealth redistribution, 60 transfer or redistribution, although in a real system, a time real efficiency gains may be unavailable. Therefore, both delay is imposed, and in the event of a temporally and/or types of economies, as well as hybrids, are available. Thus, as spatially declining value, the transfer will impose a cost. discussed in more detail below, a “fair initial (or recurring) Thus, an initial misallocation is undesired, and there will be wealth distribution may be applied, which may be supple an incentive to optimally distribute the wealth initially. Of mented with, and/or provide an output of external wealth. 65 course, if centralized control with low penalty is desired, it is The rules or proportion of external influence may be prede possible to limit the penalty, if any, for wealth redistribution termined, adaptive, or otherwise. through appropriate rules, although the time for propagation US 8,874,477 B2 55 56 through the network remains an issue, and blind nodes (i.e., bid. When the bids are “competitive', there is a substantial those which do not have an efficient communication path, or risk that the bid will be a winning bid, and thus incur the full have insufficient resources to utilize otherwise available paths bid cost. Thus, there is a disincentive to bidding a high value, through the hierarchy) may also lead to limitations on system but without an intent to win. Of course, the bid deposit may be performance. a flat fee, or Subject to a mathematical or adaptive function, In this system, there may be an economic competitive rather than directly related to administrative cost. distortion, under which a node's subjective value of a The aggregated bid deposits may, for example, be awarded resource is influenced by its then subjective wealth. If a node to a class who are optimally incentivized by the nature of this is supplied with wealth beyond its needs, the wealth is wasted, payment. For example, it may be awarded to those selling since it may decline in value and cannot be hoarded indefi 10 bandwidth, in a manner generally inversely proportional to nitely. (In a network wealth model in which wealth could be the value of their ask, or, for example, based on allocations hoarded indefinitely, small deviations from optimality and during the combinatorial (VCG) auction. This payment arbitrage opportunities may be exploited to create a percep would then incentivize sellers to offer services at a low price, tion of unfairness, thus, this is not preferred.) If a node is improving network availability. Supplied with insufficient wealth, economic Surplus through 15 Of course, there may be other classes within the auction transactional gains are lost. Thus, each node must analyze its population who may be taxed or subsidized, using value expected circumstances to retain or delegate the generator derived from the auction process. function, and to optimally allocate wealth between compet In a strategyless auction, automated bidding is quite fea ing Subordinates. Likewise, there may be a plurality of quasi sible, since the optimal bid is the computed value. For auc optimal states. tions in which a bidder does not have an incentive to bid its In any economic transaction, there is an amount that a true private value, and thus must assume a strategic play, seller requires to part with the resource, a price a buyer is automated bidding becomes more of a challenge, but may willing to pay, and a surplus between them. Typically, in a two also be automated. party transaction, the Surplus is allocated to the party initiat In a strategy-less auction, a bidder cannot gain by bidding ing the transaction, that is, the party initiating the transaction 25 over or under its private value. If a bidder bids below its uses some discovery mechanism to find the minimum price private value, it has a reduced chance of gaining the benefit of acceptable by the buyer. In brokered or agent-mediated trans the transaction. actions, a portion of the Surplus is allocated to a facilitator. In an auction which is subject to strategic bidding, the In accordance with one aspect of the present invention, strategy may be mitigated by imposing commensurate risks compliance with the community rules, as well as an incentive 30 and costs to balance the perceived advantage toward Zero. to bid oraska true private value is encouraged by distributing In particular, the competitive bidders seeking to allocate a a portion of the transaction surplus to losing competitive scarce resource for themselves receive compensation for bidders. While according to one proposal, this portion is deferring to the winning bidder in an amount commensurate allocated in accordance with their reported valuations, this with their reported value. Thus, sellers receive their minimum creates a potential incentive forbidders who know they will 35 acceptable value, buyers pay their maximum valuation, the not be winning bidders to overbid, and thereby gain an Surplus is distributed to the community in a manner tending to increased portion of the surplus. In order to reward honest promote the highest bids within the private value of the bid reporting of private values, the reward function must penalize der. In a corresponding manner, the auction rules can be both overreporting and underreporting of private values. This established to incentivized sellers to ask the minimum pos circumstance occurs if, at each bid, there is a risk of winning 40 sible amount, above their reserve. For example, a portion of commensurate with the bid, and thus the system is strat the surplus may be allocated to bidders in accordance with egyproof. In order to achieve this circumstance, for example, how close they come to the winning ask. Therefore, both a statistical noise or probability distribution may be added to incentives may be applied, for example with the Surplus split the system, with an acceptance of a bid made a statistical in two, and half allocated to the bidderpool and half allocated process. This results in a “fuzzy” boundary on the bid value, 45 to the seller pool. Clearly, other allocations or proportion although it may impose an inefficiency on the market since ations are possible. any deviation from the optimal market price represents a loss. The winning bidder and/or seller may be included within Another approach to minimizing strategic bidding is to the rebate pool. This is particularly advantageous where for impose a bid fee. That is, each bidder must offera prepayment various reasons, the winning bidder is not selected. Thus, this corresponding to a small portion of its bid, thereby disincen 50 process potentially decouples the bidding (auction) process tivizing bidding to lose. The winning bidder will then pay a and the resulting commercial transaction. second price plus the deposit bid. The sellers will receive their Because of transactional inefficiencies, human behavioral own lowest cost (or second cost) bid. Losing bidders will aspects, and a desire to avoid increased network overhead by receive a payment in accordance with the value of their bid, “false' bidders seeking a share of the allocation pool without less the bid deposit. In order to disincentivize strategic bid 55 intending to win the auction, it may be useful to limit the ding, the average return to a bidder is less than the bid cost. In allocation of the surplus pool to a subset of the bidders and/or fact, a good target for the bidder deposit is the administrative sellers, for example the top three of one or both. This therefore cost of transacting the bidding negotiations. This, in turn, encourages bidders and/or sellers to seek to be in the limited provides an incentive to keep the administrative overhead group splitting the pool, and thus incentivizes higher bids and low, thus improving overall system performance, especially 60 lower asks. Of course, a party will have a much stronger where the administrative communications compete with nor incentive to avoid bidding outside its valuation bounds, so the mal communications for bandwidth. In this circumstance, risk of this type of inefficiency is small. those bidding to win receive either the benefit of the transac As discussed above, one embodiment of the invention pro tion or a payment for deference, less the transactional fee. vides a possible redistribution or wealth among nodes within Those who are bidding strategically, in manner seeking to 65 a hierarchal chain. This redistribution may be of accumulated acquire the deference payment, must risk the transactional wealth, or of the generation function portion. Trading among cost, and to gain Substantially, must Submit a relatively high hierarchally related parties is preferred, since the perceived US 8,874,477 B2 57 58 cost is low, and the wealth can be repeatedly redistributed. In not “cost sensitive' (however the appropriate cost function fact, it is because of the possibility of wealth oscillation and might be expressed), it fails to respond to “market' forces. teaming that the declining wealth function is preferred, since Accordingly, a peer to peer mobile adhoc network Suitable this will tend to defeat closely related party control over the for respecting hierarchal organization structures is provided. network for extended periods. In this hierarchal system, the hierarchy is represented by an It is noted that, in a multihop mobile ad hoc network, ifa initial wealth or wealth generation function distribution, and communication path fail, no further transfers are possible, the hierarchally higher nodes can reallocate wealth of nodes potentially resulting in stalled or corrupt system configura beneath themselves, exercising their higher authority. This tion. It is possible to transfer an expiring or declining portion wealth redistribution can be overt or covert, and if overt, the of the generating function; however, this might lead a node 10 hierarchal orders can be imposed without nodal assent. In a which is out of range to have no ability to rejoin the network covert redistribution, trust may be required to assure redistri upon return, and thus act as an impediment to efficient net bution by a node to a grandchild node. work operation. Therefore, it is preferred that, in an artificial The wealth and its distribution can be implemented using economy, each node has some intrinsic wealth generator modified micropayment techniques and other verifiable cryp function, so an extended period of inactivity, a node gains 15 tographic techniques. This wealth can be applied to auctions wealth likely sufficient to rejoin the network as a full partici and markets, to allocate resources. Various aspects of this pant. system are discussed in more detail elsewhere in this speci In practice, in a typical military-type hierarchy, the bulk of fication. the wealth generating function will be distributed to the low In accordance with aspects of this embodiment, an est ranks with the highest numbers. Thus, under normal cir example is provided. In this scenario, a vehicle traveling cumstances, the network will appear to operate according to along a highway seeks traffic information 10-20 miles ahead a non-hierarchal (i.e., peer) model, with the distortion that not on the road. The transceiver in the vehicle has a range of about all nodes have a common generator function. On the other 0.5 miles, meaning that, assuming maximum hop range, hand, hierarchically Superior nodes either retain, or more 20-40 hope would be necessary in each direction in order to likely, can quickly recruit Surrounding Subordinates to allo 25 fulfill a response to a request for information. If we further cate their wealth generating function and accumulated wealth assume that the traffic density allows an average density of to pass urgent or valuable messages. Thus, if 85% of the compatible transceivers of 1 per 0.05 miles, then it would wealth and network resources are distributed to the lowest appear that for each hop, a number of intermediaries would be ranking members, then the maximum distortion due to hier possible. We further assume that each vehicle has a pair of archal modifications is 15%. 30 antennas (which may operate on different frequencies), for One way that this allocation of wealth may be apparent is ward and backward looking, so that forward and backward with respect to the use of expensive assets. Thus, a high level communications are non-interfering. It is noted that, in opera node might have access to a high power broadcast system or tion, it is not a single vehicle that seeks information respond licensed spectrum, while low level nodes might ordinarily be ing to a request; rather, it is likely that 2-25% of vehicles will limited to lower power transmission and/or unlicenced spec 35 seek information within a short period, especially of the cost trum or cellular wireless communications. For a low level of fulfilling a request is relatively low. We also assume that node to generate a broadcast using an expensive asset (or to there is no common trigger event. Such as an accident, which allocate a massive amount of space bandwidth product), it would provoke essentially all vehicles to request the same must pass the request up through the chain of command, until information, a circumstances that could be addressed through sufficient wealth (i.e., authority) is available to implement the 40 a multicast or broadcast. broadcast. If the vehicle sought to arrange a communication over the In fact, such communications and authorizations are quite entire 10-20 miles in advance of communicating, this would consistent with the expectations within a hierarchal organi require a multifactorial optimization likely involving over Zation, and this construct is likely to be accepted within a 100 transceivers, and if even one of the 20-40 intermediates military-type hierarchal organization. 45 fails, the entire communication fails. The administrative over Under normal circumstances, a Superior would have an head for this process may not outweigh its advantages. incentive to assure that each Subordinate node possesses Suf On the other hand, if we instead presume that the vehicle ficient wealth to carry out its function and be incentivized to only optimize a path over a limited range or number of hops, participate in the network. If a subordinate has insufficient e.g., 1 mile or 5 hops, then the optimization is facilitated and initial wealth (or wealth generating function) allocation, it 50 the administrative overhead reduced. On the other hand, this may still participate, but it must expend its internal resources requires that vehicles or nodes at the fringe arrange for to obtain wealth for participation toward its own benefit. This, completion of the communication. It is here that the statistical in turn, leads to a potential exhaustion of resources, and the aspects of the network architecture may be exploited to unavailability of the node for ad hoc intermediary use, even achieve efficiencies. Thus, in observing or participating in the for the benefit of the hierarchy. An initial surplus allocation 55 network activities over a period of time, a node can model the will lead to overbidding for resources, and thus inefficient behavior of nearby nodes, and determine a degree of risk with resource allocation, potential waste of allocation, and a dis respect to the model. That node may then undertake the risk incentive to act as an intermediary in the ad hoc network. associated with its assessment of its environment, and com While in a traditional military hierarchy, cooperation can be municate an offer to act as agent for completion of the com mandated, in Systems where cooperation is perceived as con 60 munication, without explicitly communicating the details of trary to the net personal interests of the actor, network stabil the communication. Therefore, the originating node opti ity may be poor, and defection in spite of mandate. mizes a local region ad hoc network, and then adopts an In a military system, it is thus possible to formulate an estimate of the network state beyond the edge of the local “engineered solution which forces participation and elimi region. nates defection; however, it is clear that such solutions forfeit 65 Economically, the vehicle seeking the information broad the potential gains of optimality, and incentivizes circumven casts a bid or value function of its valuation of the resources tion and non-compliance. Further, because Such a system is it requires. This bid is propagated to the local region or US 8,874,477 B2 59 60 beyond, and compared with the bids or value functions of carrier gains a benefit from the ad hoc network. One such other vehicles or nodes. A winning vehicle or node then possible benefit is extension of its fixed infrastructure to serve assumes control over the minimum temporal-spatial-fre under-covered areas. Another possible benefit is the ability to quency channel required. As stated above, at the edge of the provide information from the ad hoc network to more remote local region, nodes may act as proxies or agents, and under areas. A further benefit is the ability to use unlicensed spec take the risk of the more distant communication, adding a risk trum for its activities in a standard and non-interfering man premium to their ask. The node with the lowest ask is selected . as the agent or proxy. It is noted that the role of communica In the virtual economy, each node has a physically and/or tion intermediary and proxy or agent is discrete, and therefore logically secure cryptographic module, which sequentially need not be a single element, though certain efficiencies are 10 generates values which have a unique index number, and may gained if this is the case. The agent or proxy must also conduct be verified as to node and time of origin, and possibly chain of an auction for the next stage of the communication, in a owners. A node receiving this value can therefore verify that process which is repeated until the destination node is it is authentic, its time of creation (and therefore amortization included within the local region. schedule), and as an audit trail, the chain of ownership. Each The proxy or agent undertakes the risk of the cost of the 15 bid is also cryptographically secure and signed, so that if a downstream communications, as well as the risk of non node places a bid, and later fails to pay, a later investigation payment, and thus may well charge a substantial premium can be conducted to correctly account for the transaction, and over its actual risk-free cost. Therefore, the efficiency gained possibly penalize wrongdoing. The payments for a commu through the use of the agent or proxy derives from the admin nication are communicated after the transaction, in a crypto istrative efficiencies gained, as well as comprehension that graphic wrapper (cryptolope) destined for a target node. the risks are inherent, and must generally be undertaken by Since these are secure, the opportunity for theft is low, and some element of the network. The incrementally added risks there is little incentive for intentional delay of transmission by may be small, but are transferred. A node which promotes any intermediate. Further, these packets may be transmitted itself for acting as agent or proxy may do so because it has along redundant paths, to limit the ability of any one node to lower risks, costs or otherwise unproductive assets. For 25 disrupt communications. example, a cellular telephone carrier may choose to partici The ability of a node to spend the same value packet twice pate in the network, using its fixed infrastructure as a backup, is limited by a number of factors. First, since each node has a or bypass. In that case, if the network fails, or is less efficient, defined generator function, if its spending exceeds its genera it has the option of using its own facilities. tion capacity, this will be apparent to nearby nodes. Second, The agent or proxy therefore arbitrages the risk, based on 30 since each packet has an index value, the other nodes may its own knowledge of its local region which is different from compare these values to make Sure that they are not used more the local region of the originator of the communication. There than once by any node, before they are transferred to another may be less competition for the role of arbitrageur, allowing node. Since the value of the credit declines in value overtime, it to claim a larger portion of the economic Surplus. In fact, an indefinite period monitoring is not required. arbitrageur may pre-acquire resources at a defined price, and 35 In some instances, saving value may be an efficient strat resell them later at a profit. Thus, it is seen that economic egy. In order to take advantage of these gains, special bank efficiencies and increased profits for intermediaries are not nodes may be established which have the ability to hoard inconsistent, where opportunities for reduction in inefficien credits and then reissue new credits when required. Typically, cies exist. there will be no interest, and in fact there may be discount and Adding hierarchal element to this example, it is noted that 40 delay. The net result of promoting savings will typically be a certain risks are reduced when transactions are conducted reduction in demand with respect to Supply, thus increasing between related entities. For example, if their respective availability of resources. By allowing withdrawal of savings, wealth is interlinked, over the long term, the risk of non periods of inflation and high peak demand is possible. Fur payment is abated. Likewise, the risk of defection or non ther, if the withdrawn wealth has the same amortization compliance is reduced. Further, since it is presumed that the 45 schedule as newly generated credits, an event which provokes benefit function of related nodes is intertwined, actual costs a “run on the bank” may result in a rapid diminution of saved may be reduced, since the communication itself is a counter wealth, unless the immediate recipients bank the newly trans Vailing benefit to the cost of a related node conveying the ferred wealth. message or packet. Thus, there will likely be a preference for As is seen, many of the economic institutions of the real communications between more closely related nodes than 50 economy have equivalents in the virtual economy, and there between more distantly related or unrelated nodes. On the fore may be employed in their traditional and known roles to other hand, since wealth (virtual or real) itself is desirable, improve efficiency where the self-organizing features of the and inter-party transactions limit wealth gain opportunities, network alone incur corresponding inefficiencies, thus creat there will also be an incentive to conduct transactions with ing opportunities. Where necessary, links to a real economy, unrelated nodes for full value. As discussed above, in a hier 55 in order to pay for capital investment, efforts, or compensate archy, a top level node is initially allocated the entire wealth for risks, may be employed, however it is preferred that these and/or wealth generation function for its subordinates, which links be attenuated in order to isolate the bulk of the ad hoc is then redistributed as appropriate or desired. The top level network from the influence of real-economy node wealth, and node will generally not maintain more wealth than required, therefore to promote defection of those nodes who are disad since this is inefficient, and redistributions incur their own 60 Vantaged thereby. inefficiencies. The economy is preferably virtual, employing arbitrary Second Embodiment value credits generated using a cryptographic function. One possible exception is where external elements, such as cellu Multihop Ad Hoc Networks require cooperation of nodes lar telephone carriers, are integrated into the system. Since 65 which are relatively disinterested in the content being con these are real economy agents, there must be some inter veyed. Typically, Such disinterested intermediaries incur a change in value between credits and cash, unless the cellular cost for participation, for example, power consumption or US 8,874,477 B2 61 62 opportunity cost. Economic incentives may be used to pro This patent builds upon and extends aspects of U.S. Pat. mote cooperation of disinterested intermediaries. An eco No. 6.252,544 (Hoffberg), Jun. 26, 2001, and U.S. Pat. No. nomic optimization may be achieved using a market price 6,429,812, Aug. 6, 2002, which are expressly incorporated finding process, such as an auction. In many scenarios, the herein by reference in its entirety. See, also, U.S. Pat. No. desire for the fairness of an auction is tempered by other 6,397,141 (Binnig, May 28, 2002, Method and device for concerns, i.e., there are constraints on the optimization which signaling local traffic delays), expressly incorporated herein influence price and parties of a transaction. For example, in by reference, which relates to a method and an apparatus for military communication systems, rank may be deemed an signaling local traffic disturbances wherein a decentralized important factor in access to, and control over, the commu communication between vehicles, which is performed by 10 exchanging their respective vehicle data. Through repeated nications medium. A simple process of rank-based preemp evaluation of these individual vehicle data, each reference tion, without regard for Subjective or objective importance, vehicle may determine a group of vehicles having relevance will result in an inefficient economic distortion. In order to for itself from within a maximum group of vehicles and normalize the application of rank, one is presented with two compare the group behavior of the relevant group with its own options: imposing a normalization scheme with respect to 15 behavior. The results of this comparison are indicated in the rank to create a unified economy, or considering rank using a reference vehicle, whereby a homogeneous flow of traffic set of rules outside of the economy. One way to normalize may be generated, and the occurrence of accidents is reduced. rank, and the implicit hierarchy underlying the rank, is by One aspect of the invention provides a communications treating the economy as an object-oriented hierarchy, in system, method and infrastructure. According to one pre which each individual inherits or is allocated a subset of the ferred embodiment, an ad hoc, self organizing, cellular radio rights of a parent, with peers within the hierarchy operating in system (sometimes known as a "mesh network) is provided. a purely economic manner. The extrinsic consideration of Advantageously, high gain antennas are employed, prefer rank, outside of an economy, can be denominated “respect’. ably electronically steerable or phased array antennas, to which corresponds to the Societal treatment of the issue, provide efficient communications and to increase spatial rather than normalizing this factor within the economy, in 25 communications bandwidth, both between nodes and for the order to avoid unintended secondary economic distortion. system comprising a plurality of nodes communicating with Each system has its merits and limitations. each other. See, U.S. Pat. No. 6,507,739 (Gross, et al., Jan. 14, An economic optimization is one involving a transaction in 2003), expressly incorporated herein by reference. which all benefits and detriments can be expressed in normal In general, time-critical, e.g., voice communications ized terms, and therefore by balancing all factors, including 30 require tight routing to control communications latency. On Supply and demand, at a price, an optimum is achieved. Auc the other hand, non-time critical communications generally tions are well known means to achieve an economic optimi are afforded more leeway in terms of communications path Zation between distinct interests, to transfer a good or right in ways, including a number of “hops’, retransmission latency, exchange for a market price. While there are different types of and out-of-order packet communication tolerance, between auctions, each having their limitations and attributes, as a 35 the source and destination or fixed infrastructure, and quality class these are well accepted as a means for transfer of goods of communication pathway. Further, it is possible to establish or rights at an optimum price. Where multiple goods or rights redundant pathways, especially where communications are required in a Sufficient combination to achieve a require bandwidth is available, multiple paths possible, and no single ment, a so-called Vickrey–Clarke-Groves (VCG) auction may available path meets the entire communications requirements be employed. In Such an auction, each Supplier asserts a 40 or preferences. desired price for his component. The various combinations Technologies for determining a position of a mobile device which meet the requirement are then compared, and the low are also well known. Most popular are radio triangulation est cost combination selected. In a combinatorial Supply auc techniques, including artificial satellite and terrestrial trans tion, a plurality of buyers each seek a divisible commodity, mitters or receivers, dead reckoning and inertial techniques. and each bids its best price. The bidders with the combination 45 Advantageously, a satellite-based or augmented satellite sys of prices which is maximum is selected. In a commodity tem, although other Suitable geolocation systems are appli market, there are a plurality of buyers and sellers, so the cable. auction is more complex. In a market economy, the redistri Navigation systems are also well known. These systems bution of goods or services are typically transferred between generally combine a position sensing technology with a geo those who value them least to those who value them most. The 50 graphic information system (GIS), e.g., a mapping database, transaction price depends on the balance between Supply and to assist navigation functions. Systems which integrate GPS, demand; with the Surplus being allocated to the limiting fac GLONASS. LORAN or other positioning systems into tOr. vehicular guidance systems are well known, and indeed navi gational purposes were prime motivators for the creation of Third Embodiment 55 these systems. Environmental sensors are well known. For example, sens A third embodiment of the invention, described below, ing technologies for temperature, weather, object proximity, represents a system which may employ a self-organizing location and identification, vehicular traffic and the like are network to convey information between mobile nodes. It is well developed. In particular, known systems for analyzing expressly understood that the concepts set forth above in the 60 vehicular traffic patterns include both stationary and mobile first and second embodiments are directly applicable, and sensors, and networks thereof. Most often, such networks each aspect of the third embodiment may be extended using provide a stationary or centralized system for analyzing traf the hierarchal principles and modifications, in a consistent fic information, which is then broadcast to vehicles. manner, to achieve the advantages described herein. That is, Encryption technologies are well known and highly devel while the third embodiment generally describes peer nodes, 65 oped. These are generally classified as being symmetric key, the extension of the systems and methods to non-peer nodes for example the Data Encryption Standard (DES), and the is specifically envisioned and encompassed. more recent Advanced Encryption Standard (AES), in which US 8,874,477 B2 63 64 the same key is used for encryption as decryption, and asym an event may be a reliable indicator of the actual existence of metric key cryptography, in which different and complemen the event, the failure to warn of an event does not necessarily tary keys are used to encrypt and decrypt, in which the former mean that the event does not exist. Further, as events age, their and the latter are not derivable from each other (or one from reliability often decreases. the other) and therefore can be used for authentication and The present invention therefore seeks to provide additional digital signatures. The use of asymmetric keys allows a so information which may be of use in decision-making, includ called public key infrastructure, in which one of the keys is ing a reliability of the information presented, and/or risk published, to allow communications to be directed to a pos associated with that information, if true. These types of infor sessor of a complementary key, and/or the identity of the mation are typically distinct from the objects themselves. In sender of a message to be verified. Typical asymmetric 10 order to present these additional degrees of freedom of infor encryption systems include the Rivest-Shamir-Adelman mation within the confines of efficient human cognition, a algorithm (RSA), the Diffie-Hellman algorithm (DH), elliptic new paradigm is provided. Essentially, the objects presented curve encryption algorithms, and the so-called Pretty Good (which may be, for example, identifiers of events), are Privacy (PGP) algorithm. mapped or ranked by a joint function of risk and reliability. One embodiment of the invention provides a system that 15 Typically, the joint function will adopt economic theory to analyzes both a risk and an associated reliability. Another provide a normalized cost function. Of course, the risk and embodiment of the invention communicates the risk and asso reliability need not be jointly considered, and these may ciated reliability in a manner for efficient human comprehen remain independent considerations for mapping purposes. Sion, especially in a distracting environment. See, U.S. Pat. Because of human subjective perception of risk and reliabil Nos. 6,201493; 5,977,884; 6,118,403; 5,982,325; 5,485,161; ity, it may be useful to tune the economic normalized cost WO0077539, each of which is expressly incorporated herein function for Subjective considerations, although in other by reference, and the Uniden GPSRD (see Uniden GPSRD instances, an objective evaluation is appropriate and efficient. User's Manual, expressly incorporated herein by reference). In analyzing a complex data set for both time and space See also U.S. Pat. Nos. 5,650,770; 5,450,329; 5,504,482: patterns, wavelets may be useful. While the discrete wavelet 5,504,491; 5,539,645; 5,929,753; 5,983,161; 6,084,510; 25 transform (DWT), an analogy of the discrete Fourier trans 6.255,942: 6,225,901; 5,959,529; 5,752,976; 5,748,103; form (DFT) may be employed, it is perhaps more general to 5,720,770; 6,005,517; 5,805,055; 6,147,598; 5,687.215; apply arbitrary wavelet functions to the data set, and adopting 5,838,237; 6,044,257; 6,144,336; 6,285,867; 6,340,928; mathematical efficiencies as these present themselves, rather 6.356,822; 6,353,679 each of which is expressly incorporated than mandating that an efficient and predefined transform herein by reference. 30 necessarily be employed. According to this embodiment, a vehicle equipped with See references listed in Transforms Appendix, each of sensors, or acting as a probe, acquires information from its which is expressly incorporated herein by reference. Surroundings, and communicates them to other vehicles, One embodiment of the present invention thus advances either directly or through a number of hops. By advanta the art by explicitly communicating reliability or risk infor geously employing advanced radio and communication tech 35 mation to the user. Therefore, in addition to communicating nologies, efficient and reliable transfer video data over unli an event or predicted event, the system also computes or censed spectrum may be Supported, even multiple real-time determines a reliability of the information and outputs this streams over potentially conflicting channels. information. The reliability referred to herein generally is Statistical Analysis unavailable to the original detection device, though Such It is understood that the below analysis and analytical tools, 40 device may generate its own reliability information for a as well as those known in the art, may be used individually, in sensor reading. Sub-combination, or in appropriate combination, to achieve Therefore, one user interface embodiment according to this the goals of the invention. These techniques may be imple embodiment is improved by outputting information relating mented in dedicated or reprogrammable/general purpose to both the event and a reliability or risk with respect to that hardware, and may be employed for low level processing of 45 information. signals, such as in digital signal processors, within an oper According to a preferred embodiment of the invention, a ating system or dynamic linked libraries, or within applica vehicle travel information system is provided, for example tion software. Likewise, these techniques may be applicable, integrated with a vehicular navigation system. In a symmetric for example, to low level data processing, system-level data peer-to-peer model, each vehicle includes both environmen processing, or user interface data processing. 50 tal event sensors and a user interface, but the present invention A risk and reliability communication system may be use is not dependent on both aspects being present in a device. As ful, for example, to allow a user to evaluate a set of events in the vehicle travels, and as time advances, its context sphere is statistical context. Most indicators present data by means of a altered. For any context sphere, certain events or sensed con logical indicator or magnitude, as a single value. Scientific ditions will be most relevant. These most relevant events or displays may provide a two-dimensional display of a distri 55 sensed, to the extent known by the system, are then output bution, but these typically require significant user focus to through a user interface. However, often, the nature or exist comprehend, especially where a multimodal distribution is ence of relevant or potentially relevant event is unreliable, or represented. Typically, the human visual input can best reliance thereon entail risk. accommodate a three dimensional color input representing a In the case of a vehicle traveling along a roadway, there are set of bounded objects which change partially over time, and 60 two particular risks to analyze: first, that the recorded event it is ergonomically difficult to present more degrees of free may not exist (false positive), and second, that an absence of dom of information simultaneously. That is, the spatial image indication of an event is in error (false negative). For example, is not arbitrary, but represents bounded objects (or possibly the degree of risk may be indicated by an indication of color fuZZy edges), and the sequence over time should provide (e.g., red, yellow green) or magnitude (e.g., a bar graph or transitions. User displays of a magnitude or binary value 65 dial). typically do not provide any information about a likelihood of In many cases, the degree of risk is calculable, and thus error. Thus, while a recent positive warning of the existence of may be readily available. For example, if the event sensor is a US 8,874,477 B2 65 66 detection of police radar, reliability may be inferred from a high. For this same time-period, the probability that the traffic time since last recording of an event. If a car is traveling along enforcement officer has moved up the road against the direc a highway, and receives a warning of traffic enforcement tion of traffic flow is low. A car following 3 miles behind a radar from a car one mile ahead, there is a high degree of reliable sensor at 60 mph would therefore have a highly certainty that the traffic enforcement radar will actually exist 5 reliable indication of prospective conditions. As the time as the vehicle proceeds along the highway. Further, if the increases, so does the risk; a car following ten miles behind a traffic radar is in fixed location, there is a high degree of sensor would only have a general warning of hazards, and a certainty that there is no traffic enforcement radar closer than general indication of the lack thereof. However, over time, a one mile. On the other hand, if a warning of traffic radar at a general (and possibly diurnal or other cyclic time-sensitive given location is two hours old, then the risk of reliance on this 10 information is high, and the warning should be deemed gen variation) risk of travel within a region may be established, to eral and advisory of the nature of risks in the region. Prefer provide a baseline. ably, as such a warning ages, the temporal proximity of the It is noted that the risks are not limited to traffic enforce warning is spread from its original focus. ment radar or laser. Rather, the scheme according to the On the contrary, if the warning relates to a pothole in a 15 present invention is generalized to all sorts of risks. For certain lane on the highway, the temporal range of risk is example, a sensor may detect or predict Sun glare. In this case, much broader: even a week later, the reliability of the contin a model would be quite accurate for determining changes ued existence at that location remains high. However, over the over time, and assuming a reliable model is employed, this course of a year, the reliability wanes. On the other hand, condition could generally be accurately predicted. while there may be a risk of other potholes nearby, the par Another example is road flooding. This may be detected, ticular detected pothole would not normally move. for example, through the use of optical sensors, tire drag The algorithm may also be more complex. For example, if sensors, “splash' sensors, or other known sensors. In this a traffic accident occurs at a particular location, there are case, the relevant time-constant for onset and decay will be generally acceptable predictions of the effect of the accident variable, although for a given location, the dynamics may be on road traffic for many hours thereafter. These include rub 25 modeled with some accuracy, based on sensed actual condi bernecking, migrations of the traffic pattern, and secondary tions, regional rainfall, ground Saturation, and particular accidents. These considerations may be programmed, and the storm pattern. Therefore, a puddle or hydroplaning risk may set of events and datapoints used to predict spatial and tem be communicated to the driver in terms of location, likely poral effects, as well as the reliability of the existence of such magnitude, and confidence. effects. This, in turn, may be used to advise a traveler to take 30 a certain route to a destination. It is noted that these three independent parameters need not Eventually, the reliability of the information is inferred to all be conveyed to the user. For example, the geographic be so low as to cause an expiration of the event, although proximity to an event location may be used to trigger an preferably a statistical database is maintained to indicate geo output. Therefore, no independent output of location may be graphic regional issues broadly. 35 necessary in this case. In some cases, the magnitude of the Therefore, the system and method according to the present threat is relevant, in other cases it is not. In many present invention provides an output that can be considered “two systems (e.g., radar detection), threat magnitude is used as a dimensional’ (or higher dimensional); the nature of the warn surrogate for risk. However, it is well understood that there ing, and the reliability of the warning. In conjunction, the are high magnitude artifacts, and low magnitude true threats, system may therefore output a reliability of an absence of 40 and thus this paradigm has limited basis for use. The use of warning. In order to conserve communications bandwidth, it risk or confidence as an independent factor may be express or is preferred that an absence of warning is inferred from the intermediate. Thus, a confidence threshold may be internally existence of a communications channel with a counterpart, applied before communicating an event to the user. In deter along with a failure of a detection of an event triggering a mining or predicting risk or confidence, it may be preferred to warning. Alternately, Such communications may be explicit. 45 provide a central database. Therefore, generally more com The present invention can provide a mobile warning sys plex models may be employed, Supported by a richer data set tem having a user interface for conveying an event warning derived from many measurements over an extended period of and an associated reliability or risk of reliance on the warning. time. The central database may either directly perform the Preferably, the reliability or risk of reliance is assessed necessary computations, or convey an appropriate model, based on a time between original sensing and proximity. The 50 preferably limited to the context (e.g., geography, time, gen reliability may also be based on the nature of the event or eral environmental conditions), for local calculation of risk. sensed condition. An intrinsic reliability of the original The incorporated references relate, for example, to meth sensed event or condition may also be relayed, as distinct ods and apparatus which may be used as part of, or in con from the reliability or risk of reliance assuming the event or junction with the present invention. Therefore, it is under condition to have been accurately sensed. 55 stood that the present invention may integrate other systems, In order to determine risk, often statistical and probabilistic or be integrated in other systems, having complementary, techniques may be used. Alternately, non-linear techniques, synergistic or related in Some way. For example, common Such as neural networks, may be employed. In employing a sensors, antennas, processors, memory, communications probabilistic scheme, a sensor reading at time Zero, and the hardware, Subsystems and the like may provide a basis for associated intrinsic probability of error are stored. A model is 60 combination, even if the functions are separate. associated with the sensor reading to determine a decay pat The techniques according to the present invention may be tern. Thus, in the case of traffic enforcement radar, the half applied to other circumstances. Therefore, it is understood life for a “radar trap' for K band radar being fixed in one that the present invention has, as an object to provide a user location is, for example, about 5 minutes. Thereafter, the interface harnessing the power of statistical methods. There enforcement officer may give a ticket, and proceed up the 65 fore, it is seen that, as an aspect of the present invention, a user road. Thus, for times less than three minutes, the probability interface, a method of providing a user interface, computer of the traffic enforcement radar remaining in fixed position is Software for generating a human-computer interface, and a US 8,874,477 B2 67 68 system providing Such a user interface, presents a prediction service for a communications link may be assured, a remote of a state as well as an indication of a statistical reliability of database system serving one or more devices may be pro the prediction. vided. For example, a cellular data communications system Within a vehicular environment, the statistical analysis may be used to communicate with the Internet or a service according to the present invention may also be used to provider. improve performance and the user interface of other systems. The mobile unit, in operation, determines its position, and, In particular, modern vehicles have a number of indicators though explicit user input and/or inferential analysis, deter and warnings. In most known systems, warnings are provided mines the itinerary or expected path of the device and time at pre-established thresholds. According to the present inven sequence. The device (or associated systems) then determines tion, a risk analysis may be performed on sensor and other 10 the available weather information for the route and antici data to provide further information for the user, e.g., an indi pated itinerary (which may itselfbe dependent on the weather cation of the reliability of the sensor data, or the reliability information and/or reaction thereto). This available informa under the circumstances of the sensor data as basis for deci tion is then modeled, for example using a statistical model as Sion. (For example, a temperature sensor alone does not indi described hereinabove, to predict the forthcoming weather cate whether an engine is operating normally.) 15 conditions for the device or transporting vehicle. The device then determines the anticipated conditions and Fourth Embodiment relevance sorts them. In this case, both positive and negative information may be useful, i.e., a warning about bad weather, The present example provides a mobile telecommunica ice, freezing road Surfaces, fog, sand-storms, rain, Snow, tions device having a position detector, which may be abso sleet, hail, Sun glare, etc., and an indication of dry, warm, lute, relative, hybrid, or other type, and preferably a commu well-illuminated road surfaces may both be useful informa nications device for communicating information, typically tion. location relevant information. The device may serve as a In addition, through the analysis, a number of presump transmitter, transmitting information relevant to the location tions and predictions are made, for example using a Markov (or prior locations) of the device, a receiver, receiving infor 25 chain. Therefore, while the system may predict a most likely mation relevant to the location (or prospective location) of the state of affairs, this alone does not provide sufficient infor device, or a composite. mation for full reliance thereon. For example, the present road In the case of a transmitter device or stand-alone device, a Surface freezing conditions thirty miles ahead on a road may sensor is provided to determine a condition of or about the be a poor indicator of the road conditions when the device is device or its context. This sensor may populate a map or 30 at that position. In addition to changes in the weather, human mapping system with historical map data. action may be taken, Such as road salt, sand, traffic, etc., During use, a receiving device seeks to output location which would alter the conditions, especially in response to a context-relevant information to the user, and therefore in this warning. On the other hand, a report of freezing road condi embodiment includes a human user interface. Typically, in a tions one mile ahead would generally have high predictive vehicle having a general linear or highly constrained type 35 value for the actual road conditions when the device is at that path, a position output is not a critical feature, and may be location, assuming that the vehicle is traveling in that direc suppressed in order to simplify the interface. On the other tion. In each case, the risk is similar, but the reliability is hand, there are a number of navigation system user interfaces different; a composite indication would possibly reduce the Suitable for this application, which may be employed as nec 30 mile warning below a noise threshold, even though it may essary. 40 be anticipated that this event would produce substantial traffic A relative position output, however, may be more appro and residual delays, even of the event abated before the priate, indicating a relative position (distance, time, etc.) with vehicle approached the respective Zone of interest. respect to a potential contextually relevant position. In addi In many cases, there is too much raw information to effec tion, especially in Systems where a plurality of different types tively display to the user all relevant factors in making a of sensors or sensed parameters are available, the nature of the 45 reliability or risk determination. Thus, the device outputs a relevant context is also output. Further, as a particular feature composite estimation of the reliability or risk, which may be of the present invention, a risk or reliability assessment is a numeric or non-parametric value. This is output in conjunc indicated to the user. This risk or reliability assessment is tion with the nature of the alert and its contextual proximity. preferably statistically derived, although it may be derived Likewise, a threshold function may be applied to squelch through other known means, for example Boolean analysis, 50 notifications of events with low probability, or of low risk, or fuzzy logic, or neural networks. both. In this case, risk indicates a parameter of an event, if it As stated above, there will generally be a plurality of occurs, relating to a cost or impairment. Reliability relates to events, each with an associated risk and/or reliability and the probability that the event will occur, separate from the risk respective location. The relevance of an event may be pre incurred if it did occur. While it is often possible to product a 55 dicted based on the dynamics of the vehicle in which the composite parameter resulting from a reduction in a per device is transported and the nature of the event. Thus, if the ceived cost of the risk based on its reliability or probability of vehicle requires 170 feet to stop from a speed of 60 MPH, a occurrence, this composite eliminates useful or critical infor warning which might trigger a panic stop should be issued mation. between 170-500 feet in advance. If the warning is triggered For example, the device may provide weather information 60 closer than 170 feet, preferably the warning indicates that the to the user. Through one or more of meteorological data from evasive maneuver will be necessary, rather than a panic stop. standard reporting infrastructure (e.g., NOAA, Accu If the warning is given too far inadvance, the triggering event weather R, etc.), mobile reporting nodes (e.g., mobiles may not even exist as the vehicle approaches, leading to a devices having weather sensors), satellite data, and other false alert. In that case, it is better to present an advisory rather weather data Sources, a local weather map is created, prefer 65 than an action item. ably limited to contextual relevance. In most cases, this In this case, the indication is dependent on a number of weather map is stored locally; however, if the quality of factors. First, there is the reliability of the data upon which the US 8,874,477 B2 69 70 warning is based. Second, there is the reliability of the pre part of the system as a whole. Such means need not be dictive model which extrapolates from, the time the raw data included in every apparatus embodying the invention. is acquired to the conjunction of the device and the location of Radar detectors typically are employed to detect operating the event. Third, there is an assessment of the relative risks of emitters of X (10.5 GHz), K(25 GHz) and Ka (35 GHz) radar responding to a false positive versus failing to respond to a emissions from traffic control devices or law enforcement false negative. Other risks may also be included in the analy personnel for detecting vehicle speed by the Doppler effect. sis. Together, the composite risk is output, for example as a These systems typically operate as Superheterodyne receivers color indicator. Using, for example, a tricolor (red-green which Sweep one or more bands, and detect a wave having an blue) light emitting diode (LED) or bicolor LED (red-green), energy significantly above background. As such, these types a range of colors may be presented to the user. Likewise, in an 10 of devices are subject to numerous sources of interference, audio alert, the loudness or harmonic composition (e.g., har accidental, intentional, and incidental. A known system, monic distortion) of a tone oralert signal may indicate the risk Safety Warning System (SWS) licensed by Safety Warning or reliability. (In the case of loudness, preferably a micro System L.C., Englewood Fla., makes use of such radar detec phone measures ambient noise to determine a minimum loud 15 tors to specifically warn motorists of identified road hazards. ness necessary to indicate an alert). In this case, one of a set of particular signals is modulated The position detector is preferably a GPS or combined within a radar band by a transmitter operated near the road GPS-GLONASS receiver, although a network position detec way. The receiver decodes the transmission and warns the tion system (e.g., Enhanced 911 type system) may also be driver of the hazard. employed. Preferably, the position detector typically achieves LIDAR devices emit an infrared laser signal, which is then an accuracy oft10 meters, and preferably provides redundant reflected off a moving vehicle and analyzed for delay, which sensors, e.g., GPS and inertial sensors, in case of failure or relates to distance. Through successive measurements, a sped error of one of the systems. However, for Such purposes as can be calculated. A LIDAR detector therefore seeks to detect pothole reporting, positional accuracies of 1 to 3 meters are the characteristic pulsatile infrared energy. preferred. These may be obtained through a combination of 25 Police radios employ certain restricted frequencies, and in techniques, and therefore the inherent accuracy of any one Some cases, police vehicles continuously transmit a signal. technique need not meet the overall system requirement. While certain laws restrict interception of messages sent on The position detector may also be linked to a mapping police bands, it is believed that the mere detection and local and/or navigation system and possibly a dead reckoning sys ization of a carrier wave is not and may not be legally tem, in order to pinpoint a position with a geographic land 30 restricted. These radios tend to operate below 800 MHz, and mark. Thus, while precise absolute coordinate measurements thus a receiver may employ standard radio technologies. of position may be used, it may also be possible to obtain Potholes and other road obstructions and defects have two useful data at reduced cost by applying certain presumptions characteristics. First, they adversely affect vehicles which to available data. In an automotive system, steering angle, encounter them. Second, they often cause a secondary effect compass direction, and wheel revolution information may be 35 of motorists seeking to avoid a direct encounter or damage, by available, thereby giving a rough indication of position from slowing or executing an evasive maneuver. These obstruc a known starting point. When this information is applied to a tions may therefore be detected in three ways, first, by ana mapping system, a relatively precise position may be esti lyzing the Suspension of the vehicle for unusual shocks mated. Therefore, the required precision of another position indicative of Such vents; second, by analyzing speed and ing system used in conjunction need not be high, in order to 40 steering patterns of the Subject vehicle and possibly Surround provide high reliability position information. For example, ing vehicles; and third, by a visual, ultrasonic, or other direct where it is desired to map potholes, positional accuracy of 10 sensor for detecting the pothole or other obstruction. Such cm may be desired, far more precise than might be available direct sensors are known; however, their effectiveness is lim from a normal GPS receiver mounted in a moving automo ited, and therefore an advance mapping of such potholes and bile. Systems having Such accuracy may then be used as part 45 other road obstructions greatly facilitates avoiding vehicle of an automated repair system. However, when combined damage and executing unsafe or emergency evasive maneu with other data, location and identification of such events is vers. An advance mapping may also be useful in remediation possible. Further, while the system may include or tolerate of such road hazards, as well. inaccuracies, it is generally desired that the system have high Traffic jams occur for a variety of reasons. Typically, the precision, as compensation for inaccuracies may be applied. 50 road carries traffic above a threshold, and for some reason the A typical implementation of the device provides a memory normal traffic flow patterns are disrupted. Therefore, there is for storing events and respective locations. Preferably, further a dramatic slowdown in the average vehicle speed, and a information is also stored. Such as a time of the event, its reduced throughput. Because of the reduced throughput, even character or nature, and other quantitative or qualitative after the cause of the disruption has abated, the roadways may aspects of the information or its source and/or conditions of 55 take minutes to hours to return to normal. Therefore, it is acquisition. This memory may be a solid state memory or typically desired to have advance warnings of disruptions, module (e.g., Flash memory), rotating magnetic and/or opti which include accidents, icing, rain, Sun glare, lane closures, cal memory devices, or other known types of memory. road debris, police action, exits and entrances, and the like, in The events to be stored may be detected locally, such as order to allow the driver to avoid the involved region or plan through a detector for radar and/or laser emission source, 60 accordingly. Abnormal traffic patterns may be detected by radio Scanner, traffic or road conditions (mechanical vehicle comparing a vehicle speed to the speed limit or a historical sensors, visual and/or infrared imaging, radar or LIDAR average speed, by a visual evaluation of traffic conditions, or analysis, acoustic sensors, or the like), places of interest by broadcast road advisories. High traffic conditions are asso which may be selectively identified, itinerary stops, and/or ciated with braking of traffic, which in turn results in decel fixed locations. The events may also be provided by a remote 65 eration and the illumination of brake lights. Brake lights may transmitter, with no local event detection. Therefore, while be determined by both the specific level of illumination and means for identifying events having associated locations is a the center brake light, which is not normally illuminated. US 8,874,477 B2 71 72 Deceleration may be detected by an optical, radar or LIDAR deceleration. As the vehicle comes to a stop, the springs and sensor for detecting the speed and/or acceleration state of shock absorbers of the Suspension undergo a damped oscil nearby vehicles. lation, which is relatively comfortable, and destabilizes the While a preferred embodiment of the present invention vehicle and its contents. To the extent that a rolling stop is employs one or more sensors, broadcast advisories, including 5 safe, efficient and illegal, a change in law is advocate. those from systems according to or compatible with the According to one aspect of the present invention, the driver present invention, provide a valuable source of information may locate a deceleration target and/or a target speed. The relating to road conditions and information of interest at a vehicle navigation system may assist, recording an exact particular location. Therefore, the sensors need not form a location of a stop line, geographic (hills, curves, lane marker part of the core system. Further, some or all of the required 10 locations, etc.), weather conditions (ice, sand, puddles, etc.) sensors may be integrated with the vehicle electronics and other circumstances Surrounding the vehicle. Other (“vetronics’), and therefore the sensors may be provided vehicles obstructions or pedestrians, etc. may also be identi separately or as options. It is thereforean aspect of an embodi fied and modeled. Using models of the various components, ment of the invention to integrate the transceiver, and event as well as cost functions associated with each, as well as database into a Vetronics system, preferably using a digital 15 Subjective factors, which may include vehicle occupant time Vetronics data bus to communicate with existing systems, cost and comfort functions, an optimal acceleration or decel Such as speed sensors, antilock brake sensors, cruise control, eration profile may be calculated. The system may therefore automatic traction system, Suspension, engine, transmission, express control over throttle, brakes, transmission shifts, and other vehicle systems. clutch, valve timing, Suspension controls, etc., in order to According to one aspect of the invention, an adaptive cruise optimize vehicle performance. Of course, a fail-safe and control system is provided which, in at least one mode of dead-man override would generally be provided to ensure operation, seeks to optimize various factors of vehicle opera that the vehicle operator endorses the automated control tion, Such as fuel efficiency, acceleration, comfort, tire wear, operation. etc. For example, an automatic acceleration feature is pro See U.S. Pat. Nos. (expressly incorporated herein by ref vided which determines and/or implements a most fuel-effi 25 erence): U.S. Pat. Nos. 6,503,170; 6,470,265; 6,445,308: cient acceleration for a vehicle, for example by controlling 6,292,743; 6,292,736; 6,233,520; 6,230,098: 6,220,986: throttle and transmission. Too slow an acceleration will result 6,202,022: 6,199,001: 6,182,000; 6,178,377; 6,174,262: in increased time at Suboptimal gear ratios, while too fast 6,098,016; 6,092,014; 6,092,005; 6,091,956; 6,070,118; acceleration will waste considerable fuel. In addition, it is 6,061,003; 6,052,645; 6,034,626; 6,014,605; 5,990,825; Sometimes more efficient to vary operating speed than to 30 5,983,154; 5,938,707; 5,931,890; 5,924.406; 5,835,881: maintain a constant speed, even if time is considered a cost 5,774,073; 6,442,473; 4,704,610; 5,712,632; 5,973,616; and factor. Actual operating efficiency may be measured during 6,008,741. vehicle use, allowing an accurate prediction of fuel efficiency The radio used for the communications Subsystem can be under dynamically changing conditions, such as acceleration. radio frequency AM, FM, spread spectrum, microwave, light Vehicle sensors may assist in making a determination that 35 (infrared, visible, UV) or laser or maser beam (millimeter optimum acceleration is safe; objects both in front and behind wave, infrared, visible), or for short distance communica the vehicle may be sensed. If an object is in front of the tions, acoustic or other communications may be employed. vehicle, and the closing speed would predict a collision, then The system preferably employs an intelligent transportation the acceleration is decreased, or even brakes applied. If an system (ITS) or Industrial, Scientific and Medical (ISM) allo object is rapidly advancing from the rear, the acceleration 40 cated band, such as the 91.5 MHz, 2.4 MHZ, 5.8GHz or DSRC may be increased in order to avoid impact or reduce speed band. (The 2.350-2.450 GHz band corresponds to the emis differential. See, U.S. Pat. No. 6,445,308 (Koike, Sep. 3, sion of microwave ovens, and thus the band suffers from 2002, Positional data utilizing inter-vehicle communication potentially significant interference). The 24.125 GHz band, method and traveling control apparatus), U.S. Pat. No. 6,436, corresponding to K-band police radar, may also be available; 005 (Bellinger, Aug. 20, 2002, System for controlling driv 45 however, transmit power in this band is restricted, e.g., less etrain components to achieve fuel efficiency goals), U.S. Pat. than about 9 mW. The signal may be transmitted through free No. 6,418.367 (Toukura, et al., Jul. 9, 2002, Engine transmis space or in paths including fiber optics, waveguides, cables or sion control system), expressly incorporated herein by refer the like. The communication may be short or medium range CCC. omnidirectional, line of sight, reflected (optical, radio fre Likewise, the operation of a vehicle may be optimized 50 quency, retroreflector designs), satellite, secure or non-se approaching a stop. Such as a stop sign, red light, or the like. cure, or other modes of communications between two points, In this case, the system optimization may be more complex. that the application or state-of-the-art may allow. The particu In addition to fuel economy, wear on brakes, engine (espe lar communications methodology is not critical to the inven cially if engine drag braking is employed), transmission, tires, tion, although a preferred embodiment employs a spread Suspension, time, accident-related risks, and the like, may 55 spectrum microwave transmission. also be included. In the case of a stop sign, the issue also arises A particularly preferred communications Scheme employs with respect to a so-called “rolling stop'. Such a practice steerable high gain antennas, for example a phased array or provides that the vehicle does not actually stop, but reaches a mechanically steered directional antenna, which allows a sufficiently low speed that the driver could stop if required by higher spatial reuse of communications bands and higher circumstances. While this practice is technically considered a 60 signal to noise ratio that an omnidirectional antenna. traffic violation, in many instances, it is both efficient and A number of Dedicated Short Range Communications useful. For example, a stop line is often located behind an (DSRC) systems have been proposed or implemented in order intersection, with impaired visibility. Thus, the vehicle might to provide communications between vehicles and roadside come to a complete stop, begin to accelerate, and then find systems. These DSRC systems traditionally operate in the that the intersection is not clear, and be forced to stop again. 65 900 MHz band for toll collection, while the FCC has recently One particular reason for a rolling stop is the storage of made available 75 MHz in the 5.850-5.925 GHz range for energy in the vehicular Suspension during acceleration and Such purposes, on a co-primary basis with microwave com US 8,874,477 B2 73 74 munications, satellite uplinks, government radar, and other Range Communications (DSRC) Medium Access Control uses. However, spectrum is also available in the so-called (MAC) and Physical Layer (PHY) Specifications (This U-NII band, which encompasses 5.15-5.25 GHZ (indoors, 50 standard, ASTM E2213-02 Standard Specification for mW) and 5.25-5.35 (outdoors, 250 mW). A Japanese ITS Telecommunications and Information Exchange Between (“ETC) proposal provides a 5.8 GHZ full duplex interroga Roadside and Vehicle Systems—5 GHZ Band Dedicated tion system with a half duplex transponder, operating at about Short Range Communications (DSRC) Medium Access 1 megabit per second transmission rates. Control (MAC) and Physical Layer (PHY) Specifications, In August 2001, the DSRC standards committee (ASTM describes a medium access control layer (MAC) and physi 17.51) selected 802.11a as the underlying radio technology cal layer (PHY) specification for wireless connectivity for DSRC applications within the 5.850 to 5.925 GHz band. 10 using dedicated short-range communications (DSRC) ser The IEEE 802.11a standard was modified, in a new standard vices. This standard is based on and refers to the Institute of referred to as 802.11a R/A (roadside applications) to meet Electrical and Electronics Engineers (IEEE) standard DSRC deployment requirements, and includes OFDM modu 802.11 (Wireless LAN Medium Access Control and Physi lation with a lower data rate, 27 MBS for DSRC instead of 54 cal Layer specifications), and standard 802.11a (Wireless MBS for 802.11a. 15 LAN Medium Access Control and Physical Layer specifi Proposed DSRC applications include: cations High-Speed Physical Layer in the 5 GHz band). Emergency Vehicle Warning Currently, emergency This standard is an extension of IEEE 802.11 technology vehicles only have sirens and lights to notify of their into the high-speed vehicle environment. It contains the approach. With DSRC, the emergency vehicle can have the information necessary to explain the difference between traffic system change traffic lights to clear traffic along its IEEE 802.11 and IEEE 802.11a operating parameters intended route. Also, this route information can be broadcast required to implement a mostly high-speed data transfer to other cars to provide user/vehicle specific directions to service in the 5.9-GHz Intelligent Transportation Systems reduce collisions. Radio Service (ITS-RS) band or the Unlicensed National Traffic congestion data can be exchanged between Information Infrastructure (UNII) band, as appropriate). vehicles. On-coming traffic exchanges information on traffic 25 ANSI X3.38-1988 (R1994) Codes Identification of status ahead so that vehicle navigation systems can dynami States, the District of Columbia, and the Outlying and cally provide the best route to a destination. Associated Areas of the United States for Information An industry standard interoperable tolling platform could Interchange expand the use of toll Systems or processing payments at ASTM PS111-98-Specification for Dedicated Short Range parking lots, drive-through establishments (food, gas), etc. 30 Communication (DSRC) Physical Layer Using Micro Safety applications could benefit from use of DSRC. The wave in the 902 to 928 MHZ Band DSRC automaker consortium (DaimlerChrysler, GM, Ford, ASTM PS105-99 Specification for Dedicated Short Range Toyota, Nissan, & VW) are seeking ways to enhance passen Communication (DSRC) Data Link Layer: Medium ger safety with DSRC communications. For example, in a Access and Logical Link Control typical collision, a car has only 10 milliseconds to tighten 35 CEN Draft Document: prENV278/9/#65 Dedicated Short seatbelts, deploy airbags, etc. If an additional advance warn Range Communication (DSRC)—Application Layer ing of 5 milliseconds was provided, one could tighten seat (Layer 7) belts, warm-up the airbags, etc. to prepare the car for colli IEEE Std 1489-1999 Standard for Data Dictionaries for Sion. Using radar, GPS data, etc. a car can determine that a Intelligent Transportation Systems—Part 1: Functional collision is imminent, and it can then notify the car about to be 40 Area Data Dictionaries hit to prepare for collision. GSS Global Specification for Short Range Communication. It is noted that the present technology has the capability for The platform for Interoperable Electronic Toll Collection streamlining transportation systems, by communicating traf and Access Control fic conditions almost immediately and quickly allowing deci ISO 3166-1: 1997 Codes for the representation of names of sions to be made by drivers to minimize congestion and avoid 45 countries and their subdivisions—Part 1: Country codes unnecessary slowdowns. A particular result of the implemen ISO 3779:1983- Road vehicles Vehicle identification tation of this technology will be a reduction in vehicular air numbering (VIN)—Content and structure pollution, as a result of reduced traffic jams and other ineffi ISO/IEC 7498-1: 1994 Information technology Open cient driving patterns. To further the environmental protec Systems Interconnection Basic Reference Model: The tion aspect of the invention, integration of the database with 50 Basic Model cruise control and driver information systems may reduce ISO 7498-2: 1989—Information processing systems—Open inefficient vehicle speed fluctuations, by communicating to Systems Interconnection Basic Reference Model Part the driver or controlling the vehicle at an efficient speed. As a 2: Security Architecture part of this system, therefore, adaptive speed limits and intel ISO/IEC 7498-3:1997 Information technology Open ligent traffic flow control devices may be provided. For 55 Systems Interconnection Basic Reference Model: Nam example, there is no need for fixed time traffic lights if the ing and addressing intersection is monitored for actual traffic conditions. By ISO/IEC 7498-4:1989 Information processing systems— providing intervehicle communications and identification, Open Systems Interconnection Basic Reference Such an intelligent system is easier to implement. Likewise, Model Part 4: Management framework static speed limits may be eliminated in favor of a system 60 ISO 3780:1983- Road vehicles World manufacturer iden which employs intelligence to optimize the traffic flow pat tifier (WMI) code terns based on actual existing conditions and safe and reason ISO/IEC 8824-1:1995 Information technology Abstract able limits, rather than a static set of rules which are applied Syntax Notation One (ASN.1): Specification of basic nota universally and without intelligence. See: tion ASTM E2213-02—Standard Specification for Telecommu 65 ISO/IEC 8825-2:1996 Information technology ASN.1 nications and Information Exchange Between Roadside encoding rules: Specification of Packed Encoding Rules and Vehicle Systems—5 GHZ, Band Dedicated Short (PER) US 8,874,477 B2 75 76 ISO TC204 WG15 Committee Of Japan TICS/DSRC retroreflector systems exist. This technique may also be used DSRC Application Layer High Data Rate mobile environ to augment active communications schemes, for example ment allowing a scanning or array antenna to determine an optimal ASTM E2158-01- Standard Specification for Dedicated position or spatial sensitivity or gain, or a phase array or Short Range Communication (DSRC) Physical Layer synthetic aperture array to define an optimal spatial transfer Using Microwave in the 902-928 MHz Band function, even in the presence of multipath and other types of ASTM PS 105-99 Standard Provisional Specification for signal distortion and/or interference. Dedicated Short Range Communication (DSRC) Data According to one embodiment of the invention, a plurality Link Layer ofantenna elements are provided. These may be, for example, IEEE Std 1455-1999—Standard for Message Sets for 10 a set of high gain antennas oriented in different directions, or Vehicle/Roadside Communications an array of antennas, acting together. Accordingly, the IEEE Std 802.11-1999 Information Technology Tele antenna structure permits a spatial division multiplexing to communications and information exchange between sys separate channels, even for signals which are otherwise indis tems—Local and metropolitan area networks—Specific tinguishable or overlapping. For example, this permits a requirements—Part 11: Wireless LAN Medium Access 15 single antenna system to communicate with a plurality of Control and Physical Layer specifications other antenna systems at the same time, with reduced mutual IEEE Std 802.11a-1999 Information Technology Tele interference. Of course, these communications channels may communications and information exchange between sys be coordinated to further avoid overlap. For example, the tems—Local and metropolitan area networks—Specific communications band may be Subdivided into multiple chan requirements—Part 11: Wireless LAN Medium Access nels, with respective communications sessions occurring on Control and Physical Layer specifications: High Speed different channels. Likewise, a plurality of different bands Physical Layer in the 5 GHz band may be simultaneously employed, for example 802.11g (2.4 Each of which is expressly incorporated herein in its GHz), 802.11a (5.4 GHz), and 802.11a R/A (5.9 GHz). In entirety. another embodiment, a mechanically scanning high gain It is noted that the present technology has the capability for 25 antenna may provide directional discrimination. Such an streamlining transportation systems, by communicating traf antenna may be, for example, a cylindrical waveguide elec fic conditions almost immediately and quickly allowing deci tromagnetically reflective at one end, having a diameter cor sions to be made by drivers to minimize congestion and avoid responding to the wavelength of the band, and with a probe unnecessary slowdowns. A particular result of the implemen extending about half-way into the cylinder perpendicularly to tation of this technology will be a reduction in vehicular air 30 its axis, at about a quarter wavelength from the reflective end. pollution, as a result of reduced traffic jams and other ineffi Likewise, a so-called “Pringles Can Antenna', which has cient driving patterns. To further the environmental protec been termed a Yagi design, and known modifications thereof, tion aspect of the invention, integration of the database with have been deemed useful for extending the range of 802.11b cruise control and driver information systems may reduce communications. inefficient vehicle speed fluctuations, by communicating to 35 According to one embodiment, a radome may be provided the driver or controlling the vehicle at an efficient speed. As a on the roof of a vehicle, having therein an antenna array with, part of this system, therefore, adaptive speed limits and intel for example, 4-64 separate elements. These elements, are, for ligent traffic flow control devices may be provided. For example, simple omnidirectional dipole antennas. The size example, there is no need for fixed time traffic lights if the and spacing of the antenna elements is generally determined intersection is monitored for actual traffic conditions. By 40 by the wavelength of the radiation. However, this distance providing intervehicle communications and identification, may be reduced by using a different dielectric than air. For Such an intelligent system is easier to implement. Likewise, example, see U.S. Pat. No. 6,452.565, expressly incorporated the 55 miles per hour speed limit that was initially presented herein by reference. See, also antenova.com (Antenova Ltd., in light of the “oil crisis” in the 1970s, and parts of which Stow-cum-Quy, Cambridge, UK). persist today even in light of relatively low petroleum pricing 45 A preferred radome also includes GPS antenna, as well as and evidence that the alleged secondary health and safety cellular radio antenna (IS-95, PCS, GSM, etc.). benefit is marginal or non-existent, may be eliminated in In a preferred embodiment, the communications device favor of a system which employs intelligence to optimize the employs an unlicensed band, such as 900 MHz (902-928 traffic flow patterns based on actual existing conditions, MHz), FRS, 49 MHz, 27 MHz, 2.4-2.5 GHz, 5.4 GHz, 5.8 rather than a static set of rules which are applied universally 50 GHz, etc. Further, in order to provide noise immunity and and without intelligence. band capacity, spread spectrum RF techniques are preferred. The communications device may be a transmitter, receiver As appropriate, multiple-input multiple-output (MIMO) or transceiver, transmitting event information, storing antenna radio technologies may be employed to increase received event information, or exchanging event information, channel bandwidth, though in a mobile system the benefits of respectively. Thus, while the system as a whole typically 55 this technology may be limited. involves a propagation of event information between remote In one embodiment, communications devices are installed databases, each system embodying the invention need not in automobiles. Mobile GPS receivers in the vehicles provide perform all functions. location information to the communications devices. These In a retroreflector system design, signal to noise ratio is GPS receivers may be integral or separate from the commu improved by spatial specificity, and typically coherent detec 60 nications devices. Event detectors, such as police radar and tion. An interrogation signal is emitted, which is modulated laser (LIDAR) speed detectors, traffic and weather condition and redirected back toward its source, within a relatively wide detectors, road hazard detectors (pot holes, debris, accidents, range, by a receiver. Thus, while the receiver may be “pas ice, mud and rock slides, drunk drivers, etc.), traffic speed sive', the return signal has a relatively high amplitude (as detectors (speedometer reading, sensors for detecting speed compared to nonretroreflective designs under comparable 65 of other vehicles), speed limits, checkpoints, toll booths, etc., conditions) and the interrogator can spatially discriminate may be provided as inputs to the system, or appropriate sen and coherently detect the return signal. Both optical and RF sors integrated therein. The system may also serve as a beacon US 8,874,477 B2 77 78 to good Samaritans, emergency workers and other motorists output produced by the system. It is noted that the timecon in the event of accident, disablement, or other status of the stant may also be replaced with an expression or analysis host vehicle. which is a function of time, that is, to account for diurnal, It is noted that at frequencies above about 800 MHz, the weekly, seasonal, annual, etc. changes. Such expression or transmitter signal may be used as a part of a traffic radar analysis need not be repetitive; for example, after an abnor system. Therefore, the transmitted signal may serve both as a mality in traffic flow, traffic patterns tend to remain distorted communications stream and a sensor emission. Advanta for a long period (e.g., hours) after the abnormality is cor geously, an electronically steerable signal is emitted from an rected, or after a driverpasses the abnormality; this distortion array. Reflections of the signal are then received and analyzed is both temporally and spatially related to the original abnor for both reflection time coefficients and Doppler shifts. Of 10 mality, and may be statistically estimated. Chaos, fractal and/ course, a radar may use static antennas and/or mechanically or wavelet theories may be particularly relevant to this analy scanning antennas, and need not completely analyze the sis, and employed as desired to analyze or present the data. return signals. In outputting information directly to a human user, thresh Functions similar to those of the Cadillac (GM) On-Star olds are preferably applied to limit output to events which are system may also be implemented, as well as alarm and Secu 15 of immediate consequence and apparent importance. For rity systems, garage door opening and "Smart home integra example, if the communications device is installed in a tion. Likewise, the system may also integrate with media and vehicle, and the information in the memory indicates that a entertainment systems. See, U.S. Pat. Nos. 6,418,424; 6,400, pothole, highway obstruction, or police radar “trap' is ahead, 996; 6,081,750; 5,920,477:5,903.454; 5,901,246; 5,875,108; the user is informed. Events in the opposite direction (as 5,867,386; 5,774,357, expressly incorporated herein by ref. determined by a path or velocity record extracted from the erence. These systems may reside in a fixed location, within position detector) are not output, nor are Such events faraway. the vehicle, or distributed between fixed and mobile loca On the other hand, events such as road icing, flooding, or the tions. The system may also integrate with a satellite radio like, are often applicable to all nearby motorists, and are system, and, for example, the satellite radio antenna may be output regardless of direction of travel, unless another com included in the antenna system for other communication sys 25 munications device with event detector indicates that the tems within the vehicle. event would not affect the local communications device or the The memory stores information describing the event as vehicle in which it is installed. well as the location of the event. Preferably, the memory is not According to one embodiment of the invention, relevance organized as a matrix of memory addresses corresponding to of information and information reliability are represented as locations, e.g., a 'map', but rather in a record format having 30 orthogonal axes. For each set of facts or interpretation (hy explicitly describing the event and location, making Storage pothesis) thereof, a representation is projected on the plane of the sparse matrix more efficient and facilitating indexing defined by these two axes. This representation for each event and Sorting on various aspects of each data record. Additional generally takes the form of a bell curve, although the statistics information, Such as the time of the event, importance and for each curve need not be Gaussian. The area under the nature of the event, expiration time of the event, source and 35 Superposed curves, representing the constellation of possible reliability of the event information, and commercial and/or risks or relevances, are then integrated, starting with rel advertising information associated with the event may be evance=1.00 (100%), proceeding toward relevance=0.00 stored. The information in the memory is processed to pro (0%). As the area under a given peak exceeds a threshold, vide a useful output, which may be a simple alphanumeric, which need not be constant, and indeed may be a function of Voice (audible) or graphic output of the telecommunications 40 relevance or reliability, and/or subjective factors, the event is system. In any case, the output is preferably presented in a presented as a warning output to the user. This method Sorted order according to pertinence, which is a combination ensures that the output includes the most relevant events of the abstract importance of the event and proximity, with before less relevant events, but excluding those events with “proximity” weighted higher than “importance'. Once a low reliability. Using a dynamic threshold, highly relevant communication or output cycle is initiated, it may continue 45 events of low reliability are presented, while low relevance until the entire memory is output, or include merely output a events of even modest reliability are suppressed. It is possible portion of the contents, for example the contextually relevant for the threshold to exceed 1.0, that is, a complete suppression portion having a risk and/or reliability exceeding a threshold. of irrelevant events, regardless of reliability. Typically, a navigation system includes a raster 'map' of Alternately, the event projection into the relevance-reli geographic regions, which is further linked to a database of 50 ability plane may be normalized by a function which accounts features, geocoded to the map. Alternately, the map may itself for the desired response function, with a static threshold be a set of geocoded features, without a raster representation. applied. Various events and features defined by the sensors provided The reason why the determination employs an integration by the present system, or received through a communications of a stochastic distribution, rather than a simple scalar repre link, may therefore be overlaid or integrated into the geo 55 sentation of events, is that this allows certain events with coded features. Advantageously, all of the geocoded features broad distributions, but a mean value below than of another are separately defined from the static geography, and there event with a narrower distribution, to be ranked ahead, as fore may be separately managed and processed. For example, being more significant. This has potentially greater impact for geologic features are persistent, and absent Substantial human events having decidedly non-normal distributions, for activity or natural disaster, are persistent. Other features, such 60 example with significant kurtosis, skew, multimodality, etc., as roads, attractions, and other conditions, are subject to and in which a mean value has no readily interpretable mean change periodically. Each geocoded feature (or indeed, any ing. feature or event, whether geocoded or not) may be associated The present invention therefore provides a method, com with a timeconstant representing an estimated life; as the time prising receiving a set of facts or predicates, analyzing the set since last verification increases, the probability of change also 65 of facts or predicates to determine possible events, determin increases. This may be used to provide a user with an estima ing, from the possible events, a relevance to a user and asso tion of the reliability of the navigation system, or indeed any ciated Statistical distribution thereof, and presenting a ranked US 8,874,477 B2 79 80 set of events, wherein said ranking is dependent on both local set of events or factual predicates are analyzed with relevance and associated Statistical distribution. The associ respect to a received context. The various hypotheses are ated statistical distribution, for example, describes a probabil projected onto a relevance-reliability plane. With respect to ity of existence of an associated event, and the relevance each bidder, the projection of each event is normalized by that comprises a value function associated with that event if it bidder's conveyed utility or valuation function. It is useful to exists, wherein said ranking comprises an analysis of prob maintain the stochastic distribution representation for each ability-weighted benefits from each event to an overall utility event, since this facilitates application of the bidder valuation function for the user. The ranking may comprises a combina utility function. The winning bidder is the bidder with the torial analysis of competing sets of rankings. highest normalized integration of the event representation in It is therefore apparent that each set of events, that is, a set 10 the relevance-reliability projection plane. of facts or factual predicates, or conclusions drawn therefrom, Advantageously, according to an embodiment of the are represented as a distribution projected into a relevance present invention, output information is presented to the user reliability plane. On the abscissa, relevance has a scale of 0 to using a statistical and/or probabilistic analysis of both risk 1. At Zero relevance, the information is considered not useful, and reliability. Riskis, for example, the estimated quantitative whereas at a relevance value approaching 1, the information 15 advantage or disadvantage of an event. In the case of compet is considered very relevant. Since the determination of rel ing risks, a cost function may be employed to provide a evance is generally neither exact nor precise, there is an normalized basis for representation and analysis. While the associated reliability, that is, there is a range of possibilities risk is generally thought of as a scalar value, there is no and their likelihoods relating to a set of presumed facts. The particular reason why this cannot itself be a vector or multi various possibilities sum to the whole, which means that the parameter function, such as a mean and Standard deviation or area under the curve (integral from 0 to 1 of the distribution confidence interval. Reliability is, for example, the probabil curve) should sum to 1, although various mathematical sim ity that the risk is as estimated. Likewise, the reliability may plifications and intentional or unintentional perturbations also be a scalar value, but may also be a complex variable, may alter this integrated area. Relevance requires a determi vector or multiparameter function. nation of context, which may include both objective and 25 Since a preferred use of the risk and reliability estimates is Subjective aspects. Communicated relevance information as part of a user interface, these are preferably represented in typically should be determined with respect to the requestor, their simplest forms, which will typically take a scalar form, not the requestee, although in certain circumstances, the Such as by projection from a high dimensionality space to a requestee (possibly with adjustments) may serve as a proxy low dimensionality space, or an elimination or truncation of for the requestor. There are a number of methods for weight 30 information which is predicted to be of low usefulness in a ing higher relevances above lower relevances. One way is to decision-making process. However, where the risk, or risk determine a transfer function which masks the normalized profile, cannot be simply represented, or such representation distribution with a weighting function. This may be a simple loses significant meaning, a higher dimensionality represen linear ramp, or a more complex function. As discussed above, tation may be employed. For human user interfaces, graphic a numeric integration from 1 to 0, with a respective decision 35 displays are common, which generally Support two-dimen made when the integral of a "curve' representing an event sional graphics, representing three dimensional distributions, exceeds a threshold, allowing multiple decisions to be ranked, for example, x, y, and brightness or color. Using a time is another possibility. sequence of graphic elements, one or more additional dimen Using a hierarchal analysis, this process may occur at sions may be represented. Likewise, some graphic displays multiple levels, until each significant hypothesis is analyzed, 40 are capable of representing depth, and thus Support an addi leaving only putative hypothesis which are insignificant, that tional degree of freedom. Therefore, it can be seen that the is, with sufficient external information to distinguish between risk and reliability are not intrinsically limited to scalar rep the respective possibilities. In order to simplify the outputset, resentations, and where the quantity and quality of the infor redundancy is resolved in favor of the most specific signifi mation to be presented warrants, a higher dimensionality or cant hypothesis, while insignificant hypotheses are truncated 45 additional degrees of freedom may be presented. (not presented). As the number of significant hypotheses In a Voice output system, a sequence of information may be becomes in excess of a reasonable number (which may be an output, trading immediacy and semantic complexity for adaptive or Subjective determination), related hypotheses (ei information content. Complex Sounds or other grammars may ther by the nature of the events or the nature of an anticipated also be employed, especially where the relevance has a short responses) may be grouped. Relatedness of hypotheses may 50 time-constant. be determined based on commonality of factual predicates, According to one embodiment of the invention, risk and resulting user action, or other commonality. That is, the reliability are separately output to the user. It is understood grouping may be the same as, or different from, the hierarchy that both risk and reliability may be output in an integral or of the analysis. interrelated form as well. For example, a driver might wish to It is also noted that the projection need not be in the rel 55 employ a radar detector. A traditional radar detector emits a evance-reliability plane. Rather, the analysis is intended to signal indicative of signal type and signal strength. Based on present useful information: that which represents information these emissions, the driver decides on a course of action. having a potential materiality to the user, and which has Ideally, the driver responds immediately (if necessary) to the significance in a statistical sense. Therefore, a data analysis first detected signal, even if this is of low signal strength or which does not purely separate relevance and reliability, but 60 potentially an artifact. On the other hand, the system accord nevertheless allows a general balancing of these issues, may ing to the present invention may analyze the reliability of the nevertheless be suitable. detected signal as an indicator of risk. For example, on a Referring now to the aforementioned auction for control of highway, an X band radar signal directed from in front of the bandwidth, in which a bidder communicates a value function vehicle, which commences at relatively high signal strength, which must be resolved by the recipient, this type of rel 65 and which occurs in a location having a past history of use as evance-reliability analysis may be used to normalize utility a location for monitoring traffic speeds for enforcement pur functions between respective bidders. To determine a cost, a poses, and which was recently confirmed (e.g., within the past US 8,874,477 B2 81 82 5 minutes) as being an active traffic enforcement site, would According to prior systems, a "risk” is presented to a user as be deemed a high reliability signal. On the other hand, on the a binary signal, modulated binary signal, and/or a scalar same highway, if a continuously emitted (or half-wave 60 Hz value. A signal type (e.g., band, SWS code, etc. for a radar emission from a microwave oven) X band signal is detected, detector, temperature, wind speed, wind direction, baromet in a location where Such a signal is consistently detected by ric pressure and trend, for a weather gauge) may also be other drivers, and none is cited for violation of traffic laws, expressed. Accordingly, as set of orthogonal scalar values is then this detection would be considered a low reliability presented representing different parameters. Certainly, detection of a risk or traffic enforcement radar. While a graphic representations of mean and standard deviation are threshold of reliability may be applied, and thus a “squelch' well known; however, the reliability aspect of the present applied to the risk output, preferably, the reliability signal is 10 presented separately. When risk and reliability are both high, invention is not analogous to a simple standard deviation it for example, an enhanced alert may be presented. When risk typically represents something qualitatively different. For is high but reliability low, an indication may be nevertheless example, a determination of the magnitude of the risk variable presented to the user for his analysis. This scheme would carries its own standard deviation, which, though a possible assist the user in dealing with statistical aberrations, as well as 15 element of a reliability determination, does not address the intentional masking of conditions. For example, a traffic issue of how the measured parameter (with its own statistical enforcement radar System may be intentionally used in an parameters of measurement) relates to the underlying issue, area of normal interference with radar detectors; the system or the impact of the underlying issue on the recipient. In some according to the present invention would present an alert to cases, there with be a direct relationship and near 100% the user of this possibility. correlation between the measured parameter and risk Vari Such analysis is not limited to radar detectors. For able; in other cases, the measured parameter has poor corre example, a bridge may be likely to freeze (i.e., become slip lation with the risk variable, and further analysis is necessary. pery) under certain conditions. Some of these conditions may The system preferably ages event data intelligently, allow be detected, such as local weather, past precipitation, and the ing certain types of events to expire or decrease in impor like. Indeed, recent road sand and salt may also be accounted 25 tance. A traffic accident event more than 12 hours old is likely for. However, uncertainty remains as to the actual road Sur stale, and therefore would not be output, and preferably is face conditions, which may change over the course of a few purged; however, locations which are the site of multiple minutes. Therefore, the system according to the present accidents may be tagged as hazardous, and the hazard event invention may determine the risk, i.e., slippery road condi output to the user as appropriate. tions, and the reliability of its determination. This reliability 30 A temporal analysis may also be applied to the event data, may be estimated from actual past experience of the system in and therefore diurnal variations and the like accounted for. question, as well as from other systems including appropriate Examples of this type of data include rush hour traffic, sun sensors, for which data is available. glare (adjusted for season, etc.), vacation routes, and the like. According to the present invention, to risk tolerance, or Thus, user outputs may be provided based on proximity, more properly stated, the reliability-adjusted risk tolerance of 35 importance, and optionally other factors. Such as direction, a user may be used to “normalize' or otherwise adjust the speed (over or under speed limit), time-of-day, date or season outputs of the system. Thus, for example, an emergency (e.g., Sun glare), freshness of event recordation, and the like. vehicle may take higher risks than would normally be accept According to the present invention, a stored event may be able. Clearly, if there is a 100% probability that the vehicle analyzed for reliability. Such reliability may be determined will skid on black ice on the road ahead, this risk would be 40 by express rules or algorithms, statistically, or otherwise, unacceptable for any rational driver seeking to continue driv generally in accordance with particular characteristics of the ing. On the other hand, an ambulance driver on an urgent call type of event. Thus, even where a detected value, at the time may be willing to undertake a 5% risk that the road is slippery, of measurement, has a high reliability for indicating an event while a normal driver might be willing to accept only a 1% or condition, over time the reliability may change. risk. The ambulance driver, in the above example, generally 45 U.S. Pat. No. 6,175,803 (Chowanic, et al., Ford Global takes a number of risks, and caution must be balanced to Technologies, Inc.), expressly incorporated herein by refer assure that the goals are met, and indeed that risks are not ence in its entirety, relates to a telematics system which increased as a result of undue caution. For example, driving at employs routing criteria which include a statistical risk index. a slow speed increases the risk that the vehicle will be rear The route and associated risks may be output together, and a ended, or that the driver will fall asleep during the trip. Even 50 risk-minimized route may be automatically selected. pulling over the side of the road does not eliminate risk to According to a preferred embodiment, audio and/or visual Zero, so it is important to do a comparative risk analysis. warnings are selectively provided. In this case, a warning of The risk/reliability analysis is not limited to driving con only a single event is provided at any given time. Typically, a dition alerts. For example, the system may be used to advise visual alert indicator illuminates, and an initial tone alert the user regarding the need for preventive maintenance or 55 indicates the nature of an urgent warning. The visual indicator repair. The system may also be used as part of an entertain also outputs a strength or proximity of the alert. Typically, ment system: What is the likelihood that a channel will broad these basic indicators are illuminated red, because this color cast an undesired commercial within the next minute? Should is Societally associated with alerts, and this causes less con a recording stored in memory be purged in favor of a new striction of the iris of the eye at night. A separate visual recording? What radio station will be most acceptable to the 60 indicator, such as a bar graph, meter, or color coded indicator set of occupants of the vehicle? (e.g., bicolor or tricolor light emitting diode) provides a sepa In some cases, therefore, the risk/reliability analysis may rate reliability or risk of reliance indication. After acoustically be used by an automated system, and need not be presented indicating the nature and strength or proximity of the warn directly to the user; in other instances, the set of information ing, an acoustic indication of reliability or risk of reliance is for presentation to the user. 65 may be enunciated. The visual reliability or risk of reliance Another aspect of the invention involves a method for indicator may be constantly active, while the warning indica presentation of a multidimensional risk profile to a user. tor is preferably selectively active when an alert is present. US 8,874,477 B2 83 84 Typically, alerts will be classified by category, and a sepa The cost function is analogous to a utility function, which rate algorithm applied to determine the appropriate reliability may be perturbed or modified based on subjective factors. As factor, for example an exponential decay. As updated infor Such, automated negotiations are possible based on bidder mation is received or becomes available, this replaces pre and auctioneer contexts, and predetermined and/or adaptive sumably less reliable older data as a basis for a reliability parameters. By communicating complex, non-normalized determination. The system may also anticipate a geographic information, and allowing an ex post facto reduction in change in location of the event, for example a traffic enforce dimensionality or degrees of freedom, greater efficiency may ment officer in motion, or a traffic jam, along with reliability be obtained. information for the prediction. In choosing which information to present, a preferred 10 embodiment according to the present invention analyzes the When multiple alerts are simultaneously active, low prior risk and reliability, to produce a composite weight or cost, ity alerts are Suppressed, and the active higher-priority alerts which may then be compared with other weights or costs, as alternate. Priority of alerts, in this case, may be determined well as a dynamic threshold, which may be separately ana based on the nature of the alert, contextual proximity, the lyzed or implemented as a part of a cost function. Taking a reliability of the measurement of the alert, the reliability of 15 simple case first, information which is immediately appli reliance on the recorded information, and a comparison of the cable, represents a high degree of risk, and which is reliable, respective alerts and potential interaction. is presented with a highest weighting. If the same indication At any time, there will likely be a number “issues to be is unreliable, then the presentation is deweighted. A high risk analyzed. In order to provide an efficient user interface, these with a low reliability would compete with a low risk with high issues are analyzed to determine urgency or importance, and reliability for presentation through the user interface. As pre only those which meet criteria are particularly presented. For viously discussed, a cost function may be used to factor in example, the fact that the fuel gauge reads half-full is not external or artificial considerations as well. normally a cause for particular alert. However, if the vehicle If the risk or reliability changes as a function of time, and is passing a gas station which has a relatively low price, the this is the significant temporal relationship, then these factors alert may be welcome. Without further information, these 25 may be changed updated, and the user interface modified facts together reach a Sufficient importance to produce an according to a present condition. In some cases, the presen alert. See, U.S. Pat. No. 6,484,088 (Reimer, Nov. 19, 2002, tation relating to an event need not be continuous. That is, as Fuel optimization system with improved fuel level sensor), a result of presentation to the user, the cost function is expressly incorporated herein by reference. That is, the risk dynamically modified, and the event is not again represented (need for fuel; capacity to purchase additional fuel; distance 30 until the cost function exceeds the presentation threshold. The to next gas station and margin of safety given present fuel change in cost function may indeed be purely a function of supply; etc.), ands the reliability (fuel price predicted to be time, or take into consideration dynamically changing vari cheaper than other fuel along predicted vehicle path before ables. For example, if a traffic jam is ten minutes ahead on the urgent need for fuel; etc.), together meet a “threshold road (using predicted travel speeds), and there are a number of (which, of course, may be particularly dynamic in nature). 35 opportunities within the path leading toward the traffic to Additional information, however, may reduce the importance circumvent it, the urgency of taking a detour is low. As the of this information below a threshold level; for example, the time until the traffic decreases, or as the last opportunities for present trip is time critical; the same gas station is predicted to detour are approaching, any decision by the user become be passed a number of times before the fuel tank is empty; critical. This required decision is, in this case, the risk. On the other stations predicted to be passed have lower prices; pric 40 other hand, the traffic may be caused by a traffic light or other ing is anticipated to be more advantageous at a later time (e.g., temporary obstruction. Therefore, the reliability of the risk gas sale on Monday; anticipated trip to another locale with indication will depend on an analysis of the Surrounding lower gas prices, etc.), etc. Thus, the set of facts including circumstances and the likelihood that the predicted risk will available information is analyzed, for example using Baye be the actual risk. Time, in this case, is not clearly indepen sian techniques, Hierarchal Markov Models or other tech 45 dent of the other factors, and therefore need not represent an niques, to predict the importance to the user. Each of these independent output to the user. It is noted that Such analysis of facts or predicates, or sets of facts and/or predicates, of risk and reliability may be facilitated by a wavelet domain course, has its own estimated reliability, and thus the overall transform, which need not be a discrete wavelet transform conclusion is thereby limited. Accordingly, this reliability of (DWT), although the binary decomposition properties of this the logical conclusion is output along with the conclusion 50 transform may prove convenient or advantageous in various itself. circumstances. In particular, the purpose of the transform is With sufficient facts or predicates available, there may be not necessarily a storage or computation-efficient represen competing outputs, both relating to fuel use, conservation, tation; rather, the purpose is to significantly separate degrees and refill strategies and otherwise. Thus, the system must of freedom to simplify the statistical and probabilistic analy compare the competing prospective outputs to determine 55 sis. It is also noted that the particular wavelets may be com which are actually presented. It may be useful in Such circum plex, high dimensionality, asymmetric functions, and need stances to compute a cost function for presenting this data. In not be wavelets of a traditional kind used in image or generic this way, for example, an advertiser or other external influ data compression. ence maybe permitted to impact the overall analysis, e.g., It may also be useful to transform the data into various presentation of data though the user interface. This cost func 60 domains, such as time, frequency, wavelet, alternate iterated tion may also balance driving conditions: for example, when function system, or the like, for filtering and denoising. Pref stopped at a traffic light, less urgent messages may be pre erably, adaptive thresholds are employed, although in many sented with lower risk of driver distraction. The user interface instances the filtering may be performed in a context-inde typically supports only a limited amount of information to be pendent manner. On the other hand, where appropriate, the conveyed, and ergonomics may further limit the amount of 65 filtering may be context sensitive, that is, the modifications of information. Thus, the issue of screening information for the data set during the filtering are dependent on a calculated presentation to the user typically arises. risk, reliability, or relevance, or other parameter. Further US 8,874,477 B2 85 86 analysis may be performed either in the transform domain, tion of phantom or spurious records. Typically, such corrup inverse transform to the original representation, or using a tions are made in Such manner that a corresponding filter in a different transform. receiving unit, with high reliability, will be able to nearly, but It is also possible to employ a fractal (iterated function not absolutely, “uncorrupt” the data. However, without system) analysis and/or transform of the data. In this case, a 5 knowledge of the actual corruption parameters, which are not function within a space, of any dimensional order, is decom transmitted, the reconstruction is statistical and not lossless. posed into a representation of a set of components, which may Therefore, with high reliability, the content of the database is include continuous functions (wavelet) or discontinuous communicated, but not in Such manner that anyone could functions (geometric shape), each of which may be trans opine that individual data within the database is real. For lated, Scaled only any axis, and amplitude scaled. Indeed, 10 example, a database with GPS and chronology will necessar where convenient, the function within a space may be decom ily include data which may be used to derive the speed of the posed into a plurality of separate representations. Thus, vehicle. When that speed is in excess of the speed limit, a according to one example, a number of feature-specific transmission or retention of the data may be used as an admis decompositions may be applied where appropriate. In the sion of transgression of speed limit laws. Using a known filter case of non-linear functions, it may be possible to decompose 15 scheme implemented at the receiver, an algorithm at the trans the function into a linear component and a non-linear com mitter may operate on the data to corrupt that data in Such ponent, wherein a relatively simplified non-linear component manner that the receiver will likely correct the data. By apply may be subjected to a type-specific analysis. Thus, it is under ing a parameter at the transmitter, the reliability of the stood that even relatively complex and seemingly intractable received data can be controlled. The transmitter may, for problems may be addressed. It is further noted that incalcu- 20 example, determine that 25% of the data is to be corrupted, lable aspects of a fact or predicate net may be represented and 1% corrupted in such manner that the receive filter does within the context of a reliability variable. As such, a network not accurately reconstruct the data. However, the corrupt 1% is analyzed, and to the extent possible, numeric analysis may be distributed such that 99% is flagged as spurious, based applied to reduce the result to low-dimensionality terms. The on, for example, excess or negative speeds, non-monotonic predicted magnitude or potential magnitude of the residual 25 travel, etc. Thus, 0.01% of the corrupt data is conveyed with function or the uncertainty bounds may then be estimated, out being caught, a statistic which is likely less than other, resulting in a contribution to the reliability output. Of course, non-intentional corrupting influences. Each of these param the reliability estimate need be limited to unknowns, and may eters may be independently controlled at the transmitter. also represent a contribution from an analytical technique Likewise, it is even possible for these corrupting parameters which produces a calculated uncertainty. 30 to be transmitted, alerting the receiver that the data may be In a typical process, a data set, which may include a plu Suspect. Again, since these are statistical processes, no single rality of dimensions, is first processed to reduce noise. For data point would have evidentiary reliability. example, error correction and detection algorithms may be Using various cryptographic techniques, such as public applied to eliminate spurious data or correct data which has key infrastructure (PKI), it may also be possible to secretly been corrupted. This process may also include a Subprocess 35 synchronize the internal filters of the communicating devices, for eliminating intentional spurious data, for example, data to maintain high reliability of user alerts, while masking the communicated by a malfeasant, or data generated automati data itself. Thus, using secure hardware and appropriate soft cally in a random or pseudorandom manner to make the entire ware techniques, all or most of the corruptions may be cor dataset unavailable as a source of forensic evidence. This is rected or eliminated. For example, the transmitter uses a discussed in more detail, below. The data may also be filtered 40 pseudorandom noise generator to control a corruption of data or denoised using one or more various algorithms, especially to be transmitted. Information related to the cryptographic where the data is obtained continuously from local sensors. key used to initialize the pseudorandom noise generator is Preferably, one or more model-based or content relevant algo securely conveyed to the receiver, for example using a Ker rithms are employed to optimally process data or portions of beros, Diffie-Hellman key exchange, ElGamal key exchange, data. This later function may be consolidated with a feature 45 or other type of cryptographic key negotiation. The receiver extractor to correlate data with patterns which likely indicate then initializes its own corresponding pseudorandom noise a known event, to classify the signal. generator to generate a synchronized stream, allowing it to A multidimensional hidden Markov tree (HMT) analysis decorrupt the data. Clearly, various techniques, including may be used to process the data. A known principal compo those known in the art, may be combined to remedy weak nent analysis (PCA) may, for example, precede the HMT, to 50 nesses of any given scheme. Preferably, a plurality of differ reduce the dimensionality of the data matrix by extracting the ent algorithms are available, should one or more prove bro linear relationship between the variables and decorrelating ken. the cross correlation in the data matrix. The hidden Markov See, Matthew Crouse and Robert Nowak and Richard tree is a statistical model governing the wavelet coefficients, Baraniuk, “Wavelet-Based Statistical Signal Processing and exploiting its tree structure in the time-frequency domain. 55 Using Hidden Markov Models', Proceedings ICASSP-97 Each wavelet coefficient is modeled as a Gaussian mixture (IEEE International Conference on Acoustics, Speech and with a hidden state variable. See, Detection and Classification Signal Processing), IEEE Transactions on Signal Processing, of Abnormal Process Situations. Using Multi-dimensional 1997, and cited references, expressly incorporated herein by Wavelet Domain Hidden Markov Trees (Nov. 9, 2000), Amid reference. See, also, B. Vidakovic, Wavelet-based nonpara Bakhtazad, www.chem.eng.usyd.edu.au/events/poster 60 metric Bayes methods, Technical Report, ISDS, Duke Uni 2000/presento/ppframe.htm versity. Merlise Clyde, and Heather Desimone and Giovanni In order to prevent a data transmission from being used as Parmigiani, Prediction Via Orthogonalized Model Mixing, self-incriminating evidence, steps may be taken to undermine Journal of the American Statistical Association, the reliability of any single piece of data within a data set. 91 (435): 1197 (1996); Katrin Keller, Souheil Ben-Yacoub, According to one possibility, a random or pseudorandom 65 and Chafic Mokbel, Combining Wavelet-domain Hidden process may be used to corrupt the database. This may take Markov Trees with Hidden Markov Models, IDIAP-RR the form of modifications of existing records and/or genera 99-14 (1999), expressly incorporated herein by reference. US 8,874,477 B2 87 88 See also, Attoor Sanju Nair, Jyh-Charn Liu, Laurence Rilett able to distinguish between an accident and construction, and and Saurabh Gupta, “Non-Linear Analysis of Traffic Flow.” indeed, the individual probabilities of these may be insignifi the 4th International IEEE conference on Intelligent Trans cant. However, together, the probabilities may be significant. portation systems, Oakland Calif., Aug. 25-29, 2001, (ac Likewise, since these are two alternative, generally inconsis cepted), expressly incorporated herein by reference. tent possibilities, the reliability of each will be greatly In like manner, additional dimensions of analysis may be reduced. Grouped together, however, their joint reliability is added, resulting in further modifications of a cost function. estimated to be about the remaining likelihood after the traffic Urgency is a Subset of relevance. Relevance may also be light is accounted for, with high reliability. With respect to treated as an independent factor; that is, not included within relevance, each of these events would have similar relevance, risk or reliability. For example, a fact representing a risk may 10 which would be high, assuming the stopped traffic is along the be known with high certainty, for example, a weather condi itinerary of the vehicle. Thus, a composite alert of "abnormal tion on a road: this fact, however, has low relevance if the car stopped traffic 1 mile ahead; reliability 33% would be a is parked in a covered garage. Thus, according to an aspect of useful compromise to maintain an efficient user interface the invention, the relevance may be considered an indepen while conveying the useful information. Of course, the under dent variable. Typically, in this case, the risk and reliability 15 lying system should generally still compute the probabilities, are together analyzed to determine a cost function; the cost reliability and relevance for each possibility, since this analy function is then filtered using a relevance criteria (which, for sis may yield more useful information and provide better example, produces a modified cost function), and typically guidance to the user. sorted or ranked by weight. This relevance therefore replaces The ranking may, for example, employ a combinatorial a simple threshold with respect to making ultimate decisions analysis of a set of rankings based on a self-consistent prob regarding information presentation to the user. Relevancy ability-weighted utility of each event within a ranked set. That may be determined by explicit input from the user, implicit is, if various events are mutually inconsistent, then a ranking user input, collaborative processes, statistical analysis of is limited by a presumption of the existence of one event, and other user under like circumstances, or the like. It is noted that competing hypotheses are established as different rankings. the cost function may be personalized for each user. 25 In a rigorous sense, the utility may be determined by a math In some cases, a dimensionless cost function is too sim ematical integration of the appropriate function, although in plistic, and leads to results which fail to convey useful infor many instances either the data will be represented as a field mation to the user; in those cases, sets of outputs may be which can be simply Summed, or simplifying presumptions presented based on one or more criteria, or an estimated may be applied to make the evaluation tractable. composite function. Therefore, it is understood that a com 30 According to an aspect of the invention, a user transmits a plex "cost function' or utility function, resulting in an output relevance or cost function to corresponding other users, having various degrees of freedom, may be employed. which then calculate the most useful information to transmit Preferably, the system according to the present invention is based on the circumstances of the intended recipient. Like integrated with a vehicular telematics system, thus providing wise, a plurality of users may exchange their respective rel access to various vehicle data, in addition to environmental 35 evance or cost functions. This relevance or cost function is, data. However, it is not so limited, and may be used in any for example, a current position, immediate itinerary, and any type of man-machine interface wherein complex data is to be other particular relevance factors. Such other factors might presented to a user for human consideration. include heavy load, urgent transit, travel preferences, or the It is noted that, in some instances, a fact or predicate set like. Upon receipt, the device of the other corresponding user may possibly represent a plurality of different events. In this 40 then calculates relevance and/or cost functions using its local case, it may sometimes be useful to group these events data set based on the received parameters. This calculation is together. This is particularly the case if the nature of the alert then used as a filter to determine a priority of data to be and likely response to the alert by the user is similar, regard transmitted. As the time available for transmission grows, the less of the particular event giving rise to the sensor readings. amount of information transmitted may be complete. For In that case, the risks, reliabilities, and relevance are aggre 45 example, two cars traveling adjacent on a highway or parked gated in an appropriate fashion, for example vector Summed near each other may conduct a complete data exchange. When or composite magnitude, and an aggregate cost function out optimizing the broadcast of databased on a plurality of user's put, along with a generic alert. This generic alert may then be relevance or cost functions, a weighting may be applied Subdivided into its components, for example in a lower-hier which balances the maximum good for the many with the archal level user interface output. In this manner, a set of 50 urgent needs of the few. Likewise, accommodations may be possible events, none of which would exceed an alert thresh made for anticipated duration of communication for the old individually, may together exceed the threshold and respective users, and the availability of packetforwarding and indeed receive a high ranking. secondary retransmission. Another way of analyzing this situation is that the system Since all devices share a common transmission medium, it may analyze the available data at a number of hierarchal 55 is generally useful to compute a cost function for use of the levels. At each level, the risk, reliability and optionally rel shared medium as well, allowing peers access to the medium evance is determined, and the result stored. The user interface after the marginal utility for the current user has declined. may then select events based on redundancy and generic Access to the shared medium may also be allocated on a alerts, Superseding the particular ranking of events at a homo round robin or other “fair basis, especially when demand is geneous level of analysis. For example, data indicating 60 highest. Each device preferably monitors all local transmis stopped traffic ahead may be consistent with an accident, stop sions, since these will likely include data of some relevance to light, or construction. These may be divided into normal each device. Likewise, by monitoring Such transmissions, events (with low risk) (traffic light) and abnormal events (with one device may make presumptions as to the state of the local high risk)(accident or construction). The former would not database of another device (especially given a knowledge of generally issue an alert, unless a suitable bypass is available 65 its present position and path), and therefore avoid redundant that would be efficient. The later, on the other hand, would transmissions of data. Likewise, in Such a peer to peer net likely generate an alert. The available information may not be work, a Voting scheme may be instituted, allowing the peer US 8,874,477 B2 89 90 with the “best data, i.e., the data which is most reliable, most In general, the auction parameters may be too complex for accurate, most recent, most detail, or other criteria to transmit interactive user entry. Rather, the user cost function itself with higher priority. represents the user valuation, and therefore is usable as a Known packet data broadcast protocols may be used to bidding function. In the cost function, both subjective and convey the information. Likewise, known peer-to-peer tech objective values are observed. With respect to objective val niques and protocols may be used to communicate, or control ues, the relationship of a user context and an event known by communications, between peers. the recipient provides a relevance, and therefore the objective According to another aspect of the invention, a user may valuation. This objective valuation is then warped by user broadcast or transmit a specific query for information, using Subjective factors. Some users may be quite willing to pay 10 more for a better quality of service. At least a portion of this as at least a part of the query a relevance or cost function. value is distributed to other users; this system allows even Recipients of the broadcast or transmission then execute a those with low valuation to access the network, since these search of their database based on the received query, and deferred users will accumulate credits. In some cases, the respond accordingly. This query may be a broad or narrow credits may be provided with cash value (i.e., an ability of a request for information, and thus need not result in a complete 15 user to extract cash proceeds from the system), while in other exchange of data. cases, these credits are limited to use with the system, with no In order to optimally allocate communications bandwidth, redemption rights. The use of a central authority for the pur users within an area may engage in a local auction, that is, chase of usage credits therefore allows a profit incentive for each user bids for use of the shared medium, with those the authority responsible for the system. A user may therefore deferred and the supplier of information receiving credits. An express a higher valuation by purchasing units from an accounting for these credits may, for example, take place each authority, or by providing value to other users of the system, time a device connects with a central database, for example, which may, for example, require enhanced hardware pur using a “hotspot' or other access to the Internet. These credits chases to include more and/or better sensors of various con may, for example, be converted into economic values. In like ditions. manner, advertisers may also bid for access to users, with 25 A negotiation or auction may also include external ele users, for example, receiving credit for receipt of advertising. ments, such as fixed infrastructure. In this case, the scarce Such bidding may be on an individual or group basis. Typi resource is, for example, the right of way. Elements of the cally, advertising will be relevant, for example as a location fixed infrastructure which are subject to negotiation include based output, but need not be. traffic lights, draw bridges, railroad crossings, etc. Typically, It is also possible to conduct auctions or otherwise account 30 Such infrastructure systems have low intelligence. By for control of the communications medium using a Zero-sum employing communications with interested parties, a more temporal averaging. That is, each user has an a priori equal efficient outcome may be predicted as compared to “fair”, right to access. As a user takes advantage of that access, its though unintelligent decisions. Thus, competing drivers may rights decrease, until exhausted. Over time, rights are added, bid for a right of way or green light. The traffic signal may be and accrued rights expire. For example, rights may have a 35 arbiter of the negotiation, or merely recipient of the defined half-life of 5 minutes, with a regression to a predetermined result. In some instances, the negotiation is free of cost, for value. As more users compete for control over the medium, example, a traffic light with but one car approaching and no cost increases. Suppliers of information may receive partial hazards Surrounding. In this case, the signal allows the driver credits from the consumer. Value transmission may take place to pass, unobstructed. In another instance, a large amount of using a modified micropayment scheme, for example a vari 40 traffic may be present, seeking to pass through an intersec ant of Agora Micropayment Protocol, Agora: A Minimal tion. All of the vehicles seeking to pass present “bids' for the Distributed Protocol for Electronic Commerce', Eran Gabber right, with bids representing common interests or outcomes and Abraham Silberschatz, Bell Laboratories or MPTP, pooled. The aggregate bids are then compared for action. In Micro Payment Transfer Protocol (MPTP) Version 0.1, W3C this case, the transaction may have no economic impact, but Working Draft 22 Nov. 1995, www.w3.org/pub/WWW/TR/ 45 rather the utility functions may be relatively non-subjective. WD-mptp-951122. For example, emergency vehicles may have a non-Subjec Thus, within a cell, each user is a primary recipient, a tively determined high valuation, cars driving toward the secondary recipient, a Supplier, or undefined. A primary intersection with a present state of traffic flow control in their recipient bids for access and control of the medium, i.e., the favor at a medium valuation, and stopped traffic with a low communications band. This bid takes the form of a cell iden 50 valuation. As the duration of the stop increases, a delay factor tification (i.e., the controlling user's location and itinerary), increases the valuation for the stopped traffic to compensate, as well as an economic function representing the required allowing or forcing the signal to change. The objective crite information and valuation thereof by the user. The bids are ria used in this circumstance (which may, for example, be broadcast and each recipient calculates an actual value for the defined by a municipality or traffic engineer) may include bid using its own database and the relevance function. The 55 optimization of pollution, energy efficiency, effects of traffic calculated economic values, filtered by the recipient data flow on other intersections, speed control, and other consid bases, are then broadcast, and the highest actual valuation is erations. deemed winner. A negotiation then occurs between the bidder It is noted that, since external elements communicate using and the holder of the valued information, for payment for the the same communications system, and indeed various com transmission, and other bidders receive a lesser value as a 60 munications systems may share the same band, the concept of concession. Secondary recipients of the information also pay bidding for use of the shared or scarce resource may transcend for the information, based on their respective bids, with a a given communications purpose, and, other than communi redistribution to the other bidders as a concession. Devices cating using a common protocol for the bidding and auction which are not active bidders have no economic effect, process, other users of the band need not communicate public although these may accumulate and use transmitted informa 65 information. tion from others. Thus, an economic redistribution occurs A user may also provide a Subjective element to a context, efficiently, while optimally allocating scarce bandwidth. for example, a driver may be in a rush or be late for a meeting. US 8,874,477 B2 91 92 This may be explicitly input by the user, as a factor which resources, quality of service, cost, and reliability. For adjusts the cost function higher, or may be derived implicitly example, by providing a mobile ad-hoc supplementation for a from observation of user behavior. Likewise, a driver may be fixed cellular infrastructure, the incidence of dropped calls in no particular rush, and therefore place a low relevance to and service unavailability may be reduced. Likewise, cellular information which might be of particular benefit to allow him carriers may allocate their infrastructure build-outs and capi to travel faster. tal investments where the return on investment will be maxi Thus, in a purely fair system, each user is allocated a “fair” mum. On the other hand, cellular users in Such regions may chance for access to the scarce bandwidth resource, and bids employ other users to act as repeaters for extending the effec using equally distributed credits to compensate those users tive range of their equipment. In this case, the compensation deferred, and particularly those users who provide useful 10 information. A user bids using a cost function, representing and/or negotiation therefore for use of the system may come, the maximum value of the resource to that user. Using known in whole, or in part, from the fixed infrastructure provider. On auction theory, for example, the cost to the winning bidder the other hand, if the system is sponsored by the fixed infra may be the price bid by the second-highest bidder. Of course, structure carrier, then the repeater services hosted by each other known auction types may be employed. The cost func 15 node may be at no incremental cost to the fixed service pro tion may be automatically generated by a user based on avail vider. able funds, likely future required use of the funds, a relevance This later possibility provides an interesting opportunity. or context which allows for adaptive bidding based on the Since the fixed cellular infrastructure providers generally value of the information to be provided, and user-subjective own licensed spectrum, the implementation of repeater or ad factors. The actual normalized bid is resolved by the respec hoc services between mobile units may be coordinated cen tive recipients, which then broadcast the results. The maxi trally, with mobile-to-mobile communications using cellular mum value bidder then controls the scarce bandwidth channels, which may be time domain (TDMA), frequency resource until the bid-for communication is completed or domain (FDMA), code division (CDMA and related sys exhausted. In order to avoid inefficient reauction overhead, a tems), or other type of band-sharing scheme in accordance periodic auction may be conducted, with all bidders placed in 25 with the more global standards generally established for these a queue. services. Typically, mobile-to-mobile communications use Clearly, in real world situations, a number of additional omnidirectional antennas, and packet data communications distortions will take place. Bidders may become unavailable may use excess System capacity, e.g., capacity not presently prior to completion of a communication. Interference may being used for voice or other toll or real-time service. The require retransmission. 30 fixed infrastructure my also provide coordination of informa As discussed above, each “limited resource' may be sub tion communication services, local buffering, ad multicast f ject to auction. Preferably, a spatial division multiplexing information of general interest. scheme is employed, wherein each band has one or more It is therefore clear that the present invention may comprise frequency channels. High gain, directional antennas are both disruptive and incremental technologies, and the under employed, such that there is a high degree of frequency reuse 35 lying business model may therefore be modified to suit. within a local area. However, there will be a statistical degree A particular issue which is advantageously addressed dur of competition for the frequencies. In addition, there will be ing the design phase is the security of the system against competition from other competing uses for the band, which “hackers' or malfeasants. This may be dealt with by provid may also engage in an auction scheme for access. Typically, ing a central database of authorized users, with peer reporting by efficiently negotiating an auction between all users of the 40 of accounting and apparent abuse. If a user is suspected of resource (i.e., the overhead for negotiation is negligible as abuse, its access rights may be extinguished. This, in turn, will compared to the actual usage), overall throughput and capac at least prevent the user from engaging in auctions, and, if ity will be increased. transmissions are encrypted or otherwise secure, may prevent For example, each system may include 8-16 transceivers or eavesdropping on normal communications streams. This the ability to conduct 8-16 communication sessions simulta 45 same result may be imposed on a user who exhausts his neously. In the former case, 8-16 directional antennas having credits, although it is preferred that a user who is otherwise in relatively low overlap are arrayed in different directions, pro compliance with normal regulations be permitted to receive viding a physical separation. In the later case, a phased array communications and indeed to gain new credits by transmit or synthetic aperture antenna system electronically defines ting useful information. 8-16 independent apertures, also with low overlap. Each spa 50 Since the system is a packet data system, similar to in many tial domain aperture and its associated coverage area repre respects, and possibly an extension of the Internet, various sents a different resource which may be allocated. Therefore, known Internet security paradigms may be applied and multiple simultaneous negotiations may occur simulta employed. neously. Each aperture may be a separate radio, with packets While it is often useful to engage in fair auctions or games, routed there-between, or the radios may be coordinated. 55 it is also possible to engage in unfair auctions. For example, It is also noted that the communications system may be since there may be external financial requirements for main used not only for packet data communications between peers, tenance of the system, these may be paid by Subscription fees, but also as a real time communication system for data or subsidized by advertisers. The advertiser may transmit its streams, such as Voice communications. In this case, hand own cost function, and bid for presentation to given users, or offs may be necessary between various nodes in order to 60 engage in a broadcast for all users. In this case, more valuable assure continuous end-to-end communications. Such hand users will gain more credits, and therefore have more control offs and multiple hop communications may be predicted in over the network. This is not “fair, but the distortions advance and pre-negotiated. Such communications predic implicit in this technique may be tolerable. Likewise, a bidder tions may, indeed, involve multiple systems, such as various may purchase credits, but typically this purchase will be cellular carriers and protocols, 802.11 hot spots, and a mobile 65 consumed by the service operator, and not paid to the users as ad-hoc network with sporadic links to fixed infrastructure. a whole. However, presumably, this will on the whole reduce This, in turn, allows a balancing of competitive uses for the normal service pricing for all users. Indeed, various promo US 8,874,477 B2 93 94 tional techniques may be used to distort allocation of band the vehicles are relatively close, a line-of-sight communica width, without departing from the general scope of the inven tions scheme may be implemented, such as infrared (e.g., tion. IRdA), while at larger distances (and/or for all distances) a It is noted that this implicit auction process, wherein a user spread spectrum 915 MHz, 2.4 GHz or 5.825 GHz. RF com bids a utility function rather than a normalized economic munications scheme implemented. value, is a distinct part of the invention, applicable to a plu Where multiple communications devices are present rality of contexts and environments, well beyond telematics. within a common communications region, these may be Likewise, the concept of bidding for quality of service to pooled, allowing transmissions from one transmitter to many control a shared resource, against competing users, is also receivers. In addition, within a band, multiple channels may applicable to other contexts, notably peer-to-peer networks, 10 be allocated, allowing multiple communications sessions. In other shared transmission medium networks, and queues. this case, a single arbitration and control channel is provided In order to define a user's subjective preferences, value to identify communications devices and communications functions, and the like, a number of methods may be parameters. Preferably, a communications device has the employed. Certainly, express and explicit inputs may be capability to monitor multiple channels simultaneously, and received from a user. Likewise, a user model may be con 15 optionally to transmit on multiple channels simultaneously, structed by observation of the user. A user may be classified as where channel congestion is low. The channels are typically a member of a group having common preferences, and then frequency division. Where such frequency division channels the preferences associated with the group may serve as a are defined, communications may be facilitated by So-called proxy for the user himself. This is called a collaborative “repeaters', which may itself be a mobile transceiver accord profile, the basis for a collaborative filter. In order to classify ing to the present invention. Preferably, Such a repeater unit a user into a group, personality tests and/or common itself monitors the data stream, and may even process the data attributes may be employed. According to a particular aspect stream based on its internal parameters before passing it on. of the invention, a user may be classified by game play. By In order to assure data integrity and optimize data band using game theory, irrational or subjective user biases may be width, both forward and retrospective error correction are revealed. By using games, a users utility function valuation 25 applied. Data is preferably packetized, with each packet may be assessed. Likewise, risk tolerance and demand for including error detection and correction information. Suc immediate gratification can be determined. Further, game cessful receipt of each packet is acknowledged on a reverse theory in the form of wagering may also assistin determining channel, optionally interspersed with corresponding data economic normalizations. Games, especially with the results packets traveling in the reverse direction (e.g., full duplex augmented by collaborative profiling, may through a limited 30 communications). Where the data error rate (raw or cor number of iterations, elicit relatively detailed information. rected) is unacceptably high, one or more “fallback’ modes Indeed, through inter-relation with commercial sponsorship may be implemented, such as reduced data rates, more fault (or state associated lotteries), the economic incentives and tolerant modulation schemes, and extended error correction risks of the game may be made quite real. and detection codes. Transmitter power may also be modu 35 lated within acceptable limits. Fifth Embodiment A central repository of event data may be provided. Such as on the Internet or an on-line database. In this case, event In communicating data to another communications device, information may be administered remotely, and local storage typically it is desired to transmit (or exchange) all of the minimized or eliminated. memory or all of a “public' portion of the memory, with the 40 Communications with the central database may be con received information sorted and processed by the receiving ducted through cellular infrastructure, wired or wireless local unit and relevant information persistently stored in the area networkhotspots, or in other communications bands and memory. After exchange, conflicts may be resolved by a other communications Schemes. further exchange of information. An error detection and cor Where primary event information storage is remote from rection (EDC) protocol may be employed, to assure accurate 45 the device, preferably local storage is based on an itinerary data transmission. (route) and frequently traveled areas, with less frequently Since the communication bandwidthis necessarily limited, traveled and not prospectively traveled routes stored and the communications channels Subject to noise and crowd remotely. This allows consolidated update of memory by a ing, it is often important to prioritize transmissions. It is noted large number of Sources, with statistical error detection and that, without a complete communication of the memory, it is 50 correction of errant event information. The itinerary informa difficult to determine which events a communications partner tion may be programmed in conjunction with a GPS system is aware of, so that an initial communication may include an and mapping/navigation software. identification of the partners as well as recent encounters with According to one embodiment of the invention, a plurality other partners, to eliminate redundant communications, of functions are integrated into a single device, a sensor or where possible. Vehicles traveling in the same direction will 55 detector for sensor emissions, for example speed control often be in close proximity longer than vehicles traveling in devices, a human computer interface, a computer system opposite directions. Further, the information of relevance to a including processor, memory, and operating System, geospa vehicle traveling in the same direction will differ from the tial positioning device, and wireless communication system. information of relevance to a vehicle traveling in the opposite Preferably, the system supports accessory inputs and outputs, direction. Thus, in addition to an identification of the com 60 which may be through wired or wireless means. The human munications device, the recent path and proposed path and computer interface preferably includes both a graphic display Velocity should also be exchanged. Based on this information, and a natural language (e.g., voice) interface. The computer the data is prioritized and sorted, formatted and transmitted. system preferably possesses Sufficient machine intelligence Since the communications channel will likely vary in depen to filter outputs based on relevance and context, as well as dence on distance between partners, the communications pro 65 interpret inputs as usable commands. tocol may be adaptive, providing increased data rate with Data communications over a wireless link, for communi decreasing distance, up to the channel capacity. Further, when cating between vehicles, preferably is highly compressed and US 8,874,477 B2 95 96 fault tolerant. For digital data, this typically requires error nications may be metered or otherwise controlled externally, detection and correction codes, while for data representing with charges assessed based on usage factors. As discussed analog information, the information may be encoded Such above, units may bid for control over the transmission that more important information is transmitted in a more medium, and an accounting may take place either between robust manner than less important information. For example, corresponding units, with a central database, or both. Thus, a image information may be communicated in a hierarchally Subscription based system is Supported. compressed manner, with higher order information transmit Techniques corresponding to the Firewire (IEEE 1394) ted in a manner less Susceptible to interference and signal copy protection scheme may be implemented, and indeed the fading than lower order information. system according to the present invention may implement or The digital data may be compressed, for example, using a 10 incorporate the IEEE 1394 interface standard. The IEEE 1394 dictionary lookup, run length encoding, and/or model-based key management scheme may be useful for implementing vector quantization method. Thus, since transceivers in accor Subscription schemes and for preventing tampering. dance with this embodiment will typically be within 2000 One way to Subsidize a Subscription-based system is meters from each other, relative position may be relayed in an through advertising revenue. Therefore, the “events' may offset format, with a grid size based on GPS precision and 15 also include messages targeted to particular users, either by required accuracy, e.g., about 50-100 meters. The encoding location, demographics, origin, time, or other factors. Thus, a may be adaptive, based, for example, on stored map informa motel or restaurant might Solicit customers who are close by tion, with information representation density highest on trav (especially in the evening), or set up transponders along high eled routes and lower in desolate areas. Thus, a sort of differ ways at desired locations. Travelers would then receive mes ential-corrected positional coding may be established sages appropriate to time and place. While the user of the between units. system according to the present invention will typically be a By integrating functions, efficiencies are achieved. Thus, a frequent motorist or affluent, the system may also provide single central processor, memory, program store and user demographic codes, which allow a customized response to interface may suffice for all functions. Further, the power each unit. Since demographic information is personal, and supply and housing are also consolidated. While GPS and 25 may indicate traveler vulnerability, this information is pref telecommunication antennas may be maintained as distinct erably not transmitted as an open message and is preferably elements, other portions of the system may also be integrated. not decodable by unauthorized persons. In fact, the demo In a device intended for vehicular applications, the GPS and graphic codes may be employed to filter received informa other functions may be available to other vehicular systems, tion, rather than to broadcast interests. or the required data received from other systems. 30 Commercial messages may be stored in memory, and Communication between communications devices may therefore need not be displayed immediately upon receipt. employ unlicensed spectrum or licensed spectrum, and may Further, such information may be provided on a so-called communicate between mobile units or between mobile and “smart card” or PC Card device, with messages triggered by fixed resources. For example, excess capacity of a traditional location, perceived events, time and/or other factors. In turn, cellular system may be used for inter-vehicle communica 35 the presentation of commercial messages may be stored for tions. Thus, the system may include or encompass a typical Verification by an auditing agency, thus allowing accounting cellular (AMPS, IS-136, IS-95, CDPD, PCS and/or GSM) for advertising fees on an “impression’ basis. type telecommunications device, or link to an external tele The communications device may also receive data through communications device. broadcasts, such as using FM sidebands, paging channels, Even where the cellular telephony infrastructure is not 40 satellite transmission and the like. Thus, locationally or tem involved, mobile hardware may be reused for the present porally distant information need not be transmitted between invention. For example, digital or software defined cellular mobile units. Satellite radio systems may also be integrated. telephone handsets may permit programmed use outside the While low power or micropower design is desirable, in an normal cellular system protocols. automobile environment, typically sufficient power is con According to the present invention, messages may be 45 tinuously available to Support Sophisticated and/or power passed between a network of free roving devices. In order to hungry electronic devices; thus, significant design freedom is maintain network integrity, spurious data should be excluded. provided to implement the present invention using available Thus, in order to prevent a “hacker or miscreant (e.g., over technologies. Zealous police official) from intentionally contaminating the dispersed database, or an innocent person from transmitting 50 Sixth Embodiment corrupted data, the ultimate source of event data is preferably recorded. When corrupt or erroneous data is identified, the FIG. 8 shows a block diagram of a communications device source may then also be identified. The identity of the cor embodiment of the present invention. The mobile communi rupting source is then transmitted or identified, for example to cations device 1 includes a location sensing system 2, pro other radios or to a central database, whereupon, units in the 55 ducing a location output 3; a memory 4, for example storing field may be programmed to ignore the corrupt unit, or to a set of locations and associated events, a telecommunica identify its location as a possible event to be aware of. Further, tions Subsystem 5, for example communicating event and assuming the hardware of the corrupted unit remains opera location information between a remote system and the tional, a code may be transmitted to it deactivating it or memory 4, and a processor 6, for example processing the resetting or reprogramming it. 60 location output in conjunction with the stored locations and Preferably, data is transmitted digitally, and may be associated events in the memory 4, to determine a priority encrypted. Encryption codes may be of a public-key/private thereof. key variety, with key lookup and/or certificate verification, The location sensing system 2 may include a known GPS either before each data exchange, or on a global basis with receiver, which produces data that is analyzed by the proces published updates. In fact, corrupt or unauthorized units may 65 sor 6. In an alternate embodiment, the GPS receiver includes be deactivated by normal and authorized units within the its own processor and outputs coordinate positions, e.g., Car network, thus inhibiting “hacking of the network. Commu tesian coordinates, latitude and longitude, to the communica US 8,874,477 B2 97 98 tions device processor 6, e.g., through a serial port or data bus, cal communication systems 13 may employ various detec such as PC card, Universal serial Bus (USB), Firewire (IEEE tors, including optical homodyne detectors, or other coherent 1394), peripheral connect interface (PCI), or other bus, such optical detectors, or other types of optical sensors, such as as that present within an automobile for communication of PIDs, CCDs, silicon photodiodes, and the like. signals between Subsystems. The location sensing system 5 Under Some circumstances, a wired or dedicated link may also determine a position based on the GLONASS sys between units may be appropriate. For example, a central tem, LORAN, inertial reference, cellular base stations 10', database 20 may provide consolidated and reliable data. The 10", triangulation with fixed radio sources, such as FM radio relevant portion of the database 20 may be downloaded by and television stations, environmental markers and/or tran telephone through a modem 21, either through a physical sponders, or the like. The location system may also be net 10 connection 23 (e.g., POTS or ISDN, line), through a broad work based, for example relying on a cellular network to band Internet connect, or other network 24, to a database produce georeferenced position information. server 25. The memory 4 of the mobile unit may also be The communications Subsystem 5 is, for example, an uploaded to the central database 20, after processing by the 802.11g wireless Ethernet local area network system, having database server 25, during the same connection or session. a router and Switch controlling communications between 8 15 Thus, according to the present invention, the public separate spatially distinct channels, defined by a 'smart switched telephone network 24 may be involved both during antenna'. These spatially distinct channels are agile and hav intermittent mass data communications with a central data ing an aperture capable of being steered in real time to adjust base 20, and also using, for example, cellular telephony 14, for a change in relative position and orientation. The router for the normal operation of the system (e.g., communications and Switch permit forwarding of packets received through between mobile units). one channel to another, as well as local communications As discussed above, general access to and control over the control. The radio transceiver 12, operates in the unlicensed communications channel may be arbitrated on a bid and auc 2.4 GHz band, according to FCC regulations for this type of tion basis, as appropriate to avoid contention. equipment. The system may alternately or additionally com The processor 6 analyzes the information stored in municate in other unlicensed bands, such as 27 MHz, 49 25 memory 4 to provide a prioritized output. Thus, the memory MHZ, FRS band, 900 MHz, 5.4 GHz, 5.8-5.9 GHz using may store information relating to a relatively large number of various known modulation schemes and data communication events, without overwhelming the capacity of a human user or protocols. Further, licensed radio bands may also be used, communications partner. Priority may be based on a number including FM radio sidebands (88-108 MHz), televisionPRO of factors, including proximity of a stored location to a sensed channel, cellular telephony channels, DECT, PCS and GSM 30 location or a spatial-temporal proximity of a stored location channels, and the like. Likewise, satellite systems 16, 17 may to a loci of an itinerary 101, a prospective conjunction 102 of be used to communicate with the mobile communications a sensed location with a stored location, a type of event 103. device 1. Thus, for example, instead of direct communication a type of event and a sensed condition associated with the between mobile units, the existing cellular telephony 10", 10" mobile communications device 104, or other factors or a infrastructure may be used to provide intercell, local, and/or 35 combination of factors. Neural networks, fuzzy logic and/or regional communications between units, controlled by cellu traditional logic paradigms may also be employed to priori lar telephone switching processors 11", 11". These communi tize the outputs. These logical paradigms are provided in cations may be given a lower priority than Voice communi known manner, and, especially in the case of neural network cations on the cellular telephone network, and therefore may based systems, a training aspect may be supplied with the use otherwise excess bandwidth, thus allowing reduced costs 40 system to allow it to adapt to the preferences and capabilities and reduced user fees or Subscription rates. of the user. Thus, for a human user, events which are forth The memory 4 may be of any standard type, for example, coming and important are output, while past events and those one or more of static random access memory, dynamic ran in the distant future, if at all, are low priority. On the other domaccess memory, ferroelectric memory, magnetic domain hand, for communications with other devices, the prioritiza memory (e.g., diskette, hard disk), non-volatile semiconduc 45 tion is primarily in consideration of the fact that the commu tor memory (e.g., UV-EPROM, EEPROM, Flash, non-stan nication between units may be only short lived; therefore, the dard electrically erasable programmable non-volatile data is communicated in order to priority, preferably of the memory), optically readable memory (e.g., R-CDROM, RW recipient device. In an adaptive device, if the user believes CDROM, R-DVD, DVD-RAM, etc.), holographic memory, that the information from the device is inappropriate, a simple and the like. Preferably, common memory devices, such as 50 input is provided, which is later analyzed to alter the infor EDO, SDRAM, RIMM, DDR, are employed, at least for a mation presentation algorithm. Likewise, if an information Volatile portion of the memory, allowing simple upgrades and alert retrospectively turns out to be erroneous is a predictable industry standard compatibility. manner, i.e., relating to a route not taken, the system may While the preferred embodiment includes a radio fre internally adjust the algorithm without user input. quency transceiver for transmitting event data and receiving 55 In order to sort the priorities, the intended recipient may, event data, embodiments are also possible which either trans for example, identify itself 201 and communicate its location mit or receive the relevant data, but not both. For example, 202 an itinerary or intended or prospective path 205. High regulations may limit certain transmissions or relevant event priority messages 204 and various codes 203 may be inter sensors, e.g., radar detectors in trucks. In these cases, a spersed through the communication string. The transmitting receive-only embodiment may be appropriate. Further, while 60 unit then outputs data 206 in order of the computed or pre radio frequency communications are preferred, due to their dicted importance of the event and the time before the recipi range, data capacity and availability, optical communications ent encounters the event. Static events, such as fixed location systems 13, e.g., infrared LEDs and laser diodes, acoustic radar emission sources, which may, for example, indicate a communication 15, passive backscatter communications source for interference with a radar detector, or a speed detec (employing an RF transceiver Such as the spread spectrum 65 tion/control device, may be transmitted as well. In the case transceiver 12), and the like may also be employed in con where there is contention for the band, the communications junction or in Substitution of a radio frequency system. Opti session is limited by the scope of the authorization for use of US 8,874,477 B2 99 100 the band. Where there is no contention, the duration of the arbitration channel, which should be available to all compat communications channel will generally control the amount of ible units, may be adaptive or changing, for example follow information communicated. ing a formula based on time, location, and/or an arbitrary Therefore, it is noted that the present invention provides a authorization code provided with a subscription update. Fur means for mapping events and for analyzing their signifi ther, the chip sequence may vary based on selective availabil cance. Thus, this embodiment does not merely rely on pro ity (SA) deviancies in GPS data, or based on the identity of cessed sensor outputs to Supply information to the user; satellites in view of the receiver. While such information rather, sensor outputs may be filtered based on past experi might be available to "pirates', miscreants, hackers and ence with the particular location in question. If a particular scofflaws, the algorithm for generating the chip sequence user does not have direct experience with a location, then the 10 might be held as confidential, and thus the system unusable experience of others at that location may be substituted or without specific authorization and incompatible with equip combined to improve analysis of the sensor signal. Therefore, ment without such algorithm. Such systems employing the signal analysis from the sensor need not be subjected to a secure encryption with open access have been employed in relatively high threshold to avoid false alarms. A low thresh satellite television (General Instrument VideoCipher II) and old is acceptable because other information is employed to 15 the like. It is noted that, in order to maska message in a spread determine the nature of the physical elements that give rise to spectrum signal, multiple active channels may be employed, the event and sensor activation. one or more of which transmits the desired data and the It is noted that, in the case of “false alarms”, the response of remainder transmitting noise or masking data. the unit is to detect the event, e.g., radar signal, correlate it Employing 2.4 or 5.8-9 GHz communications bands, data with a stored “false alarm” event, and Suppress an alarm or rates of 54-108 megabits per second (MBPS) are possible, modify the alarm signal. Thus, information stored in memory although lower rates, such as 0.5-1.0 MBPS may be preferred and/or transmitted between units, may signify an important to gain range, reduce loss due to interference or adverse alarm or a Suppression of an erroneous alarm. In this context communications conditions and maintain availability of is apparent that the integrity of the database structure, espe simultaneous communications on multiple channels within cially from corruption by the very sources of alarms which are 25 the band in a small geographic area. intended to be detected, is important. To the extent that the Where mobile devices are traveling parallel and at similar environment responds to the existence and deployment of the speeds, or both are stopped, an extended communications system according to the present invention, for example by session may be initiated. In this case, the data prioritization detecting transmissions between units to identify and locate will be weighted to completely exchange a public portion of units, and thereby alter the nature of an event to be detected, 30 the database, although emphasis will still be placed on imme the present system may also be adaptive, in terms of its diately forthcoming events, if anticipated. On the other hand, function and signature spectral patterns. In one aspect, the where computed or user-input trajectories indicate a likely system may flash memory or flash memory modules, which brief encounter, the immediate past events are weighted most controls system operation. Therefore, periodically, the sys heavily. tem operation may be altered. The communications may 35 In order to analyze temporal or spatial relevance, the selectively occur on a plurality of bands, using a plurality of memory 4 preferably stores an event identifier 301, a location protocols. Thus, for example, the system may have tri-band 302, a time of detection of an event 303, a source of the event capability, e.g., 900 MHz, 2.4 GHz and 5.8-5.9 GHz. The information 304, an encoding for a likely expiration of the mapping feature of the present invention may also be used to event 305, a reliability indicator for the event 306, and pos identify the locations of Such monitoring sites. The system 40 sibly a message associated with the event 307 including other may also maskits transmissions as other, more common types information. These data fields may each be transmitted or of transmissions or environmental sources of emissions. A received to describe the event, or selectively transmitted direct sequence spread spectrum technique maybe employed based on the nature of the event or an initial exchange that is difficult to detect without knowing the spread spectrum between units specifying the information which will be com sequence seed. Of course, an aspect of the present invention is 45 municated. Other types of relevance may also be accounted open communications, which as a matter of course are not for, as discussed above. securely encrypted and which would identify the transponder For example, in a radar detector embodiment, mobile and its location. This problem may be addressed, in part, police radar “traps' are often relocated, so that a particular relying on laws which prevent unauthorized eavesdropping location of one event should not be perpetuated beyond its and unauthorized interception and decryption of communi 50 anticipated or actual relevance. In this case, expirations may cations, unauthorized "copying of copyright works and be stored, or calculated based on a “type' of event according defeating of copy protection schemes thereof, control over to a set of rules. False alarms, due to security systems, traffic availability of authorized transceivers, and patent protection control and monitoring systems, and the like, may also be of the design and implementation. According to one embodi recorded, to increase the reliability any warnings provided. ment of the invention, channel use and control is established 55 Likewise, traffic jams often resolve after minutes or hours, over a first channel, which may be out of band with respect to and, while certain road regions may be prone to traffic jams, the normal communications channel. Preferably, this control especially at certain hours of the day and/or days of the week, channel is longer-range and more robust than the data com abnormal condition information should not persist indefi munications channel, permitting control to precede normal nitely. communications capability, and providing enhanced recover 60 The preferred embodiment according to the present inven and network reconfiguration capabilities. tion provides an event detector, which, in turn is preferably a According to one embodiment, all communications are police radar 18 and LIDAR 19 detector. Other detected events direct sequence spread spectrum over a wide band, with may include speed of vehicle, traffic conditions, weather con medium to high security codes, e.g., 10 bits or greater length ditions, road conditions, road debris or potholes, site desig chip sequence and 12 bits or greater data encryption, and 65 nation, Sources of radio signals or interference or false alarms more preferably 16 bit or greater chip sequence and 16 bit or for other event detectors, and particular vehicles, such as greater data encryption. The chip sequence of the control and drunk drivers or unmarked police cars (possibly by manual US 8,874,477 B2 101 102 event input). The event detector may include, for example, a sequence may be based on an encrypted code. Ultrawideband sensor, such as a camera 26, which may analyze traffic control (UWB) communications techniques may also be employed. indicia (such as speed limits, cautions, traffic lights). The In order to provide flexibility in financing the communica event may also include a commercial message or advertise tions devices, the commercial messages 307 discussed above ment, received, for example from a fixed antenna beside a may be employed. Further, by circulating authorization road, which, for example, is stored as the message 307. Such tokens or codes 203, a subscription service may be provided. a commercial message 307 may be presented immediately or Thus, in a simplest Subscription scheme, the communications stored for future output. The received message, whether com device has a timer function, which may be a simple clock or mercial or not, may be a static or motion graphic image, text GPS referenced. The user must input an authorization code or Sound message. The user output of the system 27 may thus 10 periodically in order for the device to continue operating. be visual. Such as a graphic or alphanumeric (text) display, Thus, similarly to satellite television receivers and some indicator lights or LEDs 28, audible alerts or spoken voice addressable cable television decoders, failure to provide the through an audio transducer 29. authorization code, which may be entered, e.g., by telephone The camera is, for example, a color or infrared charge communication or through a keypad 30, renders the device coupled device (CCD) or complementary metal oxide silicon 15 temporarily or permanently inoperative. In order to reduce field effect transistor (CMOS) imager, having resolution of the burden of reauthorizations, the authorization codes or 0.3 to 13 megapixels. Image communication may be, for tokens may be passed through the communications "cloud' example H.261 or H.263+, using H.323 or H.324 protocol, or 24, so that devices 1, if used, will eventually receive the MPEG-4. The imager may also be incorporated as part of a authorization data. Conversely, a code 203 may be circulated mobile videoconferencing system, although a dual imager which specifically deactivates a certain device 1, for example system (one for imaging persons and the other for imaging for non-payment of the subscription fee or misuse of the road conditions) may be implemented. Other ITU standards, device (e.g., in an attempt to corrupt other users databases). e.g., T. 120, may be employed for data communications, The authorization process is preferably integral to the core although the particular nature of the data communications operation of the system, making bypassing authorization dif channel(s) may compel other communications protocols. 25 ficult. In order to maintain the integrity of the database stored in Where a number of communications devices are in prox memory 4, 20, it may be useful to store the originator of a imity, a multi-party communication session may be initiated. record, i.e., its source 304. Thus, if event information from For example, the communications Subsystem may have that origin is deemed unreliable, all records from that source simultaneous multi-channel capability, allowing each unit to may be purged, and future messages ignored or "flagged'. As 30 transmit on a separate channel or use a shared channel. Where stated above, even the proximity of an unreliable or modified the number of channels or channel capacity is insufficient, unit may be detrimental to system operation. Therefore, units may take turns transmitting event information on the where the location of Such a unit is known, other units in same channel (e.g., according to estimated priority), or time proximity may enter into a silent mode. Further, normal units division multiplex (TDM) the channel(s). Preferably, the may transmit a "kill message to the unreliable unit, causing 35 communication scheme involves a number of channels within it to cease functioning (at least in a transmit mode) until the a band, e.g., 1 common control channel and 24 data commu problem is rectified or the unit reauthorized. nications channels. Since Some communication sessions may The unit is preferably tamper-proof, for example, codes be relatively short, e.g., limited to a few seconds, a data necessary for unit activation and operation are corrupted or communications channel preferably has a maximum capacity erased if an enclosure to the unit is opened. Thus, techniques 40 of tens of kilobits per second or higher. In some cases, hun such as employed in the General Instrument VideoCipher II dreds of kilobits, or megabit range bandwidths are achiev and disclosed in Kaish et al., U.S. Pat. No. 4,494,114, may be able, especially with a small number of channels (e.g., one employed. channel). For example, so-called third generation (3G) cellu The communications Subsystem preferably employs an lar communications protocols may be employed. error correction/error detection protocol, with forward error 45 Thus, for example, an 802.11g or 802.11n compatible correction and confirmation of received data packet. The spread spectrum transceiver operating in the 2.4 GHz band scheme may be adaptive to the quality of the communication might have a usable bandwidth of 10-108 megabits per sec channel(s), with the packet length, encoding scheme, trans ond, even while sharing the same band with other transceivers mit power, bandwidth allocation, data rate and modulation in close proximity. Where necessary, directional antennas, or scheme varied in an adaptive scheme to optimize the commu 50 phased arrays may be employed to provide spatial discrimi nication between units. In many cases, units engaged in com nation. munication will exchange information bidirectionally. In that The system preferably has advanced ability to detect chan case, a full duplex communication protocol is preferred; on nel conditions. Thus, where communications are interrupted the other hand, where communication is unidirectional, by physical limitations in the channel, the impairment to the greater data communication rates may be achieved employ 55 communications channel is detected and the communications ing the available bandwidth and applying it to the single session paused until the impairment abates. This, in turn, will communication session. allow other units, which might not be subject to the impair In some instances, it may be desired to maintain privacy of ment, to use the same channel during this interval. The chan communications. In that case, two possibilities are available; nel impairment may be detected by a feedback protocol spread spectrum communications, preferably direct sequence 60 between communications partners, or by means of symmetric spread spectrum communications is employed, to limiteaves antennas and communications systems, by which an impair dropping possibilities. Second, the data itself may be ment of a received signal may be presumed to affect the encrypted, using, for example, a DES, PGP. elliptic keys, or transmitted signal as well. The latter requires a high degree of RSA type encryption scheme. Keys may be supplied or standardization of equipment design and installation for exchanged in advance, negotiated between partners, or 65 effectiveness. involve a public key-private key encryption algorithm. For It is particularly noted that, where the events to be detected example, the spread spectrum communications chip and the communications Subsystem operate in the same band, US 8,874,477 B2 103 104 structures may be shared between the communications and change of location of the GPS antenna, (iv) heading of the event detection systems, but this also increases the possibili GPS antenna, and (v) time to a GPS interface. Optionally, the ties for interference. GPS Smart Antenna and GPS processor are differential As one embodiment of the invention, the processor may be ready. An optional input select switch, controlled by the GPS provided as a standard personal digital assistant (PDA) with a processor upon a request from the PDA, allows a single serial PC Card or PCMCIA slot for receiving a standard GPS interface to receive either a control signal from the PDA or a receiver and another standard PC Card slot for receiving an DGPS error correction signal from an optional DGPS radio 802.11b/g/a/ R/A module. The PDA, in turn has memory, wave receiver. Alternately, a DGPS-type system may be coor which may include random access memory, flash memory, dinated between multiple mobile receivers, top provide high and rotating magnetic memory (hard disk), for example. The 10 PDA has a processing system which is capable of running relative position accuracy, even where the absolute position applications written in general purpose, high level languages accuracy is low. Since the event position calculations are Such as C. The PDA may operate under a standard operating based on the relative position frame, the effect is to accurately system, such as Microsoft WindowsCE, XP, Palm OS, Linux, position the events with respect to the vehicle. or a proprietary operating system. A Software application 15 The user device may display, for example, map features written in a high level language can normally be ported to run according to a coordinate system Such as latitude and longi in the PDA processing system. Thus, the basic elements of the tude. The display may also include an indication of the loca hardware platform are all available without customization. In tion of the GPS receiver, an itinerary, proposed route, and a preferred embodiment, an event sensor is provided, such as indications of the location of various events. By correlating a police radar and laser speed detection equipment system the GPS with a stored map, the absolute location of the (e.g., “radar detector”) is provided. This may employ a modi vehicle may be determined by map matching techniques. In fied commercially available radar detector, to produce a serial accordance with the present invention, these events are data stream or parallel signal set. For example, radar detectors derived from the event detector or the memory. Other com providing an alphanumeric display often transmit data to the munications devices may also be located on the display. display controllerby means of a serial data signal. This signal 25 The user entry system has both touchscreen keys and press may be intercepted and interfaced with a serial port or custom keys in the present embodiment. With a touchscreen, a user port of the PDA. enters a request by touching a designated portion overlying a Optionally, the GPS Smart Antenna is “differential-ready” visual display with his finger (or Soft pointer, Such as a plastic to apply differential GPS (DGPS) error correction informa pen). The touchscreen senses the touch and causes a digital tion to improve accuracy of a GPS determined location. The 30 signal to be sent to the processing system indicating where the application program for the PDA may be provided in a semi touch was made. Switches Such as rotary Switches, toggle conductor memory cartridge or stored on hard disk. switches, or other switches can equally well be applied. An The PDA 30 includes the processing system, including a advantage of the touchscreen is that a label or a placement of microprocessor, memory, precoded program instructions and the touchscreen, and a corresponding function of the touch data stored in memory, a microprocessor bus for addresses, 35 screen, may be changed by the computer controlling the data, and control, an interrupt bus for interrupt signals, and display any number of times without changing electrical or associated hardware, operates in a conventional manner to mechanical hardware. In the present embodiment, Zoom keys receive digital signals, process information, and issue digital may be employed change scale and resolution of a map on the signals. A user interface in the PDA includes a visual display display. Zooming in decreases the scale, so that the map is or audible output to present signals received from the pro 40 viewed with greater resolution over a lesser area of the map. cessing system to a user, a user entry system to issue signals Zooming out increases the scale, so that a greater area of the from the user to the processing system. The user interface map is viewed with lesser resolution. A map orientation key may include one or more push keys, toggle Switches, proX selects an orientation of a direction on the map with a direc imity Switches, trackballs, joysticks or pressure sensitive tion on the visual display, for example, orientations of north keys, a touch-sensitive display screen, microphones oracom 45 up or current ground track up. It is noted that these map bination of any of the above used together or with other functions are generally known, and known techniques may be similar type user input methods. The PDA sends digital sig generally applied for Such map functions. According to the nals representing addresses, data, and commands to the present invention, in addition to normal map functions, the memory device and receives digital signals representing event data may be overlayed on the map to provide additional instructions and data from the memory. A PDA interface 50 dimensions of display data. Further, by providing these data, electrically connects the processing system to a GPS Smart which are dynamic, the map system becomes useful even to Antenna. If the PDA and GPS are not integrated, a preferred travelers who are well aware of the geography and layout of interface comprises a computer standard low to medium the region being traveled. speed serial data interface, such as RS-232, RS-422, or USB One communications scheme, a 900 MHz spread spectrum (1.0, 1.1, 2.0), IEEE-1394, Bluetooth (especially if the com 55 communications system, operates as follows. The RF munications system operates in another band), through a receiver demodulates the signal using known techniques, cabled interface for connection to the GPS Smart Antenna. which may include Superheterodyne and Zero-IF techniques. The GPS Smart Antenna system includes a GPS receiver Thus, in a frequency hopping embodiment, the demodulator antenna to receive GPS satellite signals from GPS satellite tracks the hop frequency sequence. In a direct sequence transmitters, a GPS frequency downconverter to downconvert 60 spread spectrum embodiment, the demodulator provides the the approximately 1.575 GHz frequency of the L1 GPS sat appropriate pseudorandom code sequence to demodulate the ellite signals to a lower frequency (LF) signal that is suitable received signal. Time synchronization may be effected by for digital processing, and to issue the LF to a GPS processor. using the timing functions of the GPS receiver. The demodu The GPS processor demodulates and decodes the LF signal lated signal is then decoded into messages, which are typi and provides location information for at least one of (i) loca 65 cally digital bitstreams. The receiver may be process a plu tion of the GPS antenna, (ii), GPS satellite pseudoranges rality of signals simultaneously, allowing multiple signals to between the GPS satellites and the GPS antenna, (iii) rate of be received simultaneously. US 8,874,477 B2 105 106 Seventh Embodiment within a time scale which is short with respect to the commu nications, thus requiring consideration of dynamic changes in Ad hoc networks are a good candidate for analysis and network architecture. optimization according to game theory. A multihop ad hoc Ad hoc networks pose control issues with respect to con network requires a communication to be passed through a tention, routing and information conveyance. There are typi disinterested node. The disinterested node incurs a cost, thus cally tradeoffs involving equipment size, cost and complex leading to a disincentive to cooperate. Meanwhile, bystander ity, protocol complexity, throughput efficiency, energy nodes must defer their own communications. By understand consumption, and “fairness” of access arbitration. Other fac ing the decision analysis of the various nodes in a network, it tors may also come into play. is possible to define a system which, in accordance with game 10 theory, provides a benefit or incentive to promote cooperation Game theory studies the interactions of multiple indepen and network reliability and stability. The incentive, in eco dent decision makers, each seeking to fulfill their own objec nomic form, may be charged to the node(s) benefiting from tives. Game theory encompasses, for example, auction theory the communication, and is preferably based on a value of the and strategic decision-making. By providing appropriate benefit received. This network optimization employs a modi 15 incentives, a group of independent actors may be persuaded, fied combinatorial (VCG) auction, which optimally compen according to self-interest, to act toward the benefit of the sates those burdened by the communication, while charging group. That is, the selfish individual interests are aligned with the benefiting participants. Equilibrium usage and headroom the community interests. In this way, the community will be may be influenced by deviating from a Zero-sum condition. both efficient and the networkofactors stable and predictable. The mechanism seeks to define fairness in terms of market Of course, any system wherein the “incentives' impose too value, providing probable participation benefit for all nodes, high a cost, themselves encourage circumvention. In this leading to network Stability. case, game theory also addresses this issue. The present example describes the application of game Cooperative Problems in Ad Hoc Networks theory concepts to the arbitration of access to bandwidth in an To understand why game theory is applicable the control ad hoc communications network, more particularly to net 25 over ad hoc networks, consider the analogy of a classroom. work including mobile nodes. According to applicable ele The teacher acts as a central authority and arbitrator to ensure ments of game theory, an agent makes a decision to cooperate decorum. The teacher recognizes one student at a time for with a system having established rules, or to circumvent it. public communication. This is an example of centralized Likewise, cheating, i.e., adopting behavior contrary to an control. If there were no teacher to recognize a student, pan expected nor, may be an option, and can be analyzed in the 30 demonium would result, unless a suitable process of self context of a decision. Therefore, a game theoretic approach organization is established, which obtains cooperation dictat addresses the situation where the operation of an agent which ing common rules, adopted according to mutual incentives. has freedom of choice, allowing optimization on a high level. Now, Suppose one student wishes to send a note across the considering the possibility of alternatives to a well designed room. Presumably, there are multiple paths to the destination. system. According to game theory, the best way to ensure that 35 But how can the student be sure that the note will be for a system retains compliant agents, is to provide the greatest warded? How does one know which neighbor to handoff to? anticipated benefit, at the least anticipated cost, compared to Suppose that forwarding the note imposes a burden, Such as the alternates. the risk of being caught and sanctioned? Consider the possi Mobile ad hoc networks encompass multihop networks, bility, after conclusion of negotiations for forwarding, a stu which, by their nature, require participation of disinterested 40 dent fails to fulfill his assumed responsibility? nodes to operate. Technically, however, the multihop scenario It is therefore clear that the issues of subjective and objec is not intrinsic, since it is reasonable to presume that in some tive costs and benefits, distance, complexity, and reliability, networks, all nodes are within range of each other. Each are therefore interrelated, and there may be practical scenario poses a classic game theory issue to each node: why restraints on achieving theoretical system capacity. defer to other nodes if no direct benefit is obtained? The 45 The game theoretic solution is to link an incentive or ben multihop network adds the further issue of: why participate in efit to the desired behavior, to promote cooperation of each communications between other nodes if no direct benefit is agent with note forwarding, on a rational basis. The ultimate obtained? We discuss a set of mechanisms, incentives and payoffshould be borne by the student receiving the benefit. rationales as a framework for analyzing node behavior and Thus, by linking benefits to costs, a stable Society is achieved, optimization, and seeks to respond to these issues by propos 50 approaching a desirable equilibrium. ing appropriate incentives to promote network efficiency and In order to incentivize the intermediaries, a student could stability. compensate them by taping dimes to the note, instructing The application of game theory to ad hoc networks has each forwarding student to remove one dime (the packet been addressed in various forms to date. In general, there is a purse model). Alternately, the recipient may be expected to divergence between approaches which define a real-world 55 pay for the transmission through an acknowledgement mes system, with all of its complexity, and required functionality, sage with attached value (the packet trade model). However, and those which seek to mathematically tractable model hav how do we know that the first recipient will not remove all the ing a definite set of rules and presumptions leading to a money and throw away the note? How can the intermediaries comprehensible and useful result. ensure, in the packet trade model, that the recipient will An ad hoc network is a wireless network which does not 60 eventually pay? How does the responsible party know how require fixed infrastructure or centralized control. The termi much must be paid? These models also require stability of the nals in the network cooperate and communicate with each route during the process, and imply a priori knowledge of the other, in a self organizing network. In a multihop network, route. This approach does not permit variations in compen communications can extend beyond the scope of a single sation, e.g., some students might accept a nickel, and others in node, employing neighboring nodes within the scope, to for 65 a critical position, might require a quarter. In cases of unreli ward messages. In a mobile ad hoc network, constraints are able routes, should the originator send two notes by alternate not placed on the mobility of nodes, that is, they can relocate paths, or attempt to pay more for a single reliable delivery? US 8,874,477 B2 107 108 Even with imposition of a traffic sensitive cost, one node of life events such as a sealed bid auction. A progressive bid the network may seek to send an inordinate number of mes auction may be modeled as an extensive game. sages, resulting in congestion. A node in a critical location Games can also be divided according to their payoffstruc may become wealthy, and its fees rise, leading to instability. tures. A game is called Zero-sum game if the sum of the Likewise, in a virtual construct, what does one use as cur utilities is constant in every outcome. Zero-sum games are rency? We see that consideration must be given to keeping considered strictly competitive games. For example, an auc traffic below capacity, since highly loaded networks often tion is a Zero Sum game, since the buyer pays the seller, with have decreased efficiency. no other gains or losses incurred. If the players are fully Game Theory informed about each other's moves, the game has perfect 10 information. Only extensive games consider the issue of per Game theory is the study of the interaction of independent fect information. In games with complete information the agents, in an environment where there are rules, decisions, utility function of each player is known. In a game with and outcomes. Game theory defines the theoretical basis for incomplete information, the privacy of a player's utility func strategy, as well as providing a framework for analyzing tion is held as a strategic advantage. real-world actors. Game theory may be applied to automated 15 Pareto efficiency exists if there is no other outcome that systems, providing a basis for the analysis and optimization would make all players better off. An equilibrium is a result of of such systems. Aspects of game theory have been applied to the optimization of the individual players, but does not imply telecommunications, for example to optimize network rout that the result is “good” or globally optimum. The solution of ing, and has quality of service implications. Communications a strategic game is a Nash equilibrium. Every strategic game resources may be treated as utilities, and auctions have been with finite number of players, each with a finite set of actions, applied to the optimization of allocation of utility resources. has an equilibrium point. This Nash equilibrium is a point Each game has a set of rules or constraints, under which the from which no single player wants to deviate unilaterally. agents operate. “Cheating, if permitted at all, is modeled as When a game is played, the rationality assumption will force an available decision of an agent to comply with other con the game into a Nash equilibrium outcome. If the outcome is straints. Therefore, the game is valid as a model only for the 25 not a Nash equilibrium, at least one player would gain a rules and constraints considered. Each decision maker pur higher payoff by choosing another action. If there are mul Sues a goal, and may take into account their knowledge or tiple equilibriums, more information on the behavior of the expectations of the other decision makers' behavior. Accord players is needed to determine the outcome of the game. ing to game theory, rationality leads to optimality, and there In the strategic and extensive games, the Solution of a game fore analyzing the game and acting logically in accordance 30 is a complete set of strategies that achieve a Nash equilibrium. with the rules leads to the best outcome. In cooperative games, the Solution comprises the Subsets of It is conceptually simple for an automated system to act players or coalitions from which no member has an incentive rationally. That is, given a set of facts and circumstances, the to break away. Cooperative games can be divided between rational analysis is fixed and obtainable. On the other hand, games in which the coalition is free to internally distribute a humans acting on purely mental consideration may deviate 35 payoff (transferable payoff), and those in which the payoff is from rationality. For example, humans exhibit a broad range personal to coalition members (non-transferable payoff). A of risk tolerance, which is not directly explained by rational dominant strategy is one in which the same decision is made analysis. It is noted that risk tolerance, and other aspects of based on the various different rational strategies an agent may behavior, have been modeled, and as such, can themselves be adopt. treated Scientifically and rationally. In fact, game theory 40 To better understand game theory, it is useful to consider expressly recognizes that agents may express private values simple games. In one game, called the prisoner's dilemma, which are not rationally explained, and that by understanding two criminals are arrested and charged with a crime. The these values, a strategic advantage is obtained. Thus, while police do not have enough evidence to convict the Suspects, rationality is assumed as an optimum strategy for each entity, unless at least one confesses. They are notable to communi real entities have imperfect estimates of payoff and risk, and 45 cate. If neither confesses, they will be convicted of a minor indeed may miscalculate or misconstrue the circumstances. crime and sentenced for one month. If one confesses, and the Such perturbations may be compensated, under certain cir other does not, the confessing one will be given immunity and cumstances, by way of various parameters added to the mod released and the other will be sentenced for nine months. If eling equation. both confess, both will be sentenced for six months. Game theory is typically encompassed in the study of 50 Another famous game is the battle of the sexes. A couple is economics, since a self-interested node will always try to going to spend an evening out. She wishes to attend the opera increase its wealth, and all other concepts may be considered and he wishes to attend a hockey match, but each gains a in terms of their subjective economic costs and benefits. benefit of the other's company. Game theory can be used not only to analyze a defined game, In the prisoner's dilemma, all the outcomes except (Con but also to define a game having a desired outcome, i.e., to 55 fess; Confess) are Pareto efficient. In the battle of the sexes, an optimize a set of rules and constraints. The preferences of a outcome in which husband and wife attend different events node can be expressed either with a utility function, or with are not Pareto efficient. The outcome (Confess: Confess) is preference relations, ranking various consequences. the equilibrium, while outcome (Don't confess; Don't con Games can be divided into noncooperative and cooperative fess) results in higher payoff for both the criminals, but it is games. In cooperative games, the joint actions of groups are 60 not an equilibrium because both the players have an incentive analyzed, i.e. what is the outcome if a group of players coop to deviate from it. In the battle of the sexes, the pure strategy erate. In noncooperative games, the actions of the single equilibrium points are (Opera; Opera) and (Hockey: players are considered. The cooperative game model may be Hockey). There is also a third Nash equilibrium with mixed used to analyze heterogeneous ad hoc networks. In strategic strategies, in which both choose their preferred option with games, decisions are made at the commencement of the game. 65 probability 2:3. The prisoner's dilemma is a good example of In extensive games, decisions may be made interactively. The a sub-optimal equilibrium. Both players would gain a higher strategic game model is Suitable for representing simple real payoff by playing (Don't confess; Don't confess). US 8,874,477 B2 109 110 Another example of game theory is the so-called tragedy of of the private value of other bidders, and thus may change the commons. In this game, a set of farmers live in a commu over the course of the auction in a successive price auction. nity with a grass-filled square. Each farmer is confronted with On the other hand, in certain situations, release or publication a decision as to whether to acquire another goat, which eats of the private value is a dominant strategy, and can result in grass in the square. So long as the benefit of having the goat is Substantial efficiency, that is, honesty in reporting the private in excess of the personal detriment of that goat’s grass con value results in the maximum likelihood of prospective gain. Sumption, it is a dominant strategy to acquire the goat, even A so-called Vickrey–Clarke-Groves, or VCG, auction, is a though the necessary result of all farmers acting rationally is type of auction Suitable forbidding, in a single auction, for the the loss, to all, of the benefits of the square. goods or services of a plurality of offerors, as a unit. In the In computer networks, issues arise as the demand for com 10 classic case, each bidder bids a value vector for each available munications bandwidth approaches the theoretical limit. combination of goods or services. The various components Under such circumstances, the behavior of nodes will affect and associated ask price are evaluated combinatorially to how close to the theoretical limit the system comes, and also achieve the minimum Sum to meet the requirement. The win which communications are permitted. The well known colli ning bid set is that which produces the maximum value of the sion sense, multiple access (CSMA) protocol allows each 15 accepted bids, although the second (Vickrey) price is paid. In node to request access to the network, essentially without cost the present context, each offeror Submits an ask price (re or penalty, and regardless of the importance of the commu serve) or evaluatable value function for a component of the nication. While the protocol incurs relatively low overhead combination. If the minimum aggregate to meet the bid and may provide fully decentralized control, under congested requirement is not met, the auction fails. If the auction is network conditions, the system may exhibit instability, that is, successful, then the set of offerors selected is that with the a decline in throughput as demand increases, resulting in ever lowest aggregate bid, and they are compensated that amount. increasing demand on the system resources and decreasing The Surplus, i.e., gap between bid and ask, is then available throughput. According to game theory, the deficit of the to compensate the deferred bidders. This surplus is distrib CSMA protocol is that it is a dominant strategy to be selfish uted proportionately to the original bid value for the bidder, and hog resources, regardless of the cost to Society, resulting 25 thus further encouraging an honest valuation of control over in “the tragedy of the commons.” the resource. Game theory is most readily applied in the optimization of The network is such that, if any offeror asks an amount that communications routes through a defined network, to achieve is too high, it will be bypassed. Since the bidder pays the the best surplus allocation. The problems of determining the second highest price, honesty in bidding the full private value network topology, and conducting the communications them 30 is encouraged. VCG auctions have found application in pro selves, are also applications of game theory. Since the com viding segment links to route goods, or information in a munications incidental to the network access arbitration network. In defining the goods and services that are the sub require consideration of some of the same issues as the under ject of the auction, it is possible to value the non-interference lying communications, elements of game theory apply corre of a competitor, that is, a competitor is both a buyer and seller spondingly. Due to various uncertainties, the operation of the 35 in the sale multi-good auction, with the purchase and sale system is stochastic. This presumption, in turn, allows esti being inconsistent combinations. mation of optimality within an acceptable margin of error, The traditional VCG auction, is postulated as being opti permitting simplifying assumptions and facilitating imple mal for allocation of multiple resources between agents. It is mentation. “strategyproof and efficient, meaning that it is a dominant In an ad hoc network used for conveying real-time infor 40 strategy for agents to report their true valuation for a resource, mation, as might be the case in a telematics system, there are and the result of the optimization is a network which maxi potentially unlimited data communication requirements, and mizes the value of the system to the agents. Game theory also network congestion is almost guaranteed. Therefore, using a allows an allocation of cost between various recipients of a CSMA protocol as the paradigm for basic information con broadcast or multicast. That is, the communication is of value veyance is destined for failure, unless there is a disincentive to 45 to a plurality of nodes, and a large set of recipient nodes may network use. On the other hand, a system which provides efficiently receive the same information. This allocation from more graceful degradation under high load, sensitivity to the multiple bidders to multiple sellers is a direct extension of importance of information to be communicated, and efficient VCG theory, and a similar algorithm may be used to optimize utilization of the communications medium would appear allocation of costs and benefit. more optimal. 50 Ad Hoc Networks One way to impose a cost which varies in dependence on In an ad hoc network, there is no central authority control the Societal value of the good or service, is to conduct an ling network operation, and there is typically a presumption auction, which is a mechanism to determine the market value that some nodes cannot directly communicate with others, of the good or service, at least between the auction partici leading to a requirement for communication intermediaries to pants. In an auction, the bidder seeks to bid the lowest value, 55 forward packets, i.e., a multihop architecture. A mobile ad up to a value less than or equal to his own private value (the hoc network adds the further presumption that nodes are not actual value which the bidder appraises the good or service, stationary, and therefore a route discovery mechanism is and above which there is no surplus), that will win the auction. required. Since competitive bidders can minimize the gains of another In order to determine the network architecture state, each bidder by exploiting knowledge of the private value attached 60 node must broadcast its existence, and, for example, a pay to the good or service by the bidder, it is generally a dominant load of information including its identity, location, itinerary strategy for the bidder to attempt to keep its private value a (navigation vector) and possibly an “information value func secret, at least until the auction is concluded, thus yielding tion' and/or “information availability function”. Typically, strategies that result in the largest potential gain. Auction the system operates in a continuous state, so that, after stabi strategies become more complex when the bidder himself is 65 lization, it is reasonable to estimate of the present state based not a consumer or collector, but rather a reseller. In this case, on the prior state information. In a system with mobile nodes, the private value of the bidder is influenced by the perception the mobility may be predicted, or updates provided as neces US 8,874,477 B2 111 112 sary. Using an in-band or out-of-band propagation mecha mized Link State Routing Protocol (OLSR), Dynamic Source nism, this information must propagate through a sphere of Routing Protocol (DSR), and Topology Dissemination Based influence or to a network edge, which may be physically or on Reverse-Path Forwarding (TBRPF). In most scenarios artificially defined. Nodes may be presumed to operate with a analyzed to date, the performance metrics studied were power Substantially common estimation of network topology, and 5 consumption, end-to-end data throughput and delay, route therefore only deviations from previously propagated infor acquisition time, percentage out-of-order delivery, and effi mation need be propagated. Of course, a mechanism should ciency. A critical variable considered in many prior studies is be provided for initialization and in case a new node joins the power cost, presuming a battery operated transceiver with network. If Such estimates were accurate, the network could finite available power. There can be significant differences in then be modeled similarly to a non-mobile network, with 10 certain extensions. On the other hand, typical implementa optimum routing depending on whether node has a transmit tions will present substantial deviations between actual net power control, which in turn controls range, and provides a work architecture and predicted network architecture, requir further control over network topology. Likewise, steerable ing Substantial fault tolerance in the fundamental operation of antennas, antenna arrays, and other forms of multiplexing the protocol and system. 15 provide further degrees of control over network topology. If we presume that there is a spatial or temporal limit to Note that the protocol-level communications are preferably relevance, for example, 5 miles or 10 hops, or 1 to 5 minutes, broadcasts, while information conveyance communications then the network State propagation may be so limited. Extend are typically point-to-point. Prior analyses typically presume ing the network to encompass a large number of nodes, will a single transceiver, with a single antenna, and thus use an necessarily reduce the tractability of the optimization, omni-directional antenna, with in-band protocol data, for all although this may also produce Substantial benefits, espe communications. The tradeoff made in limiting system cially if the hop distance is relatively short with respect to the designs according to these presumptions should be clear. desired communication range. Each node may therefore Routing protocols in ad hoc networks typically employ impose a local estimate of relevance as a filter on communi three strategies: flooding, proactive routing, and reactive rout cations, especially arbitration communications. This consid 25 ing. Flooding protocols broadcast packets to all the nodes in eration is accommodated by communicating, from each node, the network, while the remaining protocols do not. In proac an update to all other nodes within its network relevance tive routing, the protocol maintains route information all the boundary, and a state variable which represents an estimate of time, while in reactive routing, a route is discovered only relevant status beyond the arbitrarily defined boundary. The when needed. All or Some of these strategies may be boundary estimate is advantageous in order to ensure long 30 employed simultaneously. Flooding typically consumes too range consistency. On a practical note, assuming a cost is much bandwidth and energy to be efficient, as compared to incurred by employing the ad hoc network, which scales with more sophisticated strategies. However, in cases with very the number of hops, then at Some point, especially consider high mobility, the flooding protocols best ensure that the ing the latency and reliability issues of ad hoc networks with transmission reaches its destination. a large number of hops, it is more efficient to employ cellular 35 In proactive routing, each node stores and updates routing communications or the like. On the other hand, making thead information constantly. The routing tables can be updated hoc network suitable and reliable for 100 hop communica based on perceived changes in the network topology. There tions will necessarily impede communications over a much fore, a new transmission can start immediately without a Smaller number of hops, thus disincentivizing the more rea route discovery delay. However, the constant exchange of sonable uses of the network. 40 routing information adds overhead to the protocol. OLSR and For example, the propagation of network State and other TBRPF protocols use proactive routing. The overhead traffic protocol-level information may conveniently occur over a of a proactive routing protocol increases as the mobility of the finite number of hops, for example 5-10, in an outward direc nodes increases, since the routing information needs to be tion, a condition which may be assessed by GPS assistance. updated in shorter intervals. For each hop, a relatively simple protocol. Such as a collision 45 In reactive routing, when a node wishes to transmit, it starts sense-multiple access (CSMA) protocol, may be employed, a route discovery process in order to findapath to the receiver. with each node propagating information according to a set of The routes remain valid until the route is no longer needed. rules. (It is noted that, since this communication is not “lim AODV and DSR protocols use reactive routing. In the AODV itless' in contrast to bulk or streaming real-time sensor data, protocol, to find a route to a receiver, a terminal broadcasts a CSMA may be an appropriate and efficient protocol). 50 route request message containing the address of the receiver An example of the application of game theory to influence and the lifespan of the message. Terminals receiving the system architecture arises when communications latency is message add their address to the packet and forward it if the an issue. A significant factor in latency is the node hop count. lifespan is not exhausted. If a receiver or a terminal knowing Therefore, a system may seek to reduce node hop count by the route to the receiver receives the route request message, it using an algorithm other than a nearest neighbor algorithm, 55 sends a route reply back to the requester. If the sender does not bypassing some nodes with longer distance communications. receive a route reply before a timeout occurs, it sends another In analyzing this possibility, one must not only look at the cost route request with a longer lifespan. The use of sequential to the nodes involved in the communication, but also the cost route requests with incremental increases in timeout allows a to nodes which are prevented from simultaneously accessing mapping of the network by hop count. the network. As a general proposition, the analysis of the 60 In order for an ad hoc network to be effective, the nodes network must include the impact of each action, or network need to cooperate. This cooperation comes at a cost. In power state, on every node in the system, although simplifying pre constrained systems, for example, the cost is battery con Sumptions may be appropriate where information is unavail Sumption. In other scenarios, the network utilization is itself able, or the anticipated impact is trivial. a burden. The various nodes must cooperate in both arbitra Known and well analyzed routing models proposed for 65 tion and control, e.g., route discovery and optimization, and forwarding of packets in ad hoc networks. These include Ad the information forwarding itself. In fact, participation in the Hoc On-Demand Distance Vector (AODV) Routing, Opti route discovery, without notice that the node will fail to for US 8,874,477 B2 113 114 ward information packets, has been shown in studies to be an auction paying the losers is at first difficult to contemplate, more detrimental to the network than simply abstaining but on further analysis quite logical. This upsets the normal entirely. analysis, since the possibility of a payment from the winner to It is the general self-interest of a node to conserve its own the loser alters the allocation of economic surplus between resources, maintain an opportunity to access resources, while the bidder, seller, and others. Likewise, while the cost to the consuming whatever resource of other nodes as it desires. involved nodes may be real, the cost to the uninvolved nodes Clearly, this represents the “tragedy of the commons', in may be subjective. Clearly, it would appear that involved which selfish individuals fail to respect the very basis for the nodes should generally be better compensated than unin community they enjoy, and a network of rational nodes oper Volved nodes, although a formal analysis remains to be per ating without significant incentives to cooperate would likely 10 formed. fail. On the other hand, if donating a node's resources gener In a more general sense, the underlying presumption is that ated an associated benefit to that node, while consuming the network provides competitive access to the physical trans network resources imposed a cost, stability and reliability can port medium, and that cooperation with the protocol provides be achieved. So long as the functionality is sufficient to meet significant advantages over competition with it. Clearly, the the need, and the economic surplus is “fairly allocated, that 15 issues of commercial Success and market dominance are is, the cost incurred is less than the private value of the benefit, much more complex and not capable of being accurately and that cost is transferred as compensation to those burdened modeled according to known paradigms: on the other hand, a in an amount in excess of their incremental cost, adoption of system providing rational benefits will be more likely to the system should increase stability. In fact, even outside of succeed than one with irrational benefits or ill defined expli these bounds, the system may be more stable than one which cable benefits. Under normal circumstances, a well developed does not tax system use nor reward altruistic behavior. While ad hoc network system can present as a formidable coordi the system is a Zero Sum system, and over time, the economic nated competitor for access to contested bandwidth by other effects will average out, in any particular instance, the incen systems, while within the network, high valued communica tive for selfish behavior by a node will be diminished. tions may receive priority. Thus, a node presented with a The concepts of node misbehavior and maliciousness, and 25 communications requirement is presented not with the simple competing networks consuming the same resources, are also choice of participate or abstain, but rather whether to partici addressed by aspects of game theory. For example, an ad hoc pate in an ad hoc network with predicted stability and mutual network may defer to or compete with an interfering network, benefit, or one with the possibility of failure due to selfish and the decision of which strategy to adopt is within the behavior, and non-cooperation. Even in the absence of a province of game theory. The particularities of agent misbe 30 present communication requirement, a network which havior or hacking are also encompassed by GameTheory, and rewards cooperative behavior may be preferable to one which real implementations must necessarily consider these issues. simply expects altruism. Sometimes, the solution to these issues is technological, but After the network architecture is defined, compensation is in others, the reaction of other nodes to discovery of misbe paid to those nodes providing value or subjected to a burden havior may be sufficient to discourage it. The intent is to 35 (including foregoing communication opportunity) by those formulate a system which is sufficiently robust and advanta gaining a benefit. The payment may be a virtual currency, geous as to disincentivize non-compliance and non-coopera with no specific true value, although the virtual currency tion, that is, the inherent advantages of compliance with the system provides a convenient method to tax, Subsidize, or system architecture exceed the anticipated benefits of the control the system, and thus apply a normalized extrinsic alternative. 40 value. One way to remedy selfish behavior is to increase the cost Game theory also encompasses the concept of that each of acting this way, that is, to impose a cost or tax for access to node may have an associated “reputation' in the community. the network. In a practical implementation, however, this is This reputation may be evaluated as a parameter in an eco problematic, since under lightly loaded conditions, the nomic analysis, or applied separately. This reputation may be “value of the communications may not justify a fixed cost 45 anecdotal or statistical. In either case, if access to resources which might be reasonable under other conditions, and like and payments are made dependent on reputation, nodes will wise, under heavier loads, critical communications may still be incentivized to maintain a good reputation, and avoid be delayed or impeded. Note that where the network includes generating a bad reputation. Therefore, by maintaining and more nodes, the throughput may increase, since there are applying the reputation in a manner consistent with the com more potential routes and overall reliability may be increased, 50 munity goals, the nodes are compelled to advance those goals but the increased number of nodes will likely also increase in order to benefit from the community. Game theory distin network demand. A variable cost, dependent on relative guishes between good reputation and bad reputation. Nodes “importance', may be provided, and indeed, as alluded to may have a selfish motivation to assert that another node has above, this cost may be market based, in the manner of an a bad reputation, while it would have little selfish motivation, auction. In a multihop network, Such an auction is compli 55 absent collusion, for undeservedly asserting a good reputa cated by the requirement for a distribution of payments tion. On the other hand, a node may have a selfish motivation between the chain of nodes, with each node having potential in failing to reward behavior with a good reputation. alternate demands for its cooperation. The VCG auction The virtual currency and reputation may be considered mechanism excludes nodes which ask too high a price, and orthogonal, since the status of a node's currency account the auction itself may comprise a value function encompass 60 provides no information about the status of its reputation. ing reliability, latency, quality of service, or other non-eco Published Ad Hoc Network Examples nomic parameters, in economic terms. By no way a comprehensive list of published applications The network may further require compensation to nodes of game theory to the control of ad hoc networks, below are which must defer communications because of inconsistent discussed a number of prominent examples. states, such as in order to avoid interference or duplicative use 65 The Terminodes project includes many of the features of an intermediary node, and which take no direct part in the described above. This project proposes a method to encour communication. It is noted that the concept of the winner of age cooperation in ad hoc networks that is based on a virtual US 8,874,477 B2 115 116 currency called nuglets. Each node contains a tamper-proof Game theory has also been applied on flow control. Each hardware module which handles the nuglets. When a node user tries to maximize its utility, defined by the ratio of aver forwards a packet it extracts nuglets from the payload. In age throughput and average delay. It has been shown that a order to make a transmission, the sender appends nuglets unique Nash equilibrium exists in Such a system, which con needed to forward the packet through the network to its des Verges to an equilibrium point. The nodes seek to both maxi tination. However, a central node probably likely accumu mize their own quality of service, but also the fairness of lates excess nuglets, hence it has less value for additional resource allocation, resulting in a Pareto efficient solution. nuglets, leading to lower incentive for network activity. ALOHA is a wireless CSMA protocol using time division Peripheral nodes may possess insufficient nuglets to Support multiplexing. Transmission probabilities are a design speci their needs. However, the system appears to achieve a balance 10 fication of the system, but ifa player uses a higher probability, over time, assuming random node movement. The Termin odes project is notable for the depth and completeness of its his throughput will likely increase, leading to a misbehavior analysis, as well as the progress made toward implementa incentive. The selfish system appears to perform no better tion. than a centrally controlled (non-CSMA) system, and perfor Crowcroft et al. present a traffic-sensitive pricing model. 15 mance typically drops by half. A pricing mechanism may be Compensation for packet forwarding is responsive to both incorporated, thus taxing users for their bandwidth demands. required energy consumption and local congestion at a node. An extensive analysis of the Subject of the application of This mechanism both enforces cooperation and balances traf game theory to the control of adhoc networks, including both fic loads to avoid congestion. Stabilization of price and node an extensive review of the literature, and new analysis, is wealth occurs in static networks. provided in the master's Thesis of Juha Leino, entitled The Confidant protocol detects node misbehavior and Applications of GameTheory in Ad Hoc Network”. Helsinki routes traffic around the misbehaving nodes, to isolate them University Of Technology (2003). Leino modeled the inter from the network. Misbehavior of neighboring nodes is action between one node and the rest of the network as an broadcast to the network by observing nodes. A trust record is extensive game. The networks were presumed to be energy used to evaluate the validity of a report, thus disincentivizing 25 constrained, and the nodes to be selfish, with the result stated misbehavior in the reporting process. The reputation infor as the amount of contribution the network can request from a mation is applied by a path manager define a route and rejects node. Leino modeled nodes with power constrained, power access requested by misbehaving nodes. adaptive, omnidirectional transceivers, each of which have a The Core protocol is similar to Confidant; each node main uniform communication demand on the network. tains reputation information, which is updated based on both 30 When a node connects to an ad hoc network, it gains both observation and third party report. A threshold function is benefits and obligations. The other nodes forward its traffic, applied to limit access by nodes based on their reputation, hence it can save energy and reach nodes outside its own resulting in isolation. transmission range, as compared to a single hop transmission. Michiardi et al. analyze whether it is optimal to join or Correspondingly, the node should participate in the network defect from an adhoc network, based on node utility function, 35 functions like the route discovery and traffic forwarding that payoff share, cost of cooperation, number of cooperating consume the resources of the node, in addition to the basic nodes, etc. communications themselves. In order to find participation in Srinivasan et al. apply game theory to model an ad hoc the network advantageous, the node has gain greater benefits network at a connection level, providing a complicated greater than the obligations imposed. This, of course, may be extended game model. Before a user can transmit, all the 40 modeled as a game. The node seeks to minimize energy nodes along the defined route must accept the relay request. consumption, and the network seeks to ensure its functional Energy consumption of terminals is restricted by an expected ity. The decision of the node is to cooperate or to defect. lifetime of batteries, that is, the nodes are modeled as being In one of Leino's models, he requires that reward of for power constrained. A normalized acceptance rate, a propor warding needs to be proportional to the energy consumed tion of Successful and attempted relays through a node, as 45 when the packet is forwarded. He analyzes the situation of observed by neighboring nodes, is sought to be maximized. both a honest node and a cheating node, i.e., one that uses the Urpi et al. model an ad hoc network at packet level. The network's resources without full participation in the network model is based on an estimate of neighboring nodes, the overhead. He concluded that if a node has an opportunity to remaining energy of node, and various packet traffic metrics. cheat, it adversely affects the network far more than mere The payoff of the model is simply the access to packet for 50 defection. Leino also analyzed whether, under his presump warding, weighted by energy cost, provided to a node. tions, a group of honest nodes will Voluntarily aggregate as an Noncooperative game theory offers a basis for analyzing ad hoc network, or would prefer to remain as a set of inde Internet traffic, wherein each user tries to independently pendent uncooperative actors, without benefit of multihop maximize its quality of service. The network operator focuses transmissions. He concludes that under his presumptions, in on maximizing network performance as a whole. Thus, in this 55 some networks, there are nodes which have detrimental case, different players adopt different roles, with different involvement in the ad hoc network, and if all such “loser value functions. Game theory may thus by applied to opti nodes refuse to participate, the network may collapse. The mize routing, flow control, queuing disciplines and traffic proportion of losers drops with minimum energy routing, pricing. While ad hoc network routing is similar to the Inter since the average cost is lowered, making gains from partici net, there are also significant differences. In an ad hoc net 60 pation more likely. There are also networks with no losers, work, routes may be inconsistent. and these provide gains to all participants. Loser nodes tend to Nagle studied the concept of fairness in queuing in packet be in the center of the network, rather than the periphery. Switches. In a first in-first out queue, a selfish node may RealTime Telematics Information Communication saturate the queue with requests. Nagle proposes, as a solu Mobile, self organizing, ad hoc communications networks tion, distinct queues for each source with a round-robin 65 have been proposed for telematics systems, for cellular net scheduler, providing a fair queuing scheme, which encour work extension, long range (multihop) traffic information ages keeping the user's queue as short as possible. communication, and short range collision avoidance systems. US 8,874,477 B2 117 118 Telematics is a recently applied term that now encom does not intrude on a high capacity physical transport passes radio transmitters or receivers in vehicles. Three basic medium with relatively low information content signaling schemes exist: wide area broadcast communication, where all packets. relevant nodes are presumed to be within the same commu In a real time telematics ad hoc network with potentially nication Zone (e.g., satellite radio, RDDS receivers, etc.); 5 unlimited data communication requirements, network con cellular communications, where an array of fixed-position gestion is almost guaranteed, in a continuous network of low power transceivers contiguously blanket a territory, pro mobile nodes. RF interference issues will likely prevent net viding various Zones which allow multiplexing within aband; work capacity from Scaling with node density. Therefore, an and mesh network communications, which allow ad hoc for alternate to CSMA was sought that provided more graceful mation of a communications infrastructure, optionally link- 10 degradation under high load, sensitivity to the importance of ing to various fixed infrastructure. information to be communicated, and efficient utilization of Telematics systems may be used for many purposes, for the communications medium. example, real time traffic information (RTTI), which in an In order to optimize the network, additional information is extreme case may involve communication of live video employed, although this imposes a burden of increased pro streams. In a more modest system, various sensors may pro- 15 tocol overhead, complexity, and potential loss privacy. One vide road and traffic data, as well as weather and incident way to remedy selfish behavior is to increase the cost of acting information. In other words, the appetite of such a system for this way, that is, to impose a cost for access to the network. In bandwidth is potentially limitless, unless constraints are a practical implementation, however, this is problematic, imposed. On the other hand, RTTI is typically not power since under lightly loaded conditions, the “value' of the com constrained, since it is vehicle based rather than hand-held, 20 munications may not justify a fixed cost which might be and therefore the cost of using the system will focus more on reasonable under other conditions, and likewise, under competition for bandwidth (the limited physical transport heavier loads, critical communications may still be delayed or layer (PHY) resource) than power consumed in communica impeded. Therefore, a variable cost, dependent on relative tions. Likewise, communications latency is not critical, “importance', may be imposed. In determining this relative unless full duplex voice communications are Supported. It is 25 importance, a market evaluation, requires a comparison, noted that parked vehicles may also be involved in network termed an auction, is employed. In a multihop network, Such communications, and the frequency band may be shared with an auction is complicated by the requirement for distributing portable communicators with self-contained power sources, payments among the chain of nodes along the route, with each making the economic cost of communications and power node having potential alternate demands for its cooperation consumption a potential factor for some nodes, leading to 30 and resources. According to a more Sophisticated analysis, split strategies. one must also compensate nodes not directly involved in the Likewise, especially for voice communications, interfac communication for their deference. ing with the fixed infrastructure through cellular telephone In a scenario involving a request for information, the auc towers or 802.11 hotspots may impose additional economic tion is complicated by the fact that the information resource constraints on the system. Telematics systems typically 35 content is unknown to the recipient, and therefore the bid is include a GPS geolocation system, which may be of great use blind, that is, the value of the information to the recipient is in mapping nodes for routing navigation functions. Indeed, indeterminate. However, game theory Supports the commu the telematics system may be integrated within a navigation nication of a value function or utility function, which can then system and/or entertainment system. This is relevant to the be evaluated at each node possessing information to be com extent that one considers the incremental cost of the hardware 40 municated, to normalize its value. Fortunately, it is a domi and its market placement. nant strategy in a VCG auction to communicate a truthful The system is designed to operate over a wide range of value. In this case, a value function may instead be commu node densities, from city rush hour traffic to rural highways. nicated, which can then be evaluated at each node possessing Due to a perceived incompatibility of a RTTI system with information to be communicated. In a mere request for infor cellular infrastructure business models, as well as inconsis- 45 mation conveyance. Such as the transport nodes in a multihop tent availability of cellular coverage of roadways, the archi network, or in a cellular network infrastructure extension tecture is designed as a decentralized control, with incidental model, the bid may be a true (resolved) value, since the involvement of the cellular networks, except for voice com information content is not the subject of the bidding; rather it munications outside of the mobile ad hoc network. This is the value of the communications per se, and the bidding decentralized control introduces a substantial level of com- 50 node can reasonably value its bid. plexity, since it must account for rapidly changing network In a cellular network infrastructure extension model, the architecture, various types of channel impairments, hidden bid may represent a resolved value, since the information nodes, and temporal and spatial distance issues, and interfer content is not the subject of the bidding; rather it is the value CCC. of the communications per se. In the case of voice, however, In defining the system, both the available hardware, costs 55 the communications are bidirectional and enduring, thus rais and purposes for use must be considered. Desirable charac ing quality of service and handoff issues. teristics of a telematics system include real time telematics Game theory is most readily applied in the optimization of information communication, multihop Voice communication communications routes through a defined network, to achieve forwarding, decentralized control, and to the extent possible, the best economic surplus allocation. That is, the problem of user privacy. The hardware may include multichannel direc- 60 determining the network topology, and the communications tional Smart antennas, out-of-band signaling and control, themselves, are ancillary, though real, applications of game complex and Sophisticated computational resources, to pro theory. Since the communications incidental to the arbitration vide efficient utilization of an unlicensed or shared band. require consideration of some of the same issues as the under That is, it is clear that a system that provides an omnidi lying communications, corresponding elements of game rectional antenna system with in band signaling and control. 65 theory may apply at both levels of analysis. Due to various is inefficient as compared to a system which directs its trans uncertainties, the operation of the system is stochastic. This mitted energy only in the direction of the intended target, and presumption, in turn, allows estimation of optimality within a US 8,874,477 B2 119 120 margin of error, simplifying implementation as compared to a and overall network capacity. Therefore, a hierarchal scheme rigorous analysis without regard to statistical significance. may be implemented to geographically organize the network The VCG auction is postulated as being optimal for allo at higher analytical levels, and geographic cells may cooper cation of multiple resources between agents. It is “strat ate to appear externally as a single entity. egyproof and efficient, meaning that it is a dominant strategy A Supernode within a Zone may be selected for its Superior for agents to report their true valuation for a resource, and the capability, or perhaps a central location. The Zone is defined result of the optimization is a network which maximizes the by a communication range of the basic data interface for value of the system to the agents. communications, with the control channel having a longer Game theory also allows an allocation of cost between range, for example at least double the normal data communi various recipients of a broadcast or multicast. That is, in many 10 cations range. Communications control channel transmitters instances, telematic information is of value to a plurality of operate on a number of channels, for example at least 7. nodes, and a large set of recipient nodes may efficiently allowing neighboring Zones in a hexagonal tiled array to receive the same information. This allocation is a direct communicate simultaneously without interference. In a geo extension of VCG theory. graphic Zone system, alternate Zones which would otherwise The preferred method for acquiring an estimate of the state 15 be interfering may use an adaptive multiplexing scheme to of the network is through use of a proactive routing protocol. avoid interference. All nodes may listen on all control chan Thus, in order to determine the network architecture state, nels, permitting rapid propagation of control information. each node must broadcast its existence, and, for example, a In order to effective provide decentralized control, either payload of information including its identity, location, itiner each node must have a common set of information to allow ary (navigation vector) and “information value function'. execution of an identical control algorithm, or nodes defer to Typically, the system operates in a continuous state, so that it the control signals of other nodes without internal analysis for is reasonable to commence the process with an estimate of the optimality. A model of semi-decentralized control is also state based on prior information. Using an in-band or out-of known, in which dispersed 'Supernodes', are nominated as band propagation mechanism, this information must propa master, with other topologically nearby nodes remaining as gate to a networkedge, which may be physically or artificially 25 slave nodes. In the pure peer network, complete information defined. If all nodes operate with a substantially common conveyance to each node is required, imposing a relatively estimation of network topology, only deviations from previ high overhead. In a master-slave (or Supernode) architecture, ously propagated information need be propagated. increased reliance on a single node trades-off reliability and CSMA is proposed for the protocol-related communica robustness (and other advantages of pure peer-to-peer net tions because it is relatively simple and robust, and well suited 30 works) for efficiency. A Supernode within a cellular Zone may for ad hoc communications in lightly loaded networks. An be selected for its Superior capability, or perhaps is at a central initial node transmits using an adaptive power protocol, to location or is immobile. achieve an effective transmit range of somewhat less than Once each control node (node or Supernode) has an esti about two times the estimated average inter-nodal distance. mate of network topology, the next step is to optimize net This distance therefore promotes propagation to a set of 35 work channels. According to VCG theory, each agent has an neighboring nodes, without unnecessarily interfering with incentive to broadcast its truthful value or value function for communications of non-neighboring nodes and therefore the scarce resource, which in this case, is control over com allowing this task to be performed in parallel. Neighboring munications physical layer, and or access to information. This nodes also transmit in Succession, providing sequential and communication can be consolidated with the network discov complete protocol information propagation over a relevance 40 ery transmission. Each control node then performs a combi range. natorial solution for the set of simultaneous equations accord If we presume that there is a spatial limit to relevance, for ing to VCG theory (or extensions thereof). This solution example, 5 miles or 10 hops, then the network State propaga should be consistent between all nodes, and the effects of tion may be so limited. Extending the network to encompass inconsistent Solutions may be resolved by collision sensing, a large number of nodes will necessarily reduce the tractabil 45 and possibly an error/inconsistency detection and correction ity of the optimization. Each node has a local estimate of algorithm specifically applied to this type of information. relevance. This consideration is accommodated, along with a As part of the network mapping, communications impair desire to prevent exponential growth in protocol-related data ment and interference sources are also mapped. GPS assis traffic, by receiving an update from all nodes within a node's tance may be particularly useful in this aspect. Where inter network relevance boundary, and a state variable which rep 50 ference is caused by interfering communications, the issue is resents an estimate of relevant status beyond the arbitrarily a determination of a strategy of deference or competition. If defined boundary. The propagation of network State may thus the interfering communication is continuous or unresponsive, conveniently occur over a finite number of hops, for example then the only available strategy is competition. On the other 5-10. hand, when the competing system uses, for example, a CSMA Under conditions of relatively high nodal densities, the 55 system, such as 802.11, competition with Such a communi system may employ a Zone strategy, that is, proximate groups cation simply leads to retransmission, and therefore ulti of nodes are is treated as an entity for purposes of external mately increased network load, and deference strategy may state estimation, especially with respect to distant nodes or be more optimal (dominant), at least and until it is determined Zones. Such a presumption is realistic, since at extended that the competing communication is incessant. Other com distances, geographically proximate nodes may be modeled 60 munications protocols, however may have a more or less as being similar or inter-related, while at close distances, and aggressive strategy. By observation of a system over time, its particularly within a Zone in which all nodes are in direct strategies may be revealed, and game theory permits compo communication, internode communications may be subject sition of an optimal strategy. to mutual interference, and can occur without Substantial The optimization process produces a representation of an external influence. Alternately, it is clear that to limit latencies 65 optimal network architecture during the Succeeding period. and communication risks, it may be prudent to bypass neigh That is, value functions representing bids are broadcast, with boring nodes, thus trading latency for power consumption the system then being permitted to determine an optimal real US 8,874,477 B2 121 122 valuation and distribution of that value. Thus, prior to may be dealt with by cryptographic techniques, in particular completion of the optimization, potentially inconsistent allo public key cryptography, in which the currency is placed in a cations must be prevented, and each node must communicate cryptographic “envelope” addressed to the intended recipi its evaluation of other node's value functions, so that the ent, e.g., is encrypted with the recipient’s public key, which optimization is performed on a normalized economic basis. must be broadcast and used as, or in conjunction with, a node This step may substantially increase the system overhead, and identifier. This makes the payment unavailable to other than is generally required for completion of the auction. This valu the intended recipient. The issue of holding the encrypted ation may be inferred, however, for transit nodes in a multihop token hostage and extorting a portion of the value to forward network path, since there is little subjectivity for nodes solely the packet can be dealt with by community pressure, that is, in this role, and the respective value functions may be persis 10 any node presenting this (or other undesirable) behavior tent. For example, the valuation applied by a node to forward might be ostracized. The likelihood of this type of misbehav information is generally content and involved party indepen ior is also diminished by avoiding monetization of the virtual dent. currency. A particular complication of a traffic information system is This currency generation and allocation mechanism gen that the nature of the information held by any node is private 15 erally encourages equal consumption by the various nodes to that node (before transmission), and therefore the valuation over the long term. In order to discourage consumption of is not known until after all bids are evaluated. Thus, prior to bandwidth, an external tax may be imposed on the system, completion of optimization, each node must communicate its that is, withdrawing value from the system base on usage. evaluation of other nodes value functions, so that the opti Clearly, the effects of such a tax must be carefully weighed, mization is performed on an economic basis. This required since this will also impose an impediment to adoption as step Substantially increases the system overhead. This valua compared to an untaxed system. On the other hand, a similar tion may be inferred, however, for transit nodes in a multihop effect use-disincentive may be obtained by rewarding low network path. consumption, for example by allocating an advertising Sub After the network usage is defined, compensation is paid to sidy between nodes, or in reward of deference. In a model those nodes providing value or Subjected to a burden (includ 25 telematics system, an audio and/or visual display provides a ing foregoing communication opportunity) by those gaining a useful possibility for advertising and sponsorship; likewise, benefit. The payment is generally of a virtual currency, with location based services may include commercial services. no specific true value, although the virtual currency system Each node computes a value function, based on its own provides a convenient method to tax the system. knowledge state, risk profile and risk tolerance, and wealth, Exerting external economic influences on the system may 30 describing the value to it of additional information, as well as have various effects on the optimization, and may exacerbate its own value for participating in the communications of differences in subjective valuations. The application of a others. The value function typically includes a past travel monetary value to the virtual currency Substantially also history, future travel itinerary, present location, recent com increases the possibility of misbehavior and external attacks. munication partners, and an estimator of information strength On the other hand, a virtual currency with no assessed real 35 and weakness with respect to the future itinerary. It may be value is self-normalizing, while monetization leads to exter presumed that each node has a standard complement of sen nal and generally irrelevant influences as well as possible sors, and accurately acquired descriptive data for its past arbitrage. External economic influences may also lead to travel path. Otherwise, a description of the available informa benefits, which are discussed in various papers on non-Zero tion is required. One advantage of a value function is that it Sum games. 40 changes little over time, unless a need is satisfied or circum In order to provide fairness, the virtual currency (similar to stances change, and therefore may be a persistent attribute. the so-called “nuglets” or “nuggets' proposed for use in the Using the protocol communication system, each node Terminodes project) is self-generated at each node according transmits its value function (or change thereof), passes to a schedule, and itselfmay have a time dependent value. For through communications from neighboring nodes, and may, example, the virtual currency may have a half-life or tempo 45 for example transmit payment information for the immediate rally declining value. On the other hand, the value may peak past bid for incoming communications. at a time after generation, which would encourage deference Messages are forwarded outward (avoiding redundant and short term savings, rather than immediate spending, and propagation back to the source), with messages appended would allow a recipient node to benefit from virtual currency from the series of nodes. Propagation continues for a finite transferred before its peak value. This also means that long 50 number of hops, until the entire community has an estimate of term hoarding of the currency is of little value, since it will the state and value function of each node in the community. eventually decay in value, while the system presupposes a Advantageously, the network beyond a respective community nominal rate of spending, which is normalized among nodes. may be modeled in simplified form, to provide a better esti The variation function may also be adaptive, but this poses a mate of the network as a whole. synchronization issue for the network. An external estimate 55 After propagation, each node evaluates the set of value of node wealth may be used to infer counterfeiting, theft and functions for its community, with respect to its own informa failure to pay debts, and to further effect remediation. tion and ability to forward packets. Each node may then make The currency is generated and Verified in accordance with an offer to supply or forward information, based on the pro micropayment theory. Micropayment theory generally vided information. In the case of multihop communications, encompasses the transfer of secure tokens (e.g., cryptographi 60 the offers are propagated to the remainder of the community, cally endorsed information) having presumed value, which for the maximum number of hops, including the originating are intended for verification, if at all, in a non-real time node. At this point, each node has a representation of the State transaction, after the transfer to the recipient. The currency is of its community, with community edge estimates providing circulated (until expiration) as a token, and therefore is not consistency for nodes with differing community Scopes, the Subject to immediate authentication by source. Since these 65 valuation function each node assigns to control over portions tokens may be communicated through an insecure network, of the network, as well as a resolved valuation of each node the issue of forcing allocation of payment to particular nodes for Supplying the need. Under these circumstances, each node US 8,874,477 B2 123 124 may then evaluate an optimization for the network architec plurality of bidders, a utility function representing a value to ture, and come to a conclusion consistent with that of other the bidder to obtain of a set of resources, receiving, from each members of its community. If supported, node reputation may of a plurality of offerors, a utility function representing a be updated based on past performance, and the reputation value to the offeror to relinquisha set of resources, computing applied as a factor in the optimization and/or externally to the a set of successful bids from the plurality of bidders and optimization. As discussed above, a VCG-type auction is plurality of offers, a Successful bid comprising a matching of employed as a basis for optimization. Since each node a maximum aggregate value of the sets of resources to the receives bid information from all other nodes within the bidders and a minimum aggregate value of the sets of maximum node count, the VCG auction produces an opti resources to the offerors, wherein the maximum aggregate mized result. 10 value of bids equals or exceeds the minimum aggregate value Transmissions are made in frames, with a single bidding of offers; and receiving for each set of resources from a bidder process controlling multiple frames, for example a multiple placing a respective successful bid a Vickrey price, and pay of the maximum number of hops. Therefore, the bid encom ing to for each set of resources to an offeror of a respective passes a frame's-worth of control over the modalities. In the Successful bid each its offer price, with any surplus being event that the simultaneous use of, or control over, a modality 15 allocated to bidders based on a value bid. The bidder utility by various nodes is not inconsistent, then the value of the function may be evaluated based on private information of an respective nodes may be summed, with the resulting alloca offeror, and communicated as a normalized value. tion based on, for example, a ratio of the respective value It is noted that in an auction for a synthetic economic value functions. As a part of the optimization, nodes are rewarded generated by a generation function, the payment may be by not only for Supporting the communication, but also for defer currency or the ability to generate currency. That is, an agent ring their own respective needs. As a result, after controlling may transfer to another agent the micropayment or the ability the resources, a node will be relatively less wealthy and less to generate a micropayment, both within the auction or out able to subsequently control the resources, while other nodes side of it. Normally, the valuation of the ability to generate will be more able to control the resources. The distribution to synthetic currency would be discounted from the future value deferred nodes also serves to prevent pure reciprocal commu 25 by a Subjective discount rate dependent on the recipient. In nications, since the proposed mechanism distributes and other cases, it may be expedient to apply an objective discount dilutes the wealth to deferring nodes. rate which is calculable without requiring Substantial com Because each node in the model presented above has com munications (or even any communications) with each agent. plete information, for a range up to the maximum node count, Thus, each agent can anticipate its future communication the wealth of each node can be estimated by its neighbors, and 30 needs and target Sufficient resources at the appropriate time to payment inferred even if not actually consummated. (Failure meet that need, while leaving for other agents the communi of payment can occur for a number of reasons, including both cation resources when these are not needed. malicious and accidental). Because each hop adds significant cost, the fact that nodes beyond the maximum hop distance Eighth Embodiment are essentially incommunicado is typically of little conse 35 quence; since it is very unlikely that a node more than 5 or 10 According to a further aspect of the invention, it is desired hops away will be efficiently included in any communication, to understand the Subjective risk aversion profile of a person. due to the increasing cost with distance, as well as reduction Risk-aversion is a significant deviation from rationality in reliability and increase in latency. Thus, large area and which can be quantified and understood, and further a calcu Scalable networks may exist. 40 lus is available for applying the risk aversion to normalize Typically, cryptography is employed for both authentica systems in which rationality is presumed. Accordingly, the tion and to preserve privacy. External regulation, in a legal method comprises presenting a game for play by a person, sense at least, is typically imposed by restrictions on hard wherein the payoff of the game is real and beneficial to the ware and Software design, as well as Voluntary compliance at person. That is, the incentive and risk must be real, with some risk of detection and legal sanction. 45 limits on the ability to extrapolate beyond the scope of risk presented. The person is then sufficiently observed during the Conclusion game play to comprehend a risk aversion profile of the user. Typically, the game is automated, but this is not required, and, The use of game theory as a basis for analyzing ad hoc in fact, a competition between two or more players is pos networks provides a basis for understanding the behavior of 50 sible. This scenario is generally quite beneficial where the complex networks of independent nodes. By presuming a stakes of the game are identical or similar to the risk aversion degree of choice and decision-making by nodes, we obtain an personality attribute sought to be defined. The comprehended analysis that is robust with respect to Such considerations. risk aversion profile may then be used to modify a rationality The principal issues impeding deployment are the inherent expectation for the person. The modified rationality expecta complexity of the system, as well as the overhead required to 55 tion may then be applied to optimize an interaction with the continuously optimize the system. Further work will allow a person outside of the game play environment. determination of a set of simplifying presumptions to reduce This process is particularly useful for creating a user-agent protocol overhead and reduce complexity. to act on behalf of the user, in a manner commensurate with a The ad hoc network does not exist in a vacuum. There are subjective profile of the user, or to subjectivize a presentation various competing interests seeking to use the same band 60 of risk data to a user. For example, in the probability based width, and technological Superiority alone does not assure user interface discussed above, the event-probability map dominance and commercial Success. Game theory may also may be analyzed based on the subjective risk tolerance of the be used as a tool to analyze the entities which seek to deploy user, and the output optimized accordingly. This method may ad hoc networks, especially where they compete. also be applied for optimally pairing a user with another The present invention therefore provides an automated 65 person or process, based on compatibility. negotiation for control of a set of resources by competing There has thus been shown and described novel commu bidders and offerors, comprising receiving, from each of a nications devices and systems and methods which fulfill all US 8,874,477 B2 125 126 the objects and advantages sought therefor. Many changes, Agent-mediated Electronic Commerce: Scientific and Tech modifications, variations, combinations, Subcombinations nological. Sierra (2001) www.iiia.csic.es/-sierra/ar and other uses and applications of the Subject invention will, ticles/2001/Roadmap.AMEC/Roadmap.ps however, become apparent to those skilled in the art after On the Integration of IR and Databases—de Vries, Wilschut considering this specification and the accompanying draw (1999) www.cs.utwente.nl/-arjen/Pics/ds8 short.ps.gZ ings which disclose the preferred embodiments thereof. All Towards a Harmonization of UML-Sequence Diagrams and Such changes, modifications, variations and other uses and MSC Rudolph, Grabowski, Graubmann (1999) www.it applications which do not depart from the spirit and scope of m.mu-luebeck.de/publications/SDL99-Harmoniza the invention are deemed to be covered by the invention, tion.ps.gZ 10 Logarithmic Time Cost Optimal Parallel Sorting is Not Yet which is to be limited only by the claims which follow. Fast in. Natvig (1996) www.idintnu.no/-lasse/publics/ APPENDICES SC90.ps CMIFed: A Presentation Environment for Portable.—van Rossum. (1993) www.cwi.nl/ftp/CWIreports/CST/CS The following resources, each of which is expressly incor 15 R9305.ps.Z. porated herein by reference, provides a basis for understand Requirements of Traffic Telematics to Spatial Databases— ing aspects of the invention and has implications for the Brinkhoff (1999) www.fh-oldenburg.de/iapg/personen/ design, control, and analysis of systems and networks, and the brinkhof/SSD99.pdf method of operation thereof. A Task-Specific Ontology for the Application and. Shahar, Other Patents by Inventor Hereof Miksch, Johnson www-smistanford.edu/pubs/SMI Re The following patents are expressly incorporated herein by ports/SMI-96-0649.pdf reference: U.S. Pat. Nos. 6,865,825, 6,850,252, 6,791,472, Practical Considerations in Building a Multi-Lingual.— 6,640,145, 6,429,812, 6,418,424, 6,400.996, 6,252,544, Business Letters John acl.ldc.upenn.edu/W/W97/W97 6,230,501, 6,081,750, 5,920,477, 5,903,454, 5,901,246, 0906.pdf 5,875,108, 5,867,386, and 5,774,357. 25 BPP: A Protocol for Exchanging Pricing Information.— Telematics References Oberle, Ritter, Wehrle www.tm.uka.de/-wehrle/pubs/ On the Propagation of Long-Range Dependence in the Inter 2001 07 HPSR01 BPP.pdf net Veres, Kenesi, Molnár. (2000) galahad.elte.hu/~vat An Open Architecture for Evaluating Arbitrary Quality of tay/cikkeim/vmkv.pdf Service. Wehrle (2001) www.tm.uka.de/-wehrle/pubs/ Anthill: A Framework for the Development of Agent 30 2001 06 ICN01 KIDS.pdf Based.—Babaoglu, Meling. (2002) www.cs.unibo.it/ba Improving the performance of TCP on guaranteed band baoglu/papers/icacs02.pdf width. Ritter, Wehrle, Wolf www.tm.uka.de/-wehrle? Automatic Web Page Categorization by Link and Context.— pubs/2001 02 KIVS01 GR TCP.pdf Attardi, Gulli. (1999) aureiei.pi.cnr.it/-fabrizio/Publica Efficient Image Compression of Medical Images. Using the 35 Wavelet Transform And Fuzzy rtsimage.di.uoa.gr/publi tions/THAI99/THAI99.ps cations/euro00-1.pdf Distributed Interactive Media—Mauve (2000) www.infor Advanced Mechanisms for Available Rate Usage in ATM.— matik.uni-mannheim.de/informatik/pi4/publications/li Bless, Holzhausen. www.tm.uka.de/-wehrle/pubs/2001 brary/Mauve2000dps.gZ 04 ICATMO1 AvRateUsage.pdf Weak Bisimulation for Fully Probabilistic Processes—Baier, 40 An Extensible Agent Architecture for a Competitive. Hoen, Hermanns (1999) web.informatik.uni-bonn.de/I/papers/ Bohte: www.cwi.nl/~hoen/publications/SEN-R0217.ps acta infps Elements of an Open Framework for Pricing in the Future.— Structured Multimedia Authoring Hardman, van Rossum, Gerke, Ritter. www.tm.uka.de/-wehrle/pubs/2000 Bulterman (1993) www.cwi.nl/ftp/CWIreports/CST/ 09 Qofis00 Pricing.pdf CS R9304.ps.Z. 45 Traffic Priorization and Differentiation with Active Queue.— Solving the modeling problems of object-oriented lan Walter, Wehrle (2002) www.tm.uka.de/-wehrle/pubs/ guages.—Aksit, Tekinerdogan (1998) trese.cs.utwente.nl/ 2002 09 ICTSM02 AF AR.pdf aop-ecoop98/papers/Aksit.pdf A Simulation Suite for Internet Nodes with the Ability to.— Group Communication in Differentiated Services Net Wehrle, Reber, Kahmann www.tm.uka.de/-wehrle/pubs/ works—Bless, Wehrle (2001) www.tm.uka.de/-bless/ 50 2001 01 CNDS01 Simpdf IQ2001-ds multicast.pdf Image Compression Using the Wavelet Transform. On Tex Ad Hoc Relay Wireless Networks over Moving Vehicles on tural Regions of Interest rtsimage.di.uoa.gr/publications/ Highways—Chen, Kung, Vlah (2001) www.eecs.harvard euro 98.pdf .edu/~htk/publication/2001-mobihoc-ckv.pdf Towards Better Support of Transaction Oriented.—Bless, A Framework for Generating Network-Based Moving 55 Holzhausen. www.tm.uka.de/-bless/Qof IS2001-qf-on Objects—Brinkhoff (2002) www.fh-wilhelmshaven.de/ line.pdf oow/institute/iapg/personen/brinkhoff/paper/ User feedback on the use of public key certificates—Klobu GeoInformatica2002.pdf car (2001) www.e5.ijs.si/staff/tomaZ/erk2001.pdf The Amsterdam Hypermedia Model: extending hypertext.— Using Realistic Internet Topology Data for Large Scale Net Hardman, Bulterman. (1993) www.cwi.nl/ftp/CWIre 60 work. Bless www.tm.uka.def-blessfomnet ws 2002 ports/CST/CS-R9306.ps.Z. 1.pdf On functional equivalence of certain fuzzy controllers and— The Trust Factor in the Virtual Enterprise. Mobile Broad Koczy, Tikk, Gedeon (2000) www.mft.hu/publications/ band Service paula.oulu.fi/Publications/Submited/ tikk/tikk3.pdf ICE2000fpdf Specification of a Service Management Architecture. May 65 User and Session Mobility in a Plug-and-Play Network.— erl, Nochta. (2000) www.cooperation-management.de/ Mazen Malek Shiaa www.item.ntinu.no/-plugandplay/ publikationen/paper?usm2000-mayerl-nochta.pdf publications/eunice2002.pdf US 8,874,477 B2 127 128 Dynamic Aggregation of Reservations for Internet Ser A Framework For Video Modelling Petkovic, Jonker vices—Bless www.tm.uka.de/-bless/ictsm 10-daris.pdf (2000) www.cwi.nl/~acoi/DMW/documents/./world/ Advanced Service Creation Using Distributed Object Tech innsb.ps nology www.ee. Surrey.ac.uk/Personal/G.Pavlou/Publica An Architecture For Reverse Charging In The Internet— tions/Journal-papers/Adam-02b.pdf 5 Sprenkels, Parhonyi, Pras. (2000) www.snmp.cs.utwen Semantics and Verification of UML Activity Diagrams for te.nl/nm/research/results/publications/sprenkels/spren Workflow. Eshuis (2002) www.home.cs.utwente.nl/-es kelsram-reverse-charging.pdf huis/thesis-hyperlinks.pdf Behavior of a Bayesian Adaptation Method for. Fredouille, be cited as a National Research Council report. The Four Mariéthoz. (2000) ftp.idiap.ch/pub/marietho/publications/ Way www.7.nationalacademies.org/CSTB/wp geo arm 10 strong.pdf AdaptICASSP2000 IRR02ps.gZ Support Specification and Selection in TAPAS Aagesen, Software Process Modeling and Evolution in EPOS—Jac Anutariya, Shiaa, Helvik www.item.ntnu.no/-plugand cheri, Larsen, Conradi (1992) www.idt.unit.no/-epos/Pa play/publications/euniceCap2002.pdf pers/capri-final.ps Profiling and Internet Connectivity in Automotive. Cilia, 15 Individual Management of Personal Reachability in. Re Hasselmeyer. www.cs.ust.hk/vldb2002/VLDB2002-pa ichenbach. (1997) www.inftu-dresden.de/-hf2/publ/ pers/S33P08.pdf 1997/RDFR 971 FIPSec.pdf MIETTA—A Framework for Uniform and Multilingual A Generic Software Component Framework for Distrib Access to—Structured Database www.xtramind.com/~ uted.—Batteram, Idzenga (1997) www.trc.nl/events/ holger/paper/ira100.pdf ecscw97oogp/papers batteram.pdf Definition And Utilisation Of OMG IDL TO TTCN-3 MAP Efficient Partitioning of Sequences—Olstad, Manne (1995) PINGS Ebner, Yin, Li www.itm.mu-luebeck.de/publica www.ii.uib.no/-fredrikm/fredrik?papers/efficient.ps tions/idltottcn3 mapping.pdf Integrated Video Archive Tools—Hjelsvold, Langorgen, Campiello—New user interface approaches for commu Midtstraum. (1995) www.idt.ntinu.no/IDT/grupper/DB nity.—Antonietta Grasso Michael (1998) www.ll.informa 25 grp/tech papers/ACM-MM95.ps tik.tu-muenchen.de/publications/pdf/Grasso 1998a.pdf MULINEX: Multilingual Web Search and Navigation SDL and MSC Based Test Generation for Distributed. Joanne Capstick Abdel (1998) speech.ftw.at/ -gor/pub/ Grabowski, Koch. (1998) www.itm.mu-luebeck.de/publi twilt 98/mulinex-twilt 98.pdf cations/SDL99-DistributedTesting ps.gZ A Non-Interleaving Semantics for MSC Heymer (1998) Autolink—A Tool for Automatic Test Generation from.— 30 www.itm.mu-luebeck.de/publications/ANIS/ANISps.gZ Koch, Grabowski. (1998) www.itm.mu-luebeck.de/publi cations/A-98-05/Report-A-98-05.ps.gZ An Overview Of The Cave Project Research. Bimbot, Hut PATROCLOS: A Flexible and High-Performance Transport ter. (1998) www.ubilab.org/publications/print versions/ Subsystem Braun (1994) www.iam.unibe.ch/-braun/lit/ pdf/bim98.pdf VanC p.p.S. gZ 35 Cross Language Retrieval with the Twenty-One system— Multilingual Generation and Summarization of Job. Som Wessel Kraaij (1998) trec.nist.gov/pubs/trecó/papers/ ers, Black. (1997) sskkili.gu.se/Tree/Reports/anlp97/AN twentyone.ps LP paper.ps Verifying Business Processes using SPIN Janssen, Matee Real-time TTCN for testing real-time and multimedia sys scu, Mauw. (1998) ftp.win.tue.nl/pub/techreports/sjouke/ tems—Walter, Grabowski (1997) www.itm.mu-lue 40 spin98.ps.Z beck.de/publications/IWTCS98RT/IWTCS98-RT-TTC Implementation of a Parallel Transport Subsystem on N.ps.gZ a.—Braun, Schmidt (1993) www.telematik.informa Applying SAMSTAG to the B-ISDN protocol SSCOP tik.uni-karlsruhe.de/~schmidt/schmidt hpdc93.ps.gZ Grabowski, Scheurer, Dai. (1997) www.itm.mu-lue Uniform Versioning: The Change-Oriented Model—Munch, beck.de/publications/IWTCS97SSCOP/IWTCS97-SS 45 Larsen, Gulla. (1993) www.idt.unit.no/-bjornmu?cov COPps.gZ Scm4.ps A Continuously Available and Highly Scalable.—Hvass Architectural Considerations for Personal Mobility In The hovd. (1991) www.idintnu.no/IDT/grupper/DB-grp/tech Wireless Internet www.item.ntinu.no/-plugandplay/publi papers/hpts91.ps cations/PWC2002.pdf A Modular VLSI Implementation Architecture for.—Braun, 50 Synchronization. In Ad Hoc Networks Based On UTRA Schiller. (1994) www.iam. unibe.ch/-braun/lit/vanc bsz. TDD Ebner, Rohling, Halfmann, Lott (2002) pS.gZ www.et2...tu-harburg.de/Mitarbeiter/Ebner/ Transportation modelling methods and advanced transport PIMRC02 Ebner.pdf telematics. Toint (1993) thales.math.fundp.ac.be/pub/ Pervasive Privacy with Identity Management—Jendricke, reports/TR92-24.ps 55 Kreutzer, Zugenmaier www.teco.edu/-philip/ Impact of ATM ABR Control on the Performance of TCP ubicomp2002ws/organize/identity.pdf Tahoe and.—Feng, Ghosal networks.cs.ucdavis.edu/~ Formalising Hinting in Tutorial Dialogues—Tsovaltzi, ghosal/globecom/globecom paperps Matheson (2002) www.ltg.ed.ac.uk/edilog/papers/185.pdf Detecting Opportunities for Parallel Observations on the.— TimedTTCN-3 A Real-Time Extension For TTCN-3– Michael Lucks Space (1992) www.stsci.edu/public/sst/ 60 Dai, Grabowski, Neukirchen www.itm.mu-luebeck.de/ poms/poms.ps publications/timedttcn3.pdf Expression Of Interest Network Of Excellence www.cti Mapping CORBAIDL to TTCN-3 based on IDL to TTCN-2 tutwente.nl/internal/info-programmes/IST/EoI-NoE/ mappings—Ebner (2001) www.itm.mu-luebeck.de/publi eHealth.pdf cations/FBT2001/fbt2001 ebner.pdf Optimisation of trapezoidal membership functions in a fuZZy 65 Mobile Identity Management—Jendricke, Kreutzer, Zugen rule System. By The www.univ-savoie.fr/labos/lamii/ maier ftp.informatik.uni-freiburg.de/documents/reports/ Busefal/Documents/85-05.pdf report178/report00178-ps.gZ US 8,874,477 B2 129 130 Causal Reasoning. In A Medical-Knowledge Based System On the Queue Tail Asymptotics for General Multifractal Traf www.newcastle.edu.au/school/design-comm-info/staff fic Molnar, Dang, Maricza hsnlab.ttt.bme.hu/~molnar/ gibbon/ecis98.ps files/ifipnw02.pdf A Secure Web-based Medical Digital Library Architecture Feature Ranking Based On Interclass Separability For Based On thalis.cs.unipi.gr/-pap/medical.pdf Fuzzy. Tikk, Gedeon (2000) www.mft.hu/publications/ IPv6 Autoconfiguration in Large Scale Mobile Ad-Hoc Net tikk/tikk8.pdf works—Kilian Weniger Martina www.iponair.de/publica Public Key Certificate Revocation Schemes—Arnes (2000) tions/Weniger-EuroWireless02.pdf www.pVV.ntnu.no/-andrearn/certrev/thesis?CertRevThe Serviceware Framework For Developing 3g Mobile-Services sis 29Feb2000.ps.gZ 10 Competitive market-based allocation of consumer atten Sahin Albayrak (2002) www.ssgrrit/en/ssgrr2002w/pa tion. Bohte, Gerding. (2001) www.cwi.nl/ftp/CWIre pers/37.pdf ports/SEN/SEN-R0131ps. Z. Traffic Control methods For High Speed-Packet Switched Towards Adaptive, Resilient and Self-Organizing. Montre Networks (2002) hsnlab.ttt.bme.hu/~molnar/files/ sor, Meling. www.elet-pohmi.it/Users/DEI/Sections/Com pch2002.pdf 15 pEng/GianPietro. Picco/ntwo2-p2p/papers/18.pdf The Impact Of Filtering On Spatial Continuous Queries— Stability of interpolative fuzzy KH controllers Tikk, Joo. Brinkhoff (2002) www.fh-wilhelmshaven.de/oow/insti (2002) www.mft.hu/publications/tikk/tikka.pdf tute/iapg/personen/brinkhoff/paper/SDH2002.pdf Comprehensive analysis of a new fuZZy rule interpolation On the Generation of Time-Evolving Regional Data— method Tikk, Baranyi (2000) www.mft.hu/publications/ TZouramanis. (2002) delab.csd.auth.gr/papers/ tikk/tikk2.pdf GEO02tvm.pdf Design and Use of Clinical Ontologies:—Curricular Goals A dynamic service delivery framework based on the OSGi For www-smi. Stanford.edu/pubs/SMI Reports/SMI model Vos, Buytaert, Buytaert (2002) www.ssgrrit/en/ 1999-0767.pdf ssgrr2002w?papers/165.pdf Toward Standardization of Electronic Guideline.—Elkin, Applying bacterial algorithm to optimise trapezoidal.— 25 Peleg, Lacson. (2000) www-smi. Stanford.edu/pubs/ Botzheim, Hamori, Kóczy (2001) www.mft.hu/hallg/ SMI Reports/SMI-2001-0865.pdf 200106.pdf Extracting Information for Automatic Indexing of Multime Deadlock Probability in Unrestricted Wormhole Routing dia Material—Saggion. (2002) parlevink.cs.utwente.nl/ Networks—Folkestad, Roche (1997) ftp.cs.ucla.edu/tech projects/mumis/documents/mumis-lrec2002.pdf report/1997-reports/970008.ps.Z. 30 Investigation of a new alpha-cut based fuZZy.—Baranyi, The Infostations Challenge: Balancing Cost and Ubiquity Tikk, Yam. www.mft.hu/publications/tikk/tikk11.pdf In Delivering www.winlab.rutgers.edu/~ryates/papers/ The Evolution of Jini Technology. In Telematics. By ieeepc6b.ps www.s. sun.com/software/iini/whitepapers/PsiNapticT A Proposal for a Combination of Compression and Encryp elematics.pdf tion Lutz Vorwerk Thomas www.informatik.uni-trier.de/ 35 Desktop Synchronous Distance Learning Application—En -Vorwerk/publics/perth.ps hanced With Efficient ru6.cti.gr/Publications/645.pdf: Mobility Management In Plug And Play Network Archi ru6.cti.gr/Publications/904.pdf tecture Mazen Malek (2002) www.item.ntinu.no/-plugan Adaptive compression of DICOM-image data—Hludov, dplay/publications/smartnet2002.pdf Engel, Meinel www.informatik.uni-trier.de/-meinel/pa Adaptive Importance Sampling Simulation Of Queueing 40 pers/tes 2.ps Networks—de Boer www.informs-cs.org/wsc00papers/ Telematic Tools to Support Group Projects in Higher Educa O86.PDF tion—Jan Van Der www.ub.utwente.nl/webdocs/ctit/1/ Ics—Forth Erich Leisch Stelios (1999) www.ics.forth.gr/ 00000004.pdf ICS/acti/cmi hta/publications/papers/1999/hector solu A Receiver-initiated WDM Multicast Tree Construction Pro tions/hector Solutions.pdf 45 tocol. Ip Dense Mode www.ub.utwente.nl/webdocs/ctit/ Co-Evolution of Bargaining Strategies in a Decentralized.— 1/00000032.pdf Eymann (2001) www.iig.uni-freiburg.de/-eymann/publi AUTOLINK A Tool for the Automatic and—Schmitt, cations/TEymann01.pdf Koch. (1997) 141.83.21.121/publications/ALASA/ VBR Video Source Characterization And A Practical.—Cse SchmKochGraHog.ps.gZ lényi, Molnár hsnlab.ttt.bme.hu/-molnar/files/VBRtel 50 The Term Processor Generator Kimwitu van Eijk, Belin SyS. pS.gZ fante, Eertink. (1997) wwwtios.cs.utwente.nl/kimwitu/ An approach to dynamic reconfiguration of—Almeida, tacas97.ps.gZ Wegdam. (2001) www.cs.utwente.nl/-alme/cvitae/ Security Engineering of Lattice-Based Policies—Bryce sbrc2001-final-footnote.pdf www.ub.utwente.nl/webdocs/ (1997) set.gmd.de/-kuehnhsr/CWASAR/ ctit/1/0000004f.pdf 55 D06.1.anniS.ps.gZ A GSM/GPS Receiver With a Bandpass Sigma-Delta Ana A scheme for adaptive biasing in importance sampling— log. Müller, Boehm, Hentschel (1999) www.ifnet.tu Heegaard (1997) www.idt.unit.no/-poulh/publications/ dresden.de/MNS/veroeffentlichungen/1999/Mueller IS-adaptive.ps www.item.ntinu.no/-poulh/publications/ T EUW 99.pdf adapt-scheme-abstract.ps An Intelligent Educational Metadata Repository Bassili 60 Towards the Industrial Use of Validation.—Ek, Grabowski. ades, Kokkoras. (2002) www.csd.auth.gr/%7Elpis/publi www.itm.mu-luebeck.de/publications/A-97-03/SDL-Fo cations/crc-chapter1.pdf rum-97.ps.gZ Messor: Load-Balancing through a Swarm of Autonomous A Browser for a Versioned Entity-Relationship Database— Agents—Montresor, Meling, Babaoglu (2002) www.c- Gulla (1992) www.idt.unit.no/-epos/Papers/browser.ps S.unibo.it/babaoglu/papers/ap2p02.pdf 65 The Eternal Resource Locator: An Alternative Means of.- Exact trade-off between approximation accuracy and Tikk, Vaclav www.usenix.org/publications/library/proceedingS/ Baranyi www.mft.hu/publications/baranyi/baranyi2.pdf ec98/full papers/anderson/anderson.pdf US 8,874,477 B2 131 132 A Continuously Available and Highly Scalable.—Hvass The Process of Designing Appropriate Smart Homes: Includ hovd. (1991) www.idintnu.no/IDT/grupper/DB-grp/tech ing.—Guy Dewsbury Bruce www.smartthinkin papers/hpts91.ps g.ukideas.com/Dewsbury et al Appropriate desi A Parallel Implementation of XTP on Transputers—Braun, gn of Smart homes.pdf Zitterbart (1991) www.iam.unibe.ch/-braun?lit? 5 Transparent Dynamic Reconfiguration for CORBA lcnl6 Xtp.ps.gZ Almeida, Wegdam, van. (2001) www.cs.utwente.nl/~ Stublets: A Notion for Mobility-Aware Application Adap alme/cvitae/doa01.pdf tion Dietmar Kottmann Christian (1996) ftp.diku.dk/diku/ Dissemination of Mutable Sets of Web Objects—Buchholz, distlab/wmr96/sommer ps.gZ Goebel, Schill, Ziegert (2001) wwwrn.inf.tu-dresden.de/- ATM Traffic Measurements and Analysis on a Real Test 10 buchholz/askom/PDCS2001.pdf bed Molnár, Cselenyi. (1996) hsnlab.ttt.bme.hu/-mol How to Support the Negotiation of Service Level Agree nar/files/itc2.ps.gZ ments. Köppel, Bóning, Abeck (1999) www.coopera A CORBA Platform for Component Groupware Hofte, van tion-management.de/publikationen/paper/isas'99 Abeck der Lugt, Bakker (1996) www.telin.nl/publicaties/1996/ 15 Boening-Koeppel.pdf ozchi96.pdf PAMINA: A Certificate Based Privilege Management Sys The European Web Index: An Internet Search Service for.— tem. Nochta, Ebinger, Abeck (2002) www.isoc.org/isoc/ Lundberg, Ardó. (1996) www.lub.lu.se/desire/radar/re conferences/indsS/02/proceedings/papers/nochta.pdf ports/D3.12/D3.12.v1.0.ps Modeling IT Operations to Derive Provider Accepted Man Results Of The CEO Project WWW Management Haze agement.—Abeck, Mayerl (1999) www.cooperation-man winkel, Van Hengstum, Pras (1996) www.snmp.cs.utwen agement.de/publikationen/paper/im99 abeck-mayerl.pdf te.nl/nm/research/results/publications/pras/WWW.pdf On deriving rules for nativised pronunciation in. Trancoso, Towards Integrated QoS Management Schmidt, Zitterbart Viana. (1999) www.12finesc.pt/documents/papers/ (1995) www.telematik.informatik.uni-karlsruhe.de/ Trancoso99b.pdf ~schmidt/schmidt hipp94.ps.gZ 25 A CANDLE to light the way? Sebastian Abeck Jodok On engineering Support for business process modelling.— www.cooperation-management.de/publikationen/paper/ Franken, de Weger. (1996) www.home.cs.utwente.nl/ ACandletoLightTheWay.pdf ~pires/publications/bpr96.pdf Internet Agents for Telemedicine Services—Mea, Roberto, Optimization of Spatial Joins Using Filters Veenhof, Apers, Conti, Di. (1999) www.telemed.uniudit/papers/VDM Houtsma (1995) wwwis.cs.utwente.nl:8080/isdoc/confpa 30 MI99.pdf per/veenhof.BNCOD95.accepted.ps.gZ Supporting Secure and Transparent Delegation in the Software Architecture of Ubiquitous Scientific Comput CORBA-ZoltnNochta Taufiq (2001) www.cooperation ing. Tzvetan Drashansky (1995) www.cs.purdue.edu/ management.de/publikationen/paper/pwc2001 nrr cr.pdf homes/saw/publications/95/ubiq-pses.ps The adaptation and use of a WWW-based coursemanagement 35 system. De Boer, Collis (2000) www.ub.utwente.nl/ Wavefront implementation of Self Organizing Maps on webdocs/ctit/1/00000059.pdf RENNS Gaute Myklebust (1995) www.idt.unit.no/-ga Evaluation of Mobile Agent Systems with Respect to Fail utemyk/icasp95.ps ure. Otto Wittner October www.item.ntinu.no/-ottow/pa On the Enterprise Modelling of an Educational Informa pers/failsemMASreport.pdf tion. www.ub.utwente.nl/webdocs/ctit/1/0000002fpdf 40 Proactive Services in a Distributed Traffic Telematics.— VBR Video Source Characterization And A Practical.—Cse Gura, Held, Kaiser www.informatik.uni-ulm.de/rs/pro lenyi, Molnár hsnlab.ttt.bme.hu/-molnar/files/VBRtel jekte/core/ProctiveServ.pdf SyS. pS.gZ TCP over GPRS Performance Analysis Manner (1999) Use of the CANTOR system for collaborative learning in.— www.cs.helsinki.fi/u/manner/papers/Thesis-Manner-ps Hans Andersen Verner newmedia.colorado.edu/cscl/ 45 Ip Over Wavelength-Routed Access Networks—Marcos 270.pdf Rogrio Salvador wwwctit.cs.utwente.nl/~salvador/ Jgroup/ARM: A Distributed Object Group Platform. Mel Eunice99.pdf ing, Montresor. www.CS.UniBO.it/babaoglu/papers/ Increasing Retrievability and Reusability of Learning.—Hid group-arm.pdf dink Van Der www.ub.utwente.nl/webdocs/ctit/1/ Validation of the Open Service Access API for UMTS.– 50 00000056.pdf Maarten Wegdam Dirk-Jaap arch.cs.utwente.nl/publica Experiences from Development of Home Health Care Appli tions/papers/proms01-osa-lincs22130210.pdf cations.—Leili Lind Erik ftp.limt.liu.se/pub/bildb/MIpa pers/524 LIND.PDF A Hyperlink-Proposal Mechanism to Exemplify Cogni A Review of Parallel Implementations of Backpropaga tive.—Haffner, Roth, Meinel www.informatik.uni 55 tion. Torresen, Tomita www.ifi.uio.no/-jimtoer/chp2.ps trier.de/- meinel/papers/Paper02.ps ESCORT: Towards Integration in Intersection Control—An Digital Signatures For Automobiles?—Gollan, Meinel drea Savigni Filippo (2000) www.cs.ucl.ac.uk/staff/ www.informatik.uni-trier.de/-meinel/papers/ A.Savigni/papers/2000 jubilee escort roma.pdf DigitalSignatures.Auto02.pdf Usability Field-Test Of A Spoken Data-Entry System Mar Designing Safe Smart Home Systems for Vulnerable 60 cello Federico And poseidon.itc.it:7117/~ssi/DITELO/pa People—Dewsbury, Taylor www.smartthinkin pers/ICASSP99 1ps g.ukideas.com/ DIRC.pdf FLAMINGO: A Packet-switched IP over WDM Metro Opti Providing X.509-based user Access Control To ftp.poli cal Network Dey Koonen Geuzebroek (2001) www.c- to.it/pub/security/papers/sec.98/sec98.ps.gZ tit.cs.utwente.nl/~salvador/NOC2001.pdf Seeing Speech. In Space And Time: Psychological And..— 65 Real-time test specification with TTCN-3—Dai, Grabowski, Ruth Campbell Department www.asel.udel.edu/icslp/ Neukirchen (2001) www.itm.mu-luebeck.de/publications/ cdrom/vol3/1008/a1008.pdf FBT2001/Abstract fbt01neukdai.pdf US 8,874,477 B2 133 134 A System for Uniform and Multilingual Access to Struc Supporting IP Dense Mode Multicast Routing Protocols tured. Xu, Netter, Stenzhorn (2000) www.cs.ust.hk/ in. Marcos Rogerio Salvador (2000) wwwctit.cs.utwen acl2000/Demo/04 Xu.pdf te.nl/-salvador/OPTICOMM2000.pdf Corpus-driven learning of Event Recognition Rules Rob Report on the course for Technology Teachers WWW Course erto Basii Maria www.dcs.shef.ac.uk/-fabio/ML4IE/ of Jyrki Pulkkinen And telematics.ex.ac.uk/T3/0/down 2. PS.gZ loads/d13-1.pdf Web-Support for Activating Use of Theory in Group-Based A Multi-DSP Laboratory Course Rinner, Schneider, Ste Learning Jan Van Der www.ub.utwente.nl/webdocs/ctit/ ger, Weiss www.iti.tu-graZ.ac.at/de?people/schneider/pa 1/0000005a.pdf pers/rinner98.pdf Automated Generation of Category-Specific Thesauri for.— 10 Modularity—A Concept For New Neural Network Architec Attardi, DiMarco. (1998) faure.iei.pi.cnrit/-fabrizio/Pub tures—Schmidt, Bandar (1998) www.comp.lancs.ac.uk/~ lications/TR0698.ps albrecht/pubs/pdf/schmidt csa irbid 1998.pdf The use of CMC in applied social science training Interim Supporting IP Dense Mode Multicast Routing in All-Opti Report Merja www.stir.ac.uk/schema/deliverables/ cal. Marcos Rogrio Salvador (2001) wwwctit.cs.utwen D5.3.pdf 15 te.nl/-salvador/ONM2001.pdf The IT-Potential Of Haptics Touch access for people Dagstuhl Seminar on Ubiquitous Computing September with. Sjöström (1999) www.certec.lth.se/doc/touchac The International (2001) www.inf ethz.ch/vs/events/ cess/TouchAccess.pdf dag2001/intro/DagstuhlIntroductions.pdf An All-Optical WDM Packet-Switched Network Architec A Framework For Video Modelling Centre For Telematics ture. Marcos Rogrio Salvador (2001) wwwctit.cs.utwen www.cs.utwente.nl/~milan/docs/innsb.ps te.nl/-salvador/ICN2001.pdf Conceptual Stage in Designing Multimedia for Tele Learn An Adaptive, Collaborative Environment to Develop. Viz ing Kommers (2001) www.ub.utwente.nl/webdocs/ctit/ cano. oreto.inf-cr.uclm.es/personas/avizcainofitsenV.ps 1/00000060.pdf The HyperMuseum Theme Generator System: Ontology 25 Signed Preservation Of Online References—Heuer, Lose based. Stuer, Meersman, De. (2001) www.archi mann, Meinel www.informatik.uni-trier.de/-meinel/pa muse.com/mw2001/papers/stuer/stuer.html wise.Vu pers/webnet00b.ps b.ac.be/Download/Papers/stuerMW2001.pdf Chapter 7 Implementation of Backpropagation Neural. Jim Failure Semantics of Mobile Agent Systems Involved in.— Torresen Department www.ifi.uio.no/-jimtoer/chp3.ps Otto Wittner Carsten (1999) www.item.ntinu.no/-ottow/ 30 TIMeat a glance Braek, Gorman, Haugen, Melby. www.s- papers/failsemNIK99.pdf intef.no/time/report.pdf Executive Summary The Vision Of www.it.bton.ac.uk/re Cross-Entropy Guided Mobile Agents Finding Cyclic Paths search/seake/knowledge.pdf in. Wittner, Helvik (2002) www.item.ntinu.no/-wittner/ Supporting the Travelling Tradition: A report on the work.— aamas2002 submitted.pdf 35 Ken Marks Department (2000) ui4all.ics.forth.gr/ Cross Entropy Guided Ant-like Agents Finding Depend i3SD2000/MarkS.PDF able. Wittner, Helvik (2002) www.item.ntinu.no/-witt Circuits and Systems—Benini, De Micheli, Macii, Maloberti ner/cec2002.pdf www.nd.edu/~stjoseph/newscas/CASMagvollnol.pdf Using Information Flow Control to Evaluate Access. Man Enterprise Modelling For An Educational Information. Ing tel, Schairer. www.dfki.de/~schairer/publications/ 40 Widya Cees (2001) www.home.cs.utwente.nl/-widya/ report00159.ps.gZ webpapers/iceis2001. 205.pdf Network Architecture of a Packet-switched WDM LAN/ Telematics For Group-Based Learning: Simplicity Versus.— MAN Dey Koonen And (2000) wwwctit.cs.utwente.nl/~ van der Veen, Collis. www.ub.utwente.nl/webdocs/ctit/1/ salvador/LEOS2000.pdf 00000057.pdf Specification and Validation of a Real-Time Parallel.—de 45 Research Report 1997-1999. Department Of Computer Farias, Pires. (1997) www.ub.utwente.nl/webdocs/ctit/1/ (1997) www.cs.ucy.ac.cy/Research/archives/rr97-99.ps 00000066.pdf Translation Resources, Merging Strategies and Relevance.— Mobile Ip: Security Application Tuquerres, Salvador, Djoerd Hiemstra Wessel janus.cs.utwente.nl/~hiemstra/ Sprenkels intrg.cs.tcd.ie/htewari/papers/MobileIP-Sec.pdf papers/clefl.pdf M3POC: a multimedia multicast transport protocol for coop 50 An Information System for Long-distance Cooperation in erative.—Owezarski www.laas.fr/-owe/PUBLIS/ Medicine. Kosch, al. (2000) www.ii.uib.no/para2000/ 99525.ps.gZ program/jacek.ps User Interfaces for All Kobsa, (eds.) (1999) www.gmd.de/ Next Generation Internet in Europe Published By The publications/report/0074/Text.pdf www.infowin.org/ACTS/ANALYSYS/PRODUCTS/ Supporting PIM-SM in All-Optical Lambda-Switched Net 55 THEMATIC/NGI/ngi in europe.pdf works—Marcos Rogrio Salvador (2001) wwwctit.cs.ut R. Mu-noz, M. Saiz-Noeda, A. Suarez and M. Palomar— wente.nl/-salvador/SBRC2001.pdf Grupo De Investigaci (2000) gplsi.dlsi.ua.es/gplsi/articu Managing Distributed Personal Firewalls with Smart.— los/a2000/mt2000.ps Haffner, Roth, Heuer. www.informatik.uni-trier.de/- Lazy Users and Automatic Video Retrieval Tools in (the) meinel/papers/Managing01.ps 60 Lowlands—The Lowlands Team carol. wins.uva.nl/~cgm FINAL REPORT: LAURIN laurin.uibk.ac.at/ Version snoek/pub/trec 10video.pdf November Author germanistik.uibk.ac.at/laurin/reports/ Remote MIB item look-up service Pras, Boros, Helthuis finalrep01.pdf (2002) www.simpleweb.org/nm/research/results/publica Some Implications of MSC, SDL and TTCN Time Exten tions/pras/2002-04-noms.pdf sions.—Hogrefe, Koch. (2001) www.itm.mu-luebeck.de/ 65 Use Of Real And Contaminated Speech For Training Of publications/DH BK HN 2001 SDLForum/ A.—Matassoni Omologo And (2001) poseidon.itc.it: sdlforum2001.pdf 7117/~ssi/SHINE/ps/eurospeech01.ps.gZ US 8,874,477 B2 135 136 Extending the Data Storage Capabilities of a Java-based.— On Burst And Correlation Structure of Teletraffic Models— Clemens Cap Nico wwwiuk.informatik.uni-rostock.de/ Molnár, Miklós hsnlab.ttt.bme.hu/~molnar/files/ ~maibaum/docs/maibtune.ps ilkley97.ps.gZ Towards Precision Tools For ATM Network Design, Dimen Scaling Analysis of IP Traffic Components—Molnar, Dang sioning. Molnár, al. hsnlab.ttt.bme.hu/-molnar/files/ (2000) hsnlab.ttt.bme.hu/~molnar/files/itcssip00.ps.gZ peripol.ps.gZ A System for Supporting Cross-Lingual Information Fair Bandwidth Allocation of a Wireless Base Station Gy Retrieval Capstick, al. (1999) speech.ftw.at/ -gor/pub/ orgy Miklos Traffic hsnlab.ttt.bme.hu/-molnar/files/ ipm/mulinex-ipm99.pdf iqwim99.ps.gZ Spatiotemporal Segmentation and Tracking of Objects for.— Forecasting the Success of Telecommunication Services 10 Kompatsiaris, Strintzis (2000) egnatia.ee.auth.gr/-ikom/ in. Detlef Schoder. (2000) www.iig.uni-freiburg.de/ CSVT2000.pdf telematik/forschung/publikationen/pubfiles/Sc2000.pdf Region-Based Color Image Indexing And Retrieval—Ioannis A General Fractal Model of Internet Traffic—Molnar hsin Kompatsiaris Evagelia (2001) egnatia.ee.auth.gr/-ikom/ lab.ttt.bme.hu/~molnar/files/multifractal LCN.pdf 15 icip2001.pdf Correlations in ATM Cell Streams Exposed to Cell Delay Kompatsiaris—And Michael Strintzis egnatia.ee.auth.gr/~ Variation Molnár, Blaabjerghsnlab.ttt.bme.hu/~molnar/ ikom/ICIP00.pdf files/hung1.ps.gZ th IFLA Council and General Conference Aug. 16-25, A General Traffic Control Framework in ATM Networks— 2001—Code Number Division www.ifla.org/IV/iflao.7/ Fodor, Marosits, Molnár (1996) hsnlab.ttt.bme.hu/~mol papers/161-165e.pdf nar/files/gtfps.gZ Agent-Mediators In Media-On-Demand Eletronic Com NAVIGATION IN CYBERSPACE. Using Multi-Dimen merce—Joo Paulo Andrade www.home.cs.utwente.nl/~ sional Scaling.—Schoder Institut Fur (1999) www.iig.uni guizzard/mod-amec-cuba.pdf freiburg.de/telematik/forschung/publikationen/pubfiles/ A Web-based Distributed Environment to Support Teleteach Sc1999a.pdf 25 ing.—Ch Bouras Computer ru6.cti.gr/Publications/ Performance Measurement Tool for Packet Forwarding 261.pdf Devices—Tam As Kovacshazy www.mit.bme.hu/-khazy/ Web-Enabled Distance Education Environment—Bouras, publications/imtc2001 3472.pdf Lampsas, Bazaios. (1998) ru6.cti.gr/Publications/296.pdf Benefits of a Universal Security Framework—Report By ID-No. of presentation: t3a01391—Authors Christos Bouras Arnd (2000) www.iig.uni-freiburg.de/telematik/fors 30 ru6.cti.gr/Publications/279.pdf chung/publikationen/pubfiles/We2000fpdf Usability meets Security—The Identity-Manager as your.— Methods for Computing B-ISDN Link Blocking Probabili Jendricke, Markotten (2000) www.acsac.org/2000/papers/ ties—Molnár, Blaabjerg hsnlab.ttt.bme.hu/~molnar/files/ 90.pdf link-ps.gZ Deployment Scenarios of DVEs in Education Bouras Com Inter-organizational Networking of Small and—Framework 35 puter Technology ru6.cti.gr/Publications/324.pdf And. (1999) www.iig.uni-freiburg.de/telematik/fors Trends: Training Educators Through Networks And...—Chris chung/publikationen/pubfiles/EgEn1999.pdf tos Bouras Computer (1996) ru6.cti.gr/Publications/ Using Objects and Patterns to Implement Domain Ontolo 439.pdf gies—Guizzardi, Falbo, Filho (2001) www.home.cs.ut 40 On-Demand Hypermedia/Multimedia Service over. Bou wente.nl/~guiZZard/bcs.pdf ras, Kapoulas. (1996) ru6.cti.gr/Publications/358.pdf Highly Secure Low-cost Computers—Arnd Weber Today's Tele-working services from the Greek PTT Christos Bou (2000) www.iig.uni-freiburg.de/telematik/forschung/pub ras Vaggelis (1999) ru6.cti.gr/Publications/322.pdf likationen/pubfiles/We2000e.pdf HIPPOCRATES: A Tool for Distance Education Bouras On Modeling and Shaping Self-Similar ATM Traffic—Andor 45 Fotakis Kapoulas ru6.cti.gr/Publications/305.pdf Molnar Andhsnlab.ttt.bme.hu/-molnar/files/itc97.ps.gZ A Platform for the Implementation of the Services.—Bouras, Pitfalls in Long Range Dependence Testing and Estimation— Gkamas. (1998) ru6.cti.gr/Publications/259.pdf Molnar, Danghsnlab.ttt.bme.hu/-molnar/files/pitfalls.pd Training Centres: An Architecture for the Realisation.— fgZ Christos Bouras Computer ru6.cti.gr/Publications/292.pdf Distributed Fair Bandwidth Allocation of a Wireless Base.— 50 In-Service Training through ODL Environments: From Gyorgy Miklos Traffic (2000) hsnlab.ttt.bme.hu/-molnar/ User.—Bouras Lampsas Spirakis ru6.cti.gr/Publications/ files/netwo0.ps.gZ 275.pdf Highly Secure Low-cost PDA-phones Weber (2000) Distributed Learning Environment using Advanced Ser www.iig.uni-freiburg.de/telematik/forschung/publika vices. Ch Bouras Computer (1999) ru6.cti.gr/Publica tionen/pubfiles/We2000d.pdf 55 tions/453.pdf Sharing Telematics Courses. The CANDLE project—Aiko Routing Management Application Based On Mobile Agents Pras Centre (2001) www.simpleweb.org/nm/research/re On.—Anglica Reyes Ernesto www.tgs.cs.utwente.nl/eu Sults/publications/pras/2001-09-04-eunice.pdf nice/summerSchool/papers/paper9-2.pdf A Prototype for an Agent-based Secure Electronic.—Pado Quality Of Service Monitoring In Ip Networks. By Using.— van, Sackmann. (2001) www.iig.uni-freiburg.de/telema 60 Tams Varga Andrs www.tgs.cs.utwente.nl/eunice/summer tik/forschung/projekte?e sicherheit/comet/publikationen/ School/papers/paper4-3.pdf PaSaEyPi2001.pdf Issues on QoS based Routing in the Integrated Services.— On Measurements of Multimedia Traffic in ATM Networks— Gbor Rtvri Department wwwtgs.cs.utwente.nl/eunice/ Cselényi. hsnlab.ttt.bme.hu/~molnar/files/icomt.ps.gZ SummerSchool/papers/paper4-1.pdf Advanced Generation Tool of Application’s Network Traf 65 Usability Research in a Housing Fair: Problems and. Sa fic—Petroczi, Molnar hsnlab.ttt.bme.hu/~molnar/files/ janiemi, Tossavainen (1995) csjoensuu.fi/pub/Reports/A- agentant.ps.gZ 1995-7ps.gZ US 8,874,477 B2 137 138 Tele-Education/-Co-Operation Pilot Pilot Study Plan Verifying a Smart Design of TCAP Arts, van Langevelde www.cg...its.tudelft.nl/~charles/publications/ (1999) www.cwi.nl/ftp/CWIreports/SEN/SEN MESH Report D213.pdf R9910.ps.Z. CAVE Speaker verification in bank and telecom services— Securing Multimedia Applications over B-ISDN Jordi Forn Lindberg, Blomberg, Melin ftp.ling.umu.se/pub/phonum/ 5 Mels www-mat.upc.es/-forme/if PROMS96.pdf phonum4/65.ps Teaching and learning with the WWW in the undergradu The Catallaxy as a new Paradigm for the Design of— ate.—Oliver, Omari, Cowan elrond. Scam.ecu.edu.au/ol Eymann, Padovan, Schoder (2000) www.iig.uni iver/docs/96/AUSWEB2d.pdf freiburg.de/telematik/forschung/publikationen/pubfiles/ Toward a Standard for Guideline Representation: an-Do EyPaSc2000.pdf 10 Internet Accounting Pras, van Beijnum, Sprenkels. (2001) menico Pisanelli Aldo Saussure.irmkant.rm.cnrit?onto/ www.simpleweb.org/nm/research/results/publications/ publ/amia99/amia99.pdf pras/internet-accounting.pdf Time Domain MLE of the Parameters of FBM Traffic EWI Search and User Interface Functions—Ardó, Cao, Lun Vidács, Virtamo (1999) keskus.tct.hut.fi/tutkimus/com2/ dberg, Roslund. www.lub.lu.se/combine/docs/ 15 publ/fbm2.pdf D34 search uips Artificial Coordination—Simulating Organizational.— Algebras and Automata for Timed and Stochastic Systems— Eymann, Padovan, Schoder (1998) www.iig.uni D'Argenio www.home.cs.utwente.nl/~dargenio/disserta freiburg.de/-padovan/publications/cefes98.pdf tion/dissertation.ps.gZ REMOT A Project to Remotely Monitor and Control Sci Characterizing Video Coding Computing in Conference Sys entific—Experiments Pucillo Oat www.aps...anl.gov/ tems—By Tuquerres Tuquerre www.ub.utwente.nl/web icalepcs97/paper97/p235.pdf docs/ctit/1/0000004d.pdf Performability Analysis of Markov Reward Models with Rate A Model to Evaluate Certificate Revocation Forn Castro and.—Andor Acz And (1999) webspin.hit.bme.hu/-telek/ Department (2000) www-mat.upc.es/-forme/jf cikkek/racz99aps.gZ SCI2000 1.pdf 25 Analysis of the Completion Time of Markov Reward Mod Dublin Bus Tracking Service Design and implementation els. Miklos Telek Andras webspin.hit.bme.hu/-telek/ of a.—Fallon (2000) ftp.cs.tcd.ie/pub/tech-reports/re cikkek/tele98b.ps.gZ ports.00/TCD-CS-2000-47.pdf A New Method for Spectral Shaping Coding Peter Amos Integrating Different Strategies for Cross-Language.— Aszlo webspin.hit.bme.hu/-telek/cikkek/vamo98a.ps.gZ Buitelaar, Netter, Xu www.dfki.de/lt/mietta/mietta-twilt.ps 30 Integrating Trading and Load Balancing for Efficient. Thi MRMSolve: A Tool for Transient Analysis of Large Ben, Neukirchen (2000) www.itmmu-luebeck.de/publica Markov. Rácz, Tóth, Telek webspn.hit.bme.hu/-telek/ tions/DT HN 2000ITal Bf MoSiDS/USM2000.pdf cikkek/racz00fps.gZ Junction Point Aspect: A Solution to Simplify Implementa Managing Services in Distributed Systems by Integrating.— tion of Berger (2000) micado.project.free.fr/Publi/ 35 Thien, Neukirchen (2000) www.itm.mu-luebeck.de/pub ecoop2000.ps.gZ lications/ISCC2000/ISCC2000.pdfgZ Csaba Antal Jzsef Molnir Sindor www.cs.kau.se/~soren/dk Conformance Testing with TTCN Schieferdecker, doc/documentstore/cc fp.ps Grabowski (2000) www.itm.mu-luebeck.de/publications/ GRAVE Cave General Video Client Wasskog (1995) I J=2000 CTWT/Telektronnikka 2000 CTandTTCN. www.idi.ntinu.no/grupper/db/report diplomas/il myggo/ 40 pdf diplom.ps.Z Analysis and Modelling of Collaborative Learning Interac Z39.50 Application programmer's Interface SYSTEM.— tions—Workshop Notes (2000) collide.informatik.uni-du Document No Document ftp.ddb.de/pub/dbvosi/ isburg.de/-martin/publication/Muehlenbrock-ECAI ses V3.ps.gZ 2000.pdf Hardware Implementation of a Secure Bridge in Ethernet.— 45 Aligning IT and Organization in the MediaSite project— Forn Soriano Mels www-mat.upc.es/-forme/ Iacucci, Axelsson (2000) iris23.htu.se/proceedings/PDF/ if GLOBECOM93.pdf 20final...PDF Performance Evaluation of Strategies for Integration.— Enriching Textual Documents with Time-codes from Queija, van den Berg. (1999) www.cwi.nl/ftp/CWIreports/ Video. van der Sluis, de Jong (2000) 133.23.229.11/- PNAVPNA-R9903.ps.Z. 50 ySuzuki/Proceedingsall/RIAO2000/Wednesday/37CP1.ps Inverse Multiplexing for ATM. Operation.—Aguilar-Igartua. Modular Automated Transport Frequently Asked Ques (1999) marley.upc.es/pub/articles/icatm99.pdf tions (FAQ)—Schweizer (2000) circwww.epfl.ch/staff jo An integrated Solution for secure communications over erg/mat/docs/facq/faqps B-ISDN. Forn Mels Department www-mat.upc.es/- The CIMI Profile Release 1.OH Profile For Cultural jforme/if CMS96.pdf 55 www.cimi.org/public docs/HarmonizedProfile/ Web Representation with Dynamic Thumbnail Schmid CIMIProfile10H.pdf www.comp.lancs.ac.uk/computing/users/SSchmid/Yu Real-Time Traffic Simulation of the German Autobahn Net foric/YuforicExtAbstrps work Rickert, Wagner, Gawron (1996) www.zpr.uni Distributed educational multimedia databases: design, pro koeln.de/-mr/documents/PASA 96.ps.gZ duction. Hiddink (1998) www.home.ctitutwente.nl/ 60 Analysis of a Distributed Wireless Fair Scheduling Scheme— -hiddinkg/professional/papers/romy/romy.ps Miklós hsnlab.ttt.bme.hu/~molnar/files/itcssmob00.ps.gZ Sojourn Times in Non-Homogeneous QBD Processes with CAC Algorithm Based on Advanced Round Robin Method Processor. Queija (1999) www.cwi.nl/ftp/CWIreports/ for QoS. Marosits, Molnár hsnlab.ttt.bme.hu/~molnar/ PNAVPNA-R9901.ps.Z. files/iscc01.pdfigZ A guided tour through LGM-How to generate spoken.— 65 Link Capacity Sharing Between Guaranteed—and Best Krahmer, Landsbergen. www.ipo.tue.nl/homepages/ Effort. Rácz, Telek webspin.hit.bme.hu/-telek/cikkek/ ekrahmer/Pubs/lgmps racZ01aps.gZ US 8,874,477 B2 139 140 Quality of Service on the Internet: Evaluation of the. Elisa Needed Services For Network Performance Evaluation— bete Reis Elreis (2001) dragao.co.it.pt/conftele2001/proc/ Dung Dinh Luong gollum.ttt.bme.hu/~luong/tools/atmi pap101.pdf p.ps A MixDemonstrator for teaching Security in the Virtual.— Link Proposals with Case-Based Reasoning Techniques— Jendricke, Rannenberg www.scis.cowan.edu.au/research/ Haffner, Roth, Heuer. www.ti.fhg.de/conferences/ wise/WISEIProceedings/pdf/jendricke.pdf 200011171414080ps Fair Allocation Of Elastic Traffic For A Wireless Base Sta Remote Access to Medical Records via the Internet: Feasi tion Gyorgy Miklos Traffic (1999) hsnlab.ttt.bme.hu/~ bility. Pi Lees Ce (1999) www.ics.forth.gr/ICS/acti/ molnar/files/globe99.ps.gZ cmi hta/publications/papers/1999/cic99/lees cic99.pdf Cell Delay Variation in an ATM Multiplexer Molnár, 10 Partial Methods Versus End-to-End Measurements—Dung Blaabjerghsnlab.ttt.bme.hu/~molnar/files/cdvSnew.ps.gZ Dinh Luong gollum.ttt.bme.hu/~luong/tools/ifip.ps Supporting All Service Classes in ATM: A Novel. Marosits. (1999) hsnlab.ttt.bme.hu/~molnar/files/info99.ps.gZ The Role of Packet-dropping Mechanisms in QoS Differen tiation Goncalo Quadros Antonio (2000) www The Impact Of Long Range Dependence On Cell Loss In .dei.uc.pt/~boavida/papers/2000icon.pdf An. Vidacs, Molnar, Gordos (1998) hsnlab.ttt.bme.hu/~ 15 molnar/files/globe98.ps.gZ Component-Based Groupware Tailorability using Monitor Investigation of Fractal Properties in Data Traffic—Dinh, ing.—de Farias, Diakov (2000) amidst.ctitutwente.nl/ Molnár.hsnlab.ttt.bme.hu/~molnar/files/jc98.ps.gZ publications/cscw cbg2000.pdf Performance Evaluation of a General Traffic Control.—Ma Modeling of Time and Document Aging for Request.— rosits.hsnlab.ttt.bme.hu/-molnar/files/ipcccQ9.ps.gZ Haffner, Roth, Engel. (2000) www.ti.fhg.de/conferences/ The Demand for Stored Value Payment Instruments—Ingo 2000092.11730520-ps Pippow Detlef (2001) www.iig.uni-freiburg.de/telematik/ Mpeg-4 Authoring Tool For The Composition Of 3D.— forschung/projekte?e sicherheit/comet/publikationen/ Daras. (2001) egnatia.ee.auth.gr/-ikom/ISCAS2001.pdf PiScho2001.pdf Disambiguation Strategies for Cross-language Informa On The Effects Of Non-Stationarity In Long-Range Depen 25 tion. Djoerd Hiemstra And (1999) www.cs.utwente.nl/~ dence Tests Trang Dinh Andhsnlab.ttt.bme.hu/-molnar/ hiemstra/papers/ecd199.ps files/trendeffps.gZ the IFLA Council and General Conference Aug. 16-25, Bottlenecks on the Way Towards Fractal Characterization 2001—Code Number Division www.ifla.org/IV/iflao.7/ of—Andor Molnar Attila hsnlab.ttt.bme.hu/~molnar/ papers/161-165e.pdf files/pmccn97.ps.gZ 30 CANDLE: an European E-Education project to . . . —Bat Multimedia Databases in Education—Gerrit Hiddink Centre logg, al. (2000) www.ssgrrit/en/ssgrr2000/papers/164.pdf www.home.ctitutwente.nl/-hiddinkg/professional/pa State-dependent M/G/1 Type Queueing Analysis for.—Alt pers/dolls 97.ps man. (2000) www.cwi.nl/ftp/CWIreports/PNA/PNA Content-based video retrieval Petkovic (2000) R0005.ps.Z. www.edbt2000.uni-konstanz.de/pha-workshop/papers/ 35 Continuous Queries within an Architecture for Querying.— Petkovic.pdf Brinkhoff, Weitkämper (2001) www.fh-wilhelmshav Adaptive Optimisation of Importance Sampling for.—Hee en.de/oow/institute/iapg/personen/brinkhoff/paper/ gaard (1996) www.idi.ntinu.no/-poulh/publications/ SSTD2001.pdf nts13a.ps The Distribution And Partitioning Scheme Of The. Jimé Factors of reuse of Units of Learning Material Gerrit Hid 40 nez. ches.ing.ula.ve/INVESTIGACION/ARTICULOS/ dink Centre www.home.ctitutwente.nl/-hiddinkg/profes TANIA/SIM-71 ps.gZ sional/papers/reuse.ps An Architecture For Video On Demand Agent-Mediated.— Multilateral Security A concept and examples for balanced Almeida, Guizzardi. www.home.cs.utwente.nl/~guizzard/ security Rannenberg (2000) csirc.nist.gov/nissc/2000/ vod-amec-workcomp99.pdf proceedings/papers/202ra.pdf 45 Reusing Multi-Media Components: A Catalogue Implemen Educational Multimedia Databases: Past and Present Gerrit tation Steinmann, Shearer www.fernuni-hagen.de/DVT/ Hiddink Centre www.home.ctitutwente.nl/-hiddinkg/pro Publikationen/Papers/emmsec.pdf fessional/papers/systems.ps Multiagent Systems Instructor Prof Dh 136.159.122.221/ Resource-limited information retrieval in Web-based.— seminar/enmf619 02/enmfo2.pdf Daan Velthausz And (1997) www.trc.nl/publicaties/1997/ 50 Twenty-One at CLEF-2000: Translation resources, merg reslim/resource-limited.pdf ing.—Djoerd Hiemstra Wessel www.ieipi.cnr.it/DELOS/ CRL Supported a Smart redesign of a real-life protocol CLEF/twentyon.pdf Thomas Arts Email extranet.telin.nl/dscgi/ds.py/Get/File Papabiles Torday, Bierlaire (2001) rosowww.epfl.ch/mbi/ 8309/fmics99.ps.Z. strc-papabiles.pdf Convergence In The Digital Age Table Of Content ftp.cord 55 Using Objects and Patterns to Implement Domain Ontolo is.lu/pub/libraries/docs/proceedings.pdf gies—Guizzardi, Filho (2001) www.home.cs.utwente.nl/~ Using Automated Assistance Systems—Putting The Driver guizzard/SBES2001 Vf.pdf Into Focus—Reichardt (1998) www.daimler-benz.com/re Decision Support Systems from a Health Informatics Per search/events/pdf/IV980240.PDF spective Nykanen (2000) acta.uta.fi/pdf/951-44-4897 The Cave-Wp4 Generic Speaker Verification System—Ja 60 9.pdf boulet, KOOLWAAIJ. (1998) www.ubilab.org/publica NetTrouble: ATTS for Network Management—Lus Santos tions/print versions/pdf/jab98.pdf Pedro www.dei.uc.pt/-psimoes/papersfits98.pdf Field Test Of A Calling Card Service Based On.—den Os. Results and experience from the application of a common.— Boves. www.ubilab.org/publications/print versions/pdf/ Antonis Bouras. (1998) ru6.cti.gr/Publications/318.pdf den97.pdf 65 The Information and Communication Technologies. In Edu Distributed Electronic Commerce Systems—Sonja Zwil Er cation Christos Bouras Computer ru6.cti.gr/Publica ftp.cs.umass.edu/pub/net/pub/hgSchulz/i96/Zwisslerps.gZ tions/289.pdf US 8,874,477 B2 141 142 Internet Protocols for Synchronous Distance Learning Ch Long Cycles and Long Paths in the Kronecker Product of Bouras Computer (2000) ru6.cti.gr/Publications/431.pdf a. Jha, Agnihotri, al. (1995) www.cs.jhu.edu/~rajesh/ps/ The Euro in the Electronic Purse Allard, Alyankian, Ankri, cta.ps Collin. (2000) www.eurosmart.com/download/WhitePa Tutorial on Message Sequence Charts (MSC 96)—Rudolph, per.pdf Grabowski, Graubmann (1996) www.itm.mu-luebeck.de/ Monitoring Extensions for Component-Based Distributed.— publications/MSC96/dis-tutorial.ps.gZ Diakov, van Sinderen. (2000) amidst.ctitutwente.nl/pub The Standardization of Core INAP CS-2 by ETSI - lications/proms2000.pdf Grabowski, Hogrefe (1999) www.itm.mu-luebeck.de/pub The Q-bit Scheme Congestion Avoidance Using (1992) lications/CS2-Standardization.ps.gZ gatekeeper.dec.com/pub/doc/sigcomm/ccr/archive/1992/ 10 On The Design Of The New Testing Language TTCN-3— Grabowski, Wiles, Willcock. (2000) www.itm.mu-lue apr92/qbit.ps.Z. beck.de/publications/New TTCN3/GrabowskiEtAll-pd Performance of a Parallel Transport Subsystem Implementa fgZ tion Torsten Braun Institute www.iam.unibe.ch/-braun/ omVR—A Safety Training System for a Virtual Refinery— lit/hpcs93.ps.gZ 15 Haller, Kurka, Volkert. www.gup.uni-linz.ac.at:8001/staff/ Towards Dynamic Composition of Hybrid Communication kurka/docs/ismcr99.pdf Services—Floch, Braek (2000) www.item.ntinu.no/-jacf/ Senior Online Telematics De Report cmc.dsv.su.se/sol/sol paper/smarnett2000.pdf transfer-spec.pdf Optimising the Operation of the World WideWeb in. Had Formal Methods and Conformance Testing or What are jiefthymiades. www.cs.auc.dk/~tryfona/papers/cacher we.—Heymer, Grabowski www.itm.mu-luebeck.de/pub elps lications/FBT98/FBT98.ps.gZ Generating Test Cases for Infinite System Specifications— A Theorem Prover-Based Analysis Tool for Object-Oriented Stefan Heymer And www.itm.mu-luebeck.de/publica Databases—Spelt, Even (1999) www.home.cs.utwen tions/GTCfISS/HeymerGrabowskips.gZ te.nl/~ seven/CTIT-TR-98-22.pdf Computational Perspectives on Discourse and Dialogue— 25 Chemistry in Action: Discovering the Behaviour of a Bonnie Lynn Webber www.dai.ed.ac.uk/daidb/people/ Network.-Heymer, Grabowski (1998) www.itm.mu-lue homesbonnie/handbook.ps.gZ beck.de/publications/A-98-18/Report-A-98-18.ps.gZ Virtual Universities—Ebner, Hogrefe (1999) www.itm.mu ERP in the e-commerce era—Luttighuis, Biemans luebeck.de/publications/VFH/ extranet.telin.nl/dscgi/dspy/Get/File-2092/baanUSP.pdf ebner hogrefe waki'99.ps.gZ 30 Business-Driven Design of Transaction Services—Biemans, Compensation methods to support generic graph editing: A Janssen, Luttighuis. (1999) extranet.telin.nl/dscgi/ds.py/ case. Even, Spelt www.home.cs.utwente.nl/-seven/ Get/File-664/ICE.pdf ECOOPWS.pdf Modelling organisations Wetering (1999) extranet.telin.nl/ Towards the Generation of Distributed Test Cases Using dscgi/dspy/Get/File-2902/modellingV2.pdf Petri.-Heymer, Grabowski www.itm.mu-luebeck.de/pub 35 On Wrapping Query Languages and Ecient XML Integra lications/FBT99/fbt 99.ps.gZ tion Vassilis Christophides Sophie (2000) www.oasis Test Case Specification with Real-Time TTCN- Walter, open.org/cover/vassilisOuery Wrap.pdf Grabowski www.itm.mu-luebeck.de/publications/TC MESH Release 2 implementation at CTIT Diakov, van Sin SwRTTTCN/WalterGrabowskips.gZ deren, Koprinkov amidst.ctitutwente.nl/publications/ Scientific Approaches and Techniques for Negotiation.— 40 ctit tr99-08.pdf Gerding, van Bragt. www.cwi.nl/projects/TA/reports/ne EUROgatherer: a Personalised Gathering and Delivery.— gotiation.ps Amato, Straccia, Thanos (2000) faure.iei.pi.cnrit/%7Es Towards an Integrated Test Methodology for Advanced.— traccia/download/papers/SCI2000/SCI2000.pdf Grabowski, Walter www.itm.mu-luebeck.de/publications/ Frameworks for protocol implementation Barbosa, Pires, tes.99.ps.gZ 45 van Sinderen (1998) www.home.cs.utwente.nl/~sinderen/ TTCN-3—A new Test Specification Language for Black-Box publications/sbrc98.pdf Testing.—Grabowski (2000) www.itm.mu-luebeck.de/ See also U.S. Pat. No. 4,706,086 (November, 1987 Panizza publications/ttcn3/Grabowski.pdf.gZ 340/902), and U.S. Pat. No. 5,428,544 (June, 1995 Shyu Test Architectures for Distributed Systems—State of the Art 701/117), U.S. Pat. No. 6,473,688 (Kohno, et al., Oct. 29, and. Walter (1998) www.itm.mu-luebeck.de/publica 50 2002, Traffic information transmitting system, traffic infor tions/IWTCS98TA/IWTCS98Testarchitectures.ps.gZ mation collecting and distributing system and traffic infor Asbru: A Task-Specific, Intention-Based, and Miksch, mation collecting and distributing method), U.S. Pat. No. Shahar, Johnson ftp.ifs.tuwien.ac.at/pub/publications/ 6,304,758 (October, 2001, Iierbiget al., 701/117): U.S. Pat. milk kem197.pdf No. 6,411,221 (January, 2002, Horber, 701/117): U.S. Pat. Protocol Specifications—Written By Jacob cmc.dsv.su.se/ 55 No. 6,384,739 (May, 2002, Robert, Jr., 701/117): U.S. Pat. select/SEL-prot-spec-V11-p-99.1009.pdf No. 6,401,027 (June, 2002, Xa et al., 701/117): U.S. Pat. Towards The Third Edition Of TTCN-Grabowski, Hogrefe No. 6,411,889 (June, 2002, Mizunuma et al., 701/117), (1999) www.itm.mu-luebeck.de/publications/ U.S. Pat. No. 6,359,571 (Endo, et al., Mar. 19, 2002, iwtcs.99.ps.gZ Broadcasting type information providing system and travel A Framework for the Specification of Test Cases for.— 60 environment information collecting device); U.S. Pat. No. Walter, Grabowski (1999) www.itm.mu-luebeck.de/publi 6.338,011 (Furst, et al., Jan. 8, 2002, Method and apparatus cations/Walter-Crabowski-JIST99/ist.ps.gZ for sharing vehicle telemetry data among a plurality of Verification of Compensation Requirements for the users over a communications network); U.S. Pat. No. SEPIA. Even, Spelt (1998) www.home.cs.utwente.nl/~ 5,131,020 (July, 1992, Liebesny et al., 455/422); U.S. Pat. seven/CTIT-TR-98-25.pdf 65 No. 5,164,904 (November, 1992, Sumner, 701/117): U.S. Cote de Resyste Conformance Testing Reactive www Pat. No. 5,539,645 (July, 1996, Mandhyan et al., 701/119): tios.cs.utwente.nl/Docs/projects/cote-de-resyste/stw.ps U.S. Pat. No. 5,594,779 (January, 1997, Goodman, 455/4): US 8,874,477 B2 143 144 U.S. Pat. No. 5,689.252 (November, 1997, Ayanoglu et al., www.inference.phy.cam.ac.uk/mackay/itprinn/book.html 340/991): U.S. Pat. No. 5,699,056 (December, 1997, www.hh..se/staff/nicholas/NN Links.html Yoshida, 340/905); U.S. Pat. No. 5,864.305 (January, Xpidea.com/products/neuroVcl/neuroabout.htm 1999, Rosenquist, 340/905); U.S. Pat. No. 5,889,473 www.msm.ele.tue.nl/research/neural/ (March, 1999, Wicks, 340/825); U.S. Pat. No. 5,919,246 homepages.goldsmiths.ac.uk/nikolaev/Nnets.htm (July, 1999, Waizmann et al., 701/209); U.S. Pat. No. www.triumf.ca/project ETA/neural network.html 5,982.298 (November, 1999, Lappenbusch et al., personal.bgsu.edu/~Suny/nn.html 340/905); U.S. Pat. No. 4,860,216 (August, 1989, Linsen www.icmc.Sc.usp.br/~andre/ann links.html mayer, 342/159); U.S. Pat. No. 5,302,955 (April, 1994, www.stud.ntinu.nof-hirpa/linkS/AI links.htm Schutte et al., 342/59): U.S. Pat. No. 5,809.437 (Septem 10 it.umary.edu/Library/research/www Subjects/neural net ber, 1998, Breed, 701/29); U.S. Pat. No. 6,115,654 (Sep works.html tember, 2000, Eid et al., 701/34); U.S. Pat. No. 6,173,159 cindy.cis.nctu.edu.tw/NN/NN5/www.html (January, 2001, Wright et al., 455/66); and Japanese Patent www.public.iastate.edu/~acl/links/links.html Document Nos.JP 9-236650 (September, 1997); 10-84430 www.cs.cf.ac.uk/User/O.F.Rana/neural.html (March, 1998); 5-151496 (June, 1993); and 11-183184 15 (July, 1999), each of which is expressly incorporated www.cs.unredu/-bebis/CS791S/ herein by reference. See also: Martin E. Liggins, II, et al., www.geocities.com/fastiland/NNwww.html “Distributed Fusion Architectures and Algorithms for Tar cns-web.bu.edu/pub/snorrason/bookmarks/neural.html get Tracking'. Proceedings of the IEEE, Vol. 85, No. 1, www.open.brain.riken.go.jp/-cao/index work.html (XP-002166088) January, 1997, pp. 95-106; D. M. Hos SVr-www.eng.cam.ac.uk/research/neural/other neural mer, “Data-Linked Associate Systems”, 1994 IEEE Inter net sites.html national Conference on Systems, Man, and Cybernetics. R. O. Duda and P. E. Hart. Pattern Classication and Scene Humans, Information and Technology (Cat. No. Analysis. Wiley, New York, 1973. 94CH3571-5), Proceedings of IEEE International Confer S. Ripley, “Pattern Recognition and Neural Networks', Sta ence on Systems, Man and Cybernetics, San Antonio, Tex., 25 tistics, 33, 1065-1076. Ripley, B. D., 1996: Pattern Recog vol. 3, (XP-002166089) (1994), pp. 2075-2079. nition and Neural Networks. Cambridge: University Neural Networks References Press., 1996, citeseerist.psu.edu/ www.inference.phy.cam.ac.uk/mackay/Bayes FAQ.html ripley96complements.html, and cited references: www-2.cs.cmu.edu/Groups/AI/html/faqs/ai/neural/faq.html Aha, D. W., Kibler, D. & Albert, M. K. (1991) Instance-based www.cs.stir.ac.uk/~lss/NNIntro/InvSlides.html 30 diryahoo.com/Science/Engineering/Electrical Engineer learning algorithms. Machine Learning 6(1), 37-66. ing/Neural Networks/ Ali, K. M. & Pazzani, M.J. (1995) On the link between error www.aist.go.jp/NIBH/-b0616/Links.html correlation and error reduction in decision tree ensembles. www.creative.net.au/~adrian/mirrors/neural/ Technical Report 95-38, Department of Information and www.fi.uib.no/Fysisk/Teori/NEURO/neurons.html 35 Computer Science, University of California at Irvine. aass.oru.se/~tdt/ann/faq/FAQ.html Almond, R. G. (1995) Graphical Belief Modeling. London: www.cis.hut.fi/-jari/research.html Chapman & Hall. ISBN 0-412-06661-0. Despite the date, www.eg3.com/WebID/elect/neur-net/blank/overview/a- this book was actually published in May 1996. Z.htm Angluin, D. & Valiant, L. G. (1979) Fast probabilistic algo directory.google.com/Top/Computers/Artificial Intelli 40 rithms for Hamiltonian circuits and matchings. Journal of gence/Neural Networks/ Computer and System Sciences 18, 155-193. directory.google.com/Top/Computers/Artificial Intelli Anthony, M. & Shawe-Taylor, J. (1993) A result of Vapnik gence/Neural Networks/FAQs, Help, and Tutorials/ with applications. Discrete Applied Mathematics 47, 207 dmoz.org/Computers/Artificial Intelligence/Neural Net 217. Erratum (1994) 52,211 (the proof of theorem 2.1 is works/ 45 corrected). dmoz.org/Computers/Artificial Intelligence/Neural Net Arbib, M.A. (ed.) (1995)The Handbook of Brain Theory and works/FAQs, Help, and Tutorials/ Neural Networks. Cambridge, Mass.: MIT Press. ISBN www.cs.cqub.ac.uk/~J.Campbell/my web/book/nn.html 0-262-01 148-4. www.cere.pa.cnrit/IDAschool/lectures/neural.html Atkeson, C. G. (1991) Using locally weighted regression for cognet.mit.edu/MITECS/Entry/jordan2 50 robot learning. In Proceedings of the IEEE Conference on www.fads.org/faqS/ai-faq/neural-nets/part1/preamble.html Robotics and Automation (Sacremento, Calif., 1991), pp. Zhanshou.hypermart.net/thesis.htm 958-963. IEEE PreSS. www.links999.net/hardware/neural.html Auer, P., Holte, R. C. & Maass, W. (1995) Theory and appli www-ra.informatik.uni-tuebingen.de/linkS/neuronal/wel cation of agnostic PAC-learning with Small decision trees. come e.html 55 In Proceedings of the Twelfth International Conference on www.cogneuro.ox.ac.uk/linkS/ann.html Machine Learning, pp. 21-29. San Francisco: Morgan faculty.cs.tamu.edu/choefresources/ Kaufmann. Also NeuroCOLT Technical Report NC-TR www.galaxy.com/galaxy/Engineering-and-Technology/ 96-034 (February 1996). Electrical-Engineering/Neural-Networks/ Bartlett, P. L. & Williamson, R. C. (1996) The VC dimension mu.dmt.ibaraki.ac.jp/yanai/neu/facq/ 60 and pseudodimension of two-layer neural networks with bubl.ac.uk/link/n/neuralnetworks.htm discrete inputs. Neural Computation 8(3), 625-628. www.webopedia.com/TERM/n/neural network.html Benveniste, A., Metivier, M. & Priouret, P. (1990) Adaptive www.ie.ncsu.edu/fangroup/neural.dir/indexneural.html Algorithms and Stochastic Approximations. Berlin: www.geneticprogramming.com/AI/nn.html Springer. www.cs.utk.edu/-yarkhan/neural networks.html 65 Bratko, I. & Muggleton, S. (1995) Applications of inductive www.physiol.ox.ac.uk/~ket/nn.html logic programming. Communications of the Association www.aaai.org/AITopics/html/neural.html for Computing Machinery 38(11), 65-70. US 8,874,477 B2 145 146 Breiman, L. (1994) Bagging predictors. Technical Report Fayyad, U. M. & Irani, K. B. (1993) Multi-interval discreti 421. Department of Statistics, University of California at Zation of continuous-valued attributes in decision tree gen Berkeley. eration. In Proceedings of the Thirteenth International Breiman, L. (1996a) Bagging predictors. Machine Learning. Joint Conference on Artificial Intelligence (Chambery, Breiman, L. (1996b) The heuristics of instability in model 5 France, 1993), pp. 1022-1027. San Francisco: Morgan Selection. Annals of Statistics. Kaufmann. Breiman, L. (1996c) Bias, variance and arcing classifiers. Freund, Y. (1990) Boosting a weak learning algorithm by Technical report, Department of Statistics, University of majority. In Proceedings of the Third Workshop on Com California at Berkeley. putational Learning Theory, pp. 202-216. Morgan Kauf Brodley, C. E. & Utgoff, P. E. (1995) Multivariate decision 10 trees. Machine Learning 19, 45-77. a. Buntine, W. & Niblett, T. (1992) A further comparison of Freund, Y. (1995) Boosting a weak learning algorithm by splitting rules for decision-tree induction. Machine Learn majority. Information and Computation 121(2), 256-285. ing 8, 75-86. Freund, Y. & Schapire, R. E. (1995) A decision-theoretic Burnell, L. & Horvitz, E. (1995) Structure and chance: Meld 15 generalization of on-line learning and an application to ing logic and probability for software debugging. Commu boosting. In Proceedings of the Second European Confer nications of the ACM 38(3), 31-41, 57. ence on Computational Learning Theory, pp. 23-37. Catlett, J. (1991) On changing continuous attributes into Springer. ordered discrete attributes. In Proceedings of the European Freund, Y. & Schapire, R. E. (1996a) Game theory, on-line Working Session on Learning EWSL-91, ed. Y. Kodrat prediction and boosting. In Proceedings of the Ninth off, pp. 164-178. Berlin: Springer. Annual Conference on Computational Learning Theory. Cesa-Bianchi, M., Freund, Y., Helmbold, D. P., Haussler, D., Freund, Y. & Schapire, R. E. (1996b) Experiments with a new Schapire, R. E. & Warmuth, M. K. (1993) How to use boosting algorithm. In Proceedings of the Thirteenth Inter expert advice. In Proceedings of the Twenth-Fifth ACM national Conference on Machine Learning. Symposium on the Theory of Computing (San Diego, 25 Friedman, J. H. (1994) Flexible metric nearest neighbor clas Calif., 1993), pp. 382-391. New York: ACM Press. sification. Technical report, Department of Statistics, Stan Cesa-Bianchi, M., Freund, Y., Helmbold, D. P., Haussler, D., ford University. Schapire, R. E. & Warmuth, M. K. (1996) How to use Fukunaga, K. & Flick, T. E. (1984) An optimal global nearest expert advice. Journal of the ACM. neighbor metric. IEEE Transactions on Pattern Analysis Cortes, C. & Vapnik, V. (1995) Support-vector networks. 30 and Machine Intelligence 6, 314-318. Reprinted in Dasa Machine Learning 20, 273–297. Cost, S. & Salzberg, S. (1993) A weighted nearest neighbor rathy (1991). algorithm for learning with symbolic features. Machine Fulton, T., Kasif, S. & Salzberg, S. (1995) Efficient algo Learning 10, 57-78. rithms for finding multi-way splits for decision trees. In Cover, T. M. (1968) Capacity problems for linear machines. 35 Proceedings of the Twelfth International Conference on In Pattern Recognition, ed. L. Kanal, pp. 283-289. Thomp Machine Learning, eds. A. Prieditis & S. Russell, pp. 244 SO. 251. San Francisco: Morgan Kaufmann. Craven, M. W. & Shavlik, J. W. (1996) Extracting tree-struc Fung, R. & Del Favarro, B. (1995) Applying Bayesian net tured representations of trained networks. In Touretzky et works to information retrival. Communications of the al. (1996), pp. 24-30. ISBN 0-262-20107-0. 40 ACM 38(3), 42-48, 57. Dasarathy, B.V. (ed.) (1991) Nearest Neighbor (NN) Norms: Hampson, S. E. & Volper, D.J. (1986) Linear function neu NN Pattern Classification Techniques. Los Alamitos, rons: structure and training. Biological Cybernetics. Calif.: IEEE Computer Society Press. Hastie, T. & Tibshirani, R. (1996a) Discriminant adaptive Dougherty, J., Kohavi, R. & Sahami, M. (1995) Supervised nearest neighbor classification and regression. In and unsupervised discretization of continuous features. In 45 Touretzky et al. (1996), pp. 409-415. ISBN 0-262-20107 Proceedings of the Twelfth International Conference on O. Machine Learning, eds. A. Prieditis & S. Russell, pp. 194 Hastie, T. & Tibshirani, R. (1996b) Discriminant adaptive 202. San Francisco: Morgan Kaufmann. nearest neighbor classification. IEEE Transactions on Pat Drucker, H. (1996) Fast decision tree ensembles for optical tern Analysis and Machine Intelligence 18, 607-618. character recognition. In Proceedings of the Fifth Annual 50 Heath, D., Kasif, S. & Salzberg, S. (1993) Learning oblique Symposium on Document Analysis and Information Retrival. decision trees. In Proceedings of the Thirteenth Interna Drucker, H. & Cortes, C. (1996) Boosting decision trees. In tional Joint Conference on Artificial Intelligence (Cham Touretzky et al. (1996), pp. 479-485. ISBN 0-262-20107 bery, France, 1993), pp. 1002-1007. San Francisco: Mor O. 55 gan Kaufmann. Drucker, H., Schapire, R. & Simard, P. (1993) Boosting per Heckerman, D. & Wellman, M.P. (1995) Bayesian networks. formance in neural networks. International Journal of Pat Communications of the ACM 38(3), 26-30. tern Recognition and Artificial Intelligence 7(4), 705-719. Heckerman, D., Breese, J. S. & Rommelse, K. (1995) Deci Drucker, H., Cortes, C., Jaeckel, L. D., LeCun, Y. & Vapnik, Sion-theoretic troubleshooting. Communications of the V. (1994) Boosting and other ensemble methods. Neural 60 ACM 38(3), 49-57. Computation 6(6), 1289-1301. Helmbold, D. P. & Schapire, R. E. (1995) Predicting nearly as Edwards, D. (1995) Introduction to Graphical Modelling. well as the best pruning of a decision tree. In Proceedings Springer. of the Eight Annual Conference on Computational Learn Elomaa, T. & Rousu, J. (1996) Finding optimal multi-splits ing Theory, pp. 61-68. New York: ACM Press. for numerical attributes in decision tree learning. Neuro 65 Helmbold, D. P. & Schapire, R. E. (1996) Predicting nearly as COLT Technical Report Series NC-TR-96-041, Depart well as the best pruning of a decision tree. Machine Learn ment of Computer Science, University of Helsinki. 1ng. US 8,874,477 B2 147 148 Ho, T. K. (1995) Random decision forests. In Proceedings of Murthy, S. K. Kasif, S. & Salzberg, S. (1994) A system for the Third International Conference on Document Analysis the induction of oblique decision trees. Journal of Artificial and Recognition, pp. 278-282. IEEE Computer Society Intelligence Research 2, 1-33. PreSS. Oliver, J. J. & Hand, D.J. (1995) On pruning and averaging Hochreiter, S. & Schmidhuber, J. (1995) Simplifying neural decision trees. In Machine Learning: Proceedings of the nets by discovering flat minima. In Tesauro et al. (1995), Twelfth International Conference, pp. 430-437. Morgan pp. 529-536. ISBN 0-262-20104-6. Kaufmann. Hochreiter, S. & Schmidhuber, J. (1996) Flat minima. Neural Parberry, I. (1994) Circuit Complexity and Neural Networks. Computation. Cambridge, Mass.: MIT Press. ISBN 0-262-16148-6. Holte, R. C. (1993) Very simple classification rules perform 10 Parrondo, J. M. R. & Van der Broeck, C. (1993) Vapnik well on most commonly used datasets. Machine Learning Chervonenkis bounds for generalization. J. Phys. A 26, 11, 63-91. 2211-2223. Hoffgen, K.-U. Simon, H.-U. & Van Horn, K. S. (1995) Przytula, K. W. & Prasanna, V. K. (1993) Parallel Digital Robust trainability of single neurons. Journal of Computer 15 Implementation of Neural Networks. Englewood Cliffs, and System Sciences 50(1), 114-125. N.J.: Prentice Hall. Karpinski, M. & Macintyre, A. (1995a) Bounding VC-dimen Quinlan, J. R. (1996a) Bagging, boosting, and C4.5. In Pro sion of neural networks: Progress and prospects. In Pro ceedings of the Fourteenth National Conference on Artifi ceedings of the Second European Conference on Compu cial Intelligence. Menlo Park, Calif.: AAAI Press. tational Learning Theory (Barcelona, Spain), ed. P. Quinlan, J. R. (1996b) Improved use of continuous attributes Vitanyi, number 904 in Lecture Notes in Artificial Intelli in C4.5. Journal of Artificial Intelligence Research 4, gence, pp. 337-341. Berlin: Springer. 77-90. Karpinski, M. & Macintyre, A. (1995b) Polynomial bounds Rachlin, J., Kasif, S., Salzberg, S. & Aha, D. (1994) Towards for VC dimension of sigmoidal neural networks. In Pro a better understanding of memory-based and Bayesian ceedings of the Twenty-Seventh Annual ACM Symposium 25 classifiers. In Proceedings of the Eleventh International on Theory of Computing (Las Vegas), pp. 200-208. ACM Conference on Machine Learning (New Brunswick, N.J.), PreSS. pp. 242-250. Koiran, P. & Sontag, E. D. (1996) Neural networks with Sakurai, A. (1993) Tighter bounds of the VC-dimension of quadratic VC dimension. In Touretzky et al. (1996), pp. three-layer networks. In Proceedings of the 1993 World 197-2O3. ISBN 0-262-2O107-O. 30 Congress on Neural Networks, volume 3, pp. 540-543. Kooperberg, C., Bose, S. & Stone, C. J. (1996) Polychoto Hillsdale, N.J.: Erlbaum. mous regression. Journal of the American Statistical Asso Schapire, R. E. (1990) The strength of weak learnability. ciation. Machine Learning 5(2), 197-227. Krogh, A. & Vedelsby, J. (1995) Neural network ensembles, Shafer, G. (1996) Probabilistic Expert Systems. Number 67 cross validation, and active learning. In Tesauro et al. 35 in CBMS-NSF Regional Conference Series in Applied (1995), pp. 231-238. ISBN 0-262-20104-6. Mathematics. Philadelphia, Pa.; SIAM. ISBN 0-89871 Langley, P. (1996) Elements of Machine Learning. San Fran 373-O. cisco: Morgan Kaufmann. Shlien, S. (1990) Multiple binary decision tree classifiers. Langley, P. & Simon, H. A. (1995) Applications of machine Pattern Recognition 23(7), 757-763. learning and rule induction. Communications of the ACM 40 Short, R. D. & Fukunaga, K. (1980) A new nearest neighbor 38(11), 54-64. distance measure. In Proceedings of the Fifth IEEE Inter Lauritzen, S.L. (1996) Graphical Models. Oxford: Clarendon national Conference on Pattern Recognition (Miami PreSS. ISBN 0-19-852219-3. Beach, 1980), pp. 81-86. Los Alamitos, Calif.: IEEE Com Ljung, L. Pflug, H. & Walk, H. (1992) Stochastic Approxi puter Society Press. mation and Optimization of Random Systems. Berlin: 45 Short, R. D. & Fukunaga, K. (1981) The optimal distance Birkhauser. measure for nearest neighbor classification. IEEE Trans Lowe, D.G. (1995) Similarity metric learning for a variable actions on Information Theory 27, 622-627. Reprinted in kernel classifier. Neural Computation 7(1), 72-85. Dasarathy (1991). Maass, W. & Tur'an, G. (1994) How fast cana thresholdgate Stone, C. J., Hansen, M., Kooperberg, C. & Truong, Y. K. learn? In Computational Learning Theory and Natural 50 (1997) Polynomial splines and their tensor products in Learning Systems: Constraints and Prospects, eds S. J. extended linear modeling. Annals of Statistics. Hanson, G. A. Drastal & R. L. Rivest, volume I, pp. 381 Tesauro, G., Touretzky, D. S. & Leen, T. K. (eds) (1995) 414. MIT PreSS. Advances in Neural Information Processing Systems 7. Marchand, M., Golea, M. & Rujan, P. (1990) A convergence Proceedings of the 1994 Conference. Cambridge, Mass.: theorm for sequential learning in two-layer perceptrons. 55 MIT PreSS. ISBN 0-262-2O104-6. Europhysics Letters 11, 487-492. Touretzky, D. S., Moser, M. C. & Hasselmo, M. E. (eds) Muroga, S. (1965) Lower bounds of the number of threshold (1996) Advances in Neural Information Processing Sys functions and a maximum weight. IEEE Transactions on tems 8. Proceedings of the 1995 Conference. Cambridge, Electronic Computers 14, 136-148. MaSS.: MIT PreSS. ISBN 0-262-2O107-0. Muroga, S., Toda, I. & Takasu, S. (1961) Theory of majority 60 Utgoff. P. E. (1989) Perceptron trees: a case study in hybrid decision elements. Journal of the Franklin Institute 271, concept representations. Connection Science 1(4), 377 376-418. 391. Murthy, S. K. Kasif, S., Salzberg, S. & Beigel, R. (1993) Valentin, D., Abdi, H. O'Toole, A. J. & Cottrell, G. (1994) OC1: randomized induction of oblique decision tress. In Connectionist models of face processing: A Survey. Pattern Proceedings of the Eleventh National Conference on Arti 65 Recognition 27, 1208-1230. ficial Intelligence (Washington, D.C., 1993), pp. 322-327. Vapnik, V. N. (1995) The Nature of Statistical Learning AAAI PreSS. ISBN 0-262-51071-5. Theory. New York: Springer. US 8,874,477 B2 149 150 Vapnik, V. N. (1996) Statistical Learning Theory. New York: G. Beylkin and N. Saito, “Multiresolution Representations Wiley. using the Autocorrelation Functions of Compactly Sup Wahba, G., Wang, Y. Gu, C., Klein, R. & Klein, B. (1996) ported Wavelets.” Smoothing spline anova for exponential families, with G. Beylkin and N. Saito, “Wavelets, their Autocorrelation application to the Wisconsin epidemiological study of dia 5 Functions and Multiresolution Analysis of Signals.” betic retinopathy. Annals of Statistics 23 (6), 1865-1895. G. Beylkin and B. Torresani, “Implementation of Operators Waltz, D. (1990) Memory-based reasoning. In Natural and via Filter Banks, Autocorrelation Shell and Hardy Wave Artificial Parallel Computation, eds M. Arbib & J. Robin lets. son, pp. 251-276. Cambridge, Mass.: MIT Press. A. G. Bruce, H. Gao and D. Ragozin, “Non-smooth Wavelets: Wang, L. & Oja, E. (1993) Image compression by MLP and 10 Graphing Functions Unbounded on Every Interval.” PCA neural networks. In Proceedings of the Eight Scandi C. Cabrelli, C. Heil, and U. Molter. Accuracy of Lattice navian Conference on Image Analysis (Tromso, Norway), Translates of Several Multidimensional Refinable Func pp. 1317-1324. tions.” Werbos, P. (1987) Learning how the world works. In Proceed R. C. Calderbank, I. Daubechies, W. Sweldens and B. Yeo ings of the IEEE Conference on Systems, Man and Cyber 15 “Wavelet Transforms that Map Integers to Integers.” There netics, pp. 320-310. New York: IEEE Press. is also an uncompressed version. Werbos, P.J. (1995) Backpropagation: Basics and new devel D. Chen, “Extended Families of Cardinal Spline Wavelets.” opments. pp. 134-139 of Arbib (1995). D. Chen, “Spline Wavelets of Small Support.” Willems, F. M. J., Shtarkov, Y. M. & Tjalkens, T. J. (1993) D. Chen, “Characterization of Biorthogonal Cardinal Spline Context tree weighting: A sequential universal source cod Wavelet Bases. ing procedure for FSMX sources. In Proceedings of the D. Chen, “Cardinal Spline Wavelets, dissertation. Also part 1993 IEEE International Symposium on Information 2, part 3, part 4, part 5 and part 6. Theory, p. 59. IEEE Press. S. Chen and D. Donoho, “Atomic Decomposition by Basis Willems, F. M.J., Shtarkov, Y. M. & Tjalkens, T.J. (1995) The Pursuit. context-tree weighting method: Basic properties. IEEE 25 A. Cohen, W. Dahmen and R. DeVore, “Multiscale Decom Transactions on Information Theory pp. 653-664. positions on Bounded Domains.” Wavelets References J. Cohen, “The Foot Problem in Wavelet Packet Splitting.” A Introductions to Wavelets Mathematica notebook converted to Postscript. G. Kaiser, “A Friendly Guide to Wavelets” J. Cohen, “Schauder Basis for 0.1. A Mathematica note C. S. Burrus and R. A. Gopinath, 'A Tutorial Overview of 30 book converted to Postscript. Wavelets, Filter Banks and Interrelationships' J. Cohen, “The Littlewood-Paley-Stein Wavelet.” A Math R. A. DeVore and B. J. Lucier, “Wavelets’ ematica notebook converted to PostScript. T. Edwards, “Discrete Wavelet Transforms: Theory and J. Cohen, “Battle-Lemarie Wavelets.” A Mathematica note Application.” book converted to Postscript. E. Gootman and M. Wickerhauser, “Elementary Wavelets.” 35 J. Cohen, “The Daubechies Minimum Phase Wavelets. A A. Graps, “An Introduction To Wavelets.” Mathematica notebook converted to Postscript. C. Heil and D. Walnut, “Continuous and Discrete Wavelet J. Cohen, “Meyer Wavelets.” A Mathematica notebook con Transforms.” verted to PostScript. B. Jawerth and W. Sweldens, “An Overview of Wavelet Based R. Coifman and M. V. Wickerhauser, “Entropy-Based Algo Multiresolution Analysis”. An abstract is also available. 40 rithms for Best Basis Selection' J. Lewalle, “Tutorial on Wavelet Analysis of Experimental R. Coifman and M. Wickerhauser, “Best-adapted Wave Data' Packet Bases. P. Schröder and W. Sweldens, “Building Your Own Wavelets R. Coifman, “Numerical Harmonic Analysis.” at Home.” An abstract is also available. R. Coifman, Y. Meyer and M. Wickerhauser, “Size Properties G. Strang, “Wavelets.” 45 of Wavelets Packets. G. Strang “Wavelets and dilation equations: a brief introduc R. Coifman and Y. Meyer, “Orthonormal Wave Packet tion. Bases C. Torrence and G. P. Compo, “A Practical Guide to Wavelet R. Coifman and M. V. Wickerhauser, “Wavelets and Adapted Analysis, with Significance and Confidence Testing.” Waveform Analysis’ B. Vidakovic, “Wavelets for Kids, also part 2. 50 R. Coifman, Y. Meyer and M. Wickerhauser, Adapted Wave E. Tolkova, “Orthogonal Wavelets Construction.” Form Analysis, Wavelet Packets and Applications.” Y. Meyer, Wavelets: Algorithms and Applications, Society for D. Colella and C. Heil, “Matrix Refinement Equations: Exist Industrial and Applied Mathematics, Philadelphia, 1993, ence and Accuracy.” pp. 13-31, 101-105. S. Dahlke, W. Dahmen, E. Schmitt and I. Weinreich, “Multi G. Kaiser, A Friendly Guide to Wavelets, Birkhauser, Boston, 55 resolution Analysis and Wavelets on S2 and S3.” 1994, pp. 44-45. W. Dahmen, “Stability of Multiscale Transformations.” General Theory W. Dahmen and C. A. Micchelli, “Biorthogonal Wavelet L. Andersson, N. Hall, B. Jawerth and G. Peters, “Wavelets on Expansions.” Closed Subsets of the Real Line' G. Davis, S. Mallat and Z. Zhang, "Adaptive Nonlinear P. Auscher, G. Weiss and M.V. Wickerhauser, “Local Sine and 60 Approximations.” Cosine Bases of Coifman and Meyer and the Construction G. Davis, Adaptive Nonlinear Approximations.” of Smooth Wavelets’ G. Davis, S. Mallat and Z.Zhang, Adaptive Time-Frequency C. Basdevant and V. Perrier, "Besov Norms in Terms of Con Approximations with Matching Pursuits.” tinuous Wavelet Transforms and Application to Structure I. Daubechies and W. Sweldens, An abstract is also available. Functions.” 65 C. deBoor, R. DeVore and R. Amos, “On the Construction of B. E. Bassil, G. J. Dickson and D. M. Monro, “Orthonormal Multivariate (Pre)Wavelets.” Wavelets With Balanced Uncertainty”. R. L. deGueiroz, “On Lapped Transforms.” US 8,874,477 B2 151 152 M. Girardi and W. Sweldens, “A New Class of Unbalanced J. Benedetto, C. Heil, and D. Walnut, “Differentiation and the Haar Wavelets That Forman Unconditional Basis for Lp on Balian-Low Theorem.” General Measure Spaces.” An abstract is also available. O. Christensen and C. Heil, “Perturbations of Banach Frames S. Haykin and S. Mann, “The Chirplet Transform: A New and Atomic Decompositions.” Signal Analysis Technique Based on Affine Relationships D. M. Healy, Jr. and S. Li, "On Pseudo Frame Decomposi in the Time-Frequency Plane.” This about 3.5 MB. tions and Discrete Gabor Expansions.” C. Heil, G. Strang and V. Strela, Approximation. By Trans S. Li, "General Frame Decompositions, Pseudo-Duals and lates of Refinable Functions.” Applications for Weyl-Heisenberg Frames.” C. Heil and G. Strang, “Continuity of the Joint Spectral S. Li, “On Dimension Invariance of Discrete Gabor Expan Radius: Application to Wavelets.” 10 sions.” B. Jawerth and W. Sweldens, “Biorthogonal Smooth Local M-Band Wavelets and Filter Banks Trigonometric Bases.” An abstract is also available. C. S. Burrus and R. A. Gopinath, “On the Correlation Struc B. Jawerth and W. Sweldens, “Weighted Multiwavelets on ture of Multiplicity M Scaling Functions' General Domains.” 15 C. S. Burrus and R. A. Gopinath, “Wavelets and Filter Banks' M. K. Kwong and P. T. Peter Tang, “W-Matrices, Nonor C. S. Burrus and R. A. Gopinath, “Unitary FIR Filter Banks thogonal Multiresolution Analysis and Finite Signals of and Symmetry’ Arbitrary Length.” C. S. Burrus and R. A. Gopinath, “Theory of Modulated Filter G. Leaf, J. M. Restrepo and G. Schlossnagle, “Periodized Banks and Modulated Wavelet Tight Frames' Daubechies Wavelets. C. S. Burrus and R. A. Gopinath, “Factorization Approach to J. Lippus, “Wavelet Coefficients of Functions of Generalized Time-Varying Unitary Filter Bank Trees and Wavelets' Lipschitz Classes.” C. Herley, "Boundary Filters for Finite-Length Signals and S. Mallat and Z. Zhang, “Matching Pursuit with Time-Fre Time-Varying Filter Banks.” quency Dictionaries.” P. Steffen, P. Heller, R. A. Gopinath and C. S. Burrus, “The E.J. McCoy, D. B. Percival and A. T. Walden, “On the Phase 25 Theory of Regular M-Band Wavelets” of Least-Asymmetric Scaling and Wavelet Filters.” Wavelets and General Signal Processing R. Piessens and W. Sweldens, “Wavelet Sampling Tech M. Vetterli and J. Kovacevic, “Wavelets and Subband Cod niques.” An abstract is also available. ing, Prentice Hall, 1995. J. Shen and G. Strang, “The Zeros of the Daubechies polyno D. E. Ashpis and J. Lewalle, “Transport in bypass transition: mials.” 30 mapping the active time scales using wavelet techniques' J. Shen and G. Strang, “Asymptotics of Daubechies Filters, Scaling Functions and Wavelets.” D. E. Ashpis and J. Lewalle, “Demonstration of wavelet tech M. J. Shensa, “An Inverse DWT for Nonorthogonal Wave niques in the spectral analysis of bypass transition data' lets C. Basdevant, V. Perrier and T. Philipovitch, “Wavelet Spectra W. Sweldens, “Compactly Supported Wavelets which are 35 Compared to Fourier Spectra.” Biorthogonal to a Weighted Inner Product.” An abstract is J. P. Bonnet, J. Lewalle and M. N. Glauser, "Coherent Struc also available. tures: Past, Present and Future.” W. Sweldens, “The Lifting Scheme: A Custom-Design Con G. Buresti, J. Lewalle & P. Petagna, “Wavelet statistics and struction of Biorthogonal Wavelets.” An abstract is also the near-field structure of coaxial jets” available. 40 C. S. Burrus and R. A. Gopinath, “Wavelet-Based Lowpass/ W. Sweldens, “The Lifting Scheme: A Construction of Sec Bandpass Interpolation” ond Generation Wavelets.” An abstract is also available. C. S. Burrus, R. A. Gopinath and J. E. Odegard, “Design of B. Suter and X. Xia, “Vector Valued Wavelets and Vector Linear Phase Cosine Modulated Filter Banks for Subband Filter Banks. Image Compression' C. Taswell, “Wavelet Transform Algorithms for Finite Dura 45 S. Cabrera, V. Krienovich and O. Sirisaengtaksin, “Wavelets tion Discrete-Time Signals.” Compress Better Than All Other Methods: A 1-D Theo C. Taswell, “Near-Best Basis Selection Algorithms with Non rem. Additive Information Cost Functions.” S. Cabrera, V. Krienovich and O. Sirisaengtaksin, “Wavelet K. Urban, “On Divergence-Free Wavelets.” Nerual Networks are Optimal Approximators for Func R. O. Wells Jr., “Recent Advances in Wavelet Technology” 50 tions of One Variable. M. V. Wickerhauser, “Entropy of a Vector Relative to a R. Carmona, W. L. Hwang and B. Torresani, “Characteriza Decomposition.” tion of Signals by the Ridges of Their Wavelet Trans M. V. Wickerhauser, “Lectures on Wavelet Packet Algo forms.” rithms R. Carmona, W. L. Hwang and B. Torresani, “Multi-Ridge M. V. Wickerhauser, “Smooth Localized Orthonormal Bases' 55 Detection and Time-Frequency Reconstruction.” C. Zarowski, “Notes on Orthogonal Wavelets and Wavelet R. Coifman, Adapted Multiresolution Analysis, Computa Packets’ tion, Signal Processing and Operator Theory’ V. Zavadsky, “Multiresolution Approximations of Banach R. Coifman, Y. Meyer, S. Quake and M. Wickerhauser, “Sig Spaces.” nal Processing and Compression with Wave Packets.” V. Zavadsky, “Wavelet Approximation of Sampled Func 60 R. Coifman, Y. Meyer and M. V. Wickerhauser, “Wavelet tions.” Analysis and Signal Processing B. G. Sherlock and D. M. Monro, “On the Space of Orthonor P. Crane, H. Higuchi and J. Lewalle, “On the structure of mal Wavelets. two-dimensional wakes behind a pair of flat plates' M. Vetterli and C. Herley, “Wavelets and Filter Banks: Theory M. Goldburg, Applications of Wavelets to Quantization and and Design.” IEEE Transactions on Signal Processing, Vol. 65 Random Process Representations. About 1.1 MB. 40, 1992, pp. 2207-2232. D. M. Healy, Jr., J. Lu and J. B. Weaver, “Signal Recovery and Frame Decompositions Wavelet Reproducing Kernels.” US 8,874,477 B2 153 154 D. M. Healy, Jr., J. Lu, J. B. Weaver and Y. Xu, “Noise A. Chambole, R. A. DeVore, N. Lee, and B. J. Lucier, “Non Reduction With Multiscale Edge Representation and Per linear Wavelet Image Processing: Variational Problems, ceptual Criteria.” Compression, and Noise Removal through Wavelet D. M. Healy, Jr. and J. Lu, “Contrast Enhancement via Mul Shrinkage.” tiscale Gradient Transformations.” 5 H. Chao and P. Fisher, “An Approach of Fast Integer Revers W. Hwang and S. Mallat, “Singularity Detection and Process ible Wavelet Transforms for Image Compression.” ing with Wavelets.” R. A. DeVore and B. J. Lucier, “Fast Wavelet Techniques for B Jawerth, Y. Liu and W. Sweldens, “Signal Compression Near-Optimal Image Processing with Smooth Local Trigonometric Bases.” An abstract is B. Deng, B. D. Jawerth, G. Peters and W. Sweldens, “Wavelet also available. 10 Probing for Compression Based Segmentation'. An abstract is also available. M. M. Lankhorst and M. D. van der Laan, “Wavelet-Based J. Fan and A. Laine, “An Adaptive Approach for Texture Signal Approximation with Genetic Algorithms.” Segmentation by Multi-Channel Wavelet Frames'. J. Lewalle, K. Read and M.T. Schobeiri, “Effect of unsteady W.T. Freeman and E. P. Simoncelli, “The Steerable Pyramid: wake-passing frequency on boundary layer transition— 15 A Flexible Architecture for Multi-Scale Derivative Com experimental investigation and wavelet analysis’ putation.”/C Source Code (75k) J. Lewalle, S.J. Murphy and F. W. Peek, “Wavelet analysis of A. Grzeszczak, M. K. Mandal, S. Panchanathan and T. Yeap, olfactory nerve response to stimulus’ “VLSI Implementation of Discrete Wavelet Transform.” J. Lewalle, “Wavelet analysis of experimental data: some O. Guleryuz, M. T. Orchard and Z. Xiong, “A DCT-based methods and the underlying physics’ Embedded Image Coder.” G. Strang, “Eigenvalues of (2)H and convergence of the D. M. Healy, Jr., J. Lu and J. B. Weaver, “Contrast Enhance cascade algorithm. ment of Medical Images. Using Multiscale Edge Represen G. Strang, “Creating and comparing wavelets.” tation.” A. R. Tate, “Pattern Recognition Analysis of in vivo Magnetic C. Heil, P. N. Heller, G. Strang, V. Strela, and P. Topiwala, Resonance Spectra’ 25 Accuracy of Lattice Translates of Several Multidimen D. Donoho, “Nonlinear Wavelet Methods for Recovery of sional Refinable Functions.” Signals, Densities, and Spectra from Indirect and Noisy C. Herley, M. T. Orchard, K. Ramchandran and Z. Xiong, Data. Different Perspectives on Wavelets, Proceeding of "Flexible Tree-structured Signal Expansions for Compres Symposia in Applied Mathematics, Vol 47, 1. Daubechies sion Using Time-Varying Filter Banks.” ed. Amer. Math.Soc., Providence, R.I., 1993, pp. 173–205. 30 M. L. Hilton, B. D. Jawerth and A. Sengupta, “/Compressing Wavelets and Image Processing Still and Moving Images with Wavelets” with figure. E. Adelson and E. Simoncelli, "Subband Image Coding with P. Kovesi, "Image Features from Phase Congruency” Three-tap Pyramids.” B. Lin, “Wavelet Phase Filter for Denoising Tomographic E. H. Adelson, W. T. Freeman, D. J. Heeger and E. P. Simo Image Reconstruction' ncelli, “Shiftable Multi-Scale Transforms.' 35 M. K. Mandal, T. Aboulnasr and S. Panchanathan, “Image E. H. Adelson and E. P. Simoncelli, “Subband Transforms.' Indexing. Using Moments and Wavelets.” V. R. Algazi, R. R. Estes and J. Lu, “Comparison of wavelet M. K. Mandal, E. Chan, X. Wang and S. Panchanathan, “Mul image coders using the Picture Quality Scale (POS).” tiresolution Motion Estimation Techniques for Video M. Bhatia, W. C. Karl, and A. S. Willsky, “A Wavelet-Based Compression.” Method for Multiscale Tomographic Reconstruction.” 40 M. K. Mandal, S. Panchanathan and T. Aboulnasr, "Choice of M. Bhatia, W. C. Karl, and A. S. Willsky, “Using Natural Wavelets for Image Compression.” Wavelet Bases and Multiscale Stochastic Models for D. M. Monro and B. G. Sherlock, “Psychovisually Tuned Tomographic Reconstruction.” Wavelet Fingerprint Compression'. M. Louys, J. L. Starck, S. Mei, F. Bonnarel, and F. Murtagh, D. M. Monro and B. G. Sherlock, “Optimised Wavelets for Astronomical Image Compression.” 45 Fingerprint Compression'. M. Louys, J. L. Starck and F. Murtagh, “Lossless Compres P. Moulin, “A Multiscale Relaxation Algorithm for SNR sion of Astronomical Images.” Maximization in 2-D Nonorthogonal Subband Coding.” F. Murtagh and J. L. Starck, “Wavelets and Multiscale Trans M.T. Orchard, Z. Xiong andY. Zhang, “A Simple Deblocking forms in Massive Data Sets.” Algorithm for JPEG Compressed Images Using Overcom F. Murtagh and J. L. Starck, “Image Processing through Mul 50 plete Wavelet Representations.” tiscale Analysis and Measurement Noise Modeling.” M. T. Orchard, K. Ramchandran and Z. Xiong, “Wavelet J. L. Starck and F. Murtagh, “Multiscale Entropy Filtering.” Packets Image Coding Using Space-Frequency Quantiza J. L. Starck and F. Murtagh, “Image Filtering by Combining tion. Multiple Vision Models”. M. T. Orchard, K. Ramchandran and Z. Xiong, “Space-fre F. Murtagh, “Wedding the Wavelet Transform and Multivari 55 quency Quantization for Wavelet Image Coding.” ate Data Analysis.” H. Pan, “Uniform Full-Information Image Matching. Using F. Murtagh and J. L. Starck, “Pattern Clustering based on Complex Conjugate Wavelet Pyramids”, with figures. Noise Modeling in Wavelet Space.” H. Pan, “General Stereo Image Matching Using Symmetric G. Zheng, J. L. Starck, J. G. Campbell and F. Murtagh, “Mul Complex Wavelets.” presented at SPIE Conference: Wave tiscale Transforms for Filtering Financial Data Streams.” 60 let Applications in Signal and Image Processing, VI. Den M. Morehart, F. Murtagh and J. L. Starck, “Multiresolution ver, August 1996, Published in SPIE Proceedings, vol. Spatial Analysis.” 2825. F. Murtagh, J. L. Starck and M. W. Berry, “Overcoming the P. Schröder and W. Sweldens, “Spherical wavelets: Effi Curse of Dimensionality in Clustering by means of the ciently representing functions on the sphere. An abstractis Wavelet Transform. 65 also available. R. A. Carmona, R. D. Frostig and W. L. Hwang, “Wavelet P. Schröder and W. Sweldens, “Spherical Wavelets: Texture Analysis for Brain Function Imaging.” Processing.” An abstract is also available. US 8,874,477 B2 155 156 J. A. Solomon, J. Villasenor, A. B. Watson and G.Y. Yang, G. Beylkin and J. H. Keiser, “On the Adaptive Numerical “Visual Thresholds For Wavelet Quantization Error.” Solution of Nonlinear Partial Differential Equations in V. Strela, P. Heller, G. Strang, P. Topiwala and C. Heil, “The Wavelet Bases. application of multiwavelet filterbanks to signal and image D. M. Bond and S. A. Vavasis, “Fast Wavelet Transforms for processing.” Matrices Arising From Boundary Element Methods.” Y. Wang, “Image representations using multiscale differential T. Chan, W. Tang and W. Wan, P. Charton and V. Perrier, operators.” “Factorisation sur Bases d’Ondelettes du Noyeau de la Y. Wang and S. L. Lee, “Scale-space derived from B-splines.” Chaleur et Algorithmes Matriciels Rapides Associes.” G. Weiss, “Time-Frequency and Time-Scaling Methods in P. Charton and V. Perrier, “Towards a Wavelet Based Numeri Signal and Image Processing 10 M. Wickerhauser, “Picture Compression by Best-Basis Sub cal Scheme for the Two-Dimensional Navier-Stokes Equa band Coding.” tions.” M. V. Wickerhauser, “High-Resolution Still Picture Com P. Charton and V. Perrier, “A Pseudo-Wavelet Scheme for the pression” Two-Dimensional Navier-Stokes Equations.” Z. Xiong, “Representation and Coding of Images Using 15 S. Dahlke and A. Kunoth, “Biorthogonal Wavelets and Mul Wavelets. tigrid.” D. M. Monro and B.G. Sherlock, “Space-Frequency Balance S. Dahlke and I. Weinreich, “Wavelet-Galerkin Methods: An in Biorthogonal Wavelets.” Adapted Biorthogonal Wavelet Basis.” Xuejun Li, "Low Bit Rate Wavelet Image and Video Coding S. Dahlke and I. Weinreich, “Wavelet Bases Adapted to Algorithm and Software.” Pseudo-Differential Operators.” The FBI Wavelet Fingerprint Compression Standard W. Dahmen and A. Kunoth, “Multilevel Preconditioning.” J. N. Bradley and C. M. Brislawn, “Proposed First-Genera W. Dahmen, A. Kunoth and K. Urban "A Wavelet-Galerkin tion WSQ Bit Allocation Procedure” Method for the Stokes-Equations, also full version with J. N. Bradley and C. M. Brislawn, “The Wavelet/Scalar Quan pictures. tization Compression Standard for Digital Fingerprint 25 R. Glowinski, T. Pan, R. O. Wells, Jr. and X. Zhou, “Wavelet Images.” and Finite Element Solutions for the Neumann Problem J. Bradley, C. Brislawn and T. Hopper, “WSQ Gray-Scale Using Fictitious Domains' Fingerprint Image Compression Specification.” R. Glowinski, A. Rieder, R. O. Wells, Jr. and X. Zhou, “A J. Bradley, C. Brislawn and T. Hopper, “The FBI Wavelet/ Wavelet Multigrid Preconditioner for Dirichlet Boundary Scalar Quantization Standard for Gray-Scale Fingerprint 30 Value Problems in General Domains.” Image Compression' with figures. R. Glowinski, A. Rieder, R. O. Wells, Jr. and X. Zhou, “A C. M. Brislawn, "Classification of Nonexpansive Symmetric Preconditioned CG-Method for Wavelet-Galerkin Dis Extension Transforms for Multirate Filter Banks' C. M. Brislawn, “Fingerprints Go Digital” cretizations of Elliptic Problems’ C. M. Brislawn, “Preservation of Subband Symmetry in Mul 35 F. Heurtaux, F. Planchon and M. V. Wickerhauser, “Scale tirate Signal Coding.” Decomposition in Burgers' Equation' “The FBI Wavelet/Scalar Quantization Fingerprint Image A. Jiang, J. H. Keiser, "On I. Wavelet Based Approach to Compression Standard.” Numerical Solution on Nonlinear Partial Differential Wavelets and Speech Processing Equations and II. Nonlinear Waves in Fully Discrete E. Wesfreid and M. V. Wickerhauser, Adapted Local Trigo 40 Dynamical Systems.” nometric Transforms and Speech Processing A. Kunoth, “Multilevel Preconditioning Appending M. Wickerhauser, Acoustic Signal Compression with Wave Boundary Conditions by Lagrange Multipliers.” lets Packets. G. Leaf and J. M. Restrepo, “Wavelet-Galerkin Discretization Wavelets and Ordinary Differential Equations of Hyperbolic Equations.” G. Beylkin, “On Wavelet-based Algorithms for Solving Dif 45 J. Lewalle, “Wavelet Transforms of some Equations of Fluid ferential Equations.” Mechanics' B Jawerth and W. Sweldens, “Wavelet Multiresolution J. Lewalle, “Energy Dissipation in the Wavelet-Transformed Analyses Adapted for the FastSolution of Boundary Value Navier-Stokes Equations” Ordinary Differential Equations.” An abstract is also avail J. Lewalle, “On the effect of boundary conditions on the able. 50 multifractal statistics of incompressible turbulence' P. Monasse and V. Perrier, “Ondelettes sur l'Intervalle pour la J. Lewalle, "Diffusion is Hamiltonian'. Prise en Compte de Conditions aux Limites.” D. Lu, T. Ohyoshi and L. Zhu, “Treatment of Boundary Con A. Rieder, “Semi-Algebraic Multi-level Methods Based on ditions in the Application of Wavelet-Galerkin Method to a Wavelet Decompositions I: Application to Two-Point SH Wave Problem Boundary Problems” 55 P. Monasse and V. Perrier, “Orthonormal Wavelet Bases W. C. Shann and J. C. Xu, "Galerkin-wavelet Methods for Adapted for Partial Differential Equations with Boundary Two Point Boundary Value Problems.” Conditions.” Wavelets and Partial Differential Equations A. Rieder and X. Zhou, “On the Robustness of the Damped G. Kaiser, “Complex-Distance Potential Theory and Hyper V-Cycle of the Wavelet Frequency Decompositions Mul bolic Equations' 60 tigrid Method” A. Averbuch, G. Beylkin R. R. Coifman and M. Israeli, “Mul A. Rieder, R. O. Wells, Jr. and X. Zhou, “A Wavelet Approach tiscale Inversion of Elliptic Operators.” to Robust Multilevel Solvers for Anisotropic Elliptic Prob E. Bacry, S. Mallat and G. Papanicolaou, A Wavelet Based lems.” Space-Time Adaptive Numerical Method for Partial Dif A. Rieder, R. O. Wells, Jr. and X. Zhou, “On the Wavelet ferential Equations' 65 Frequency Decomposition Method G. Beylkin and N. Coult, “A Multiresolution Strategy for K. Urban, “A Wavelet-Galerkin Algorithm for the Driven Reduction of Elliptic PDEs and Eigenvalue Problems.” Cavity-Stokes-Problem in Two Space Dimensions.” US 8,874,477 B2 157 158 O. V. Vasilyev and S. Paolucci, A Dynamically Adaptive R. Piessens and W. Sweldens, “Quadrature Formulae and Multilevel Wavelet Collocation Method for Solving Partial Asymptotic Error Expansion of Wavelet Approximations Differential Equations in a Finite Domain.” of Smooth Functions.” An abstract is also available. O.V.Vasilyev and S. Paolucci, “Thermoacoustic Wave Propa R. Piessens and W. Sweldens, “Asymptotic Error Expansion gation Modeling. Using a Dynamically Adaptive Wavelet 5 of Wavelet Approximations of Smooth Functions II. An Collocation Method. abstract is also available. O. V. Vasilyev and S. Paolucci, “A Fast Adaptive Wavelet W. C. Shann, “Quadratures Involving Polynomials and Collocation Algorithm for Multi-Dimensional PDEs.” Daubechies Wavelets. with figures. W. Sweldens, “Construction and Application of Wavelets in O.V. Vasilyev, S. Paolucci and M. Sen, “A Multilevel Wavelet 10 Numerical Analysis.” Collocation Method for Solving Partial Differential Equa M. Wickerhauser, “Nonstandard Matrix Multiplication.” tions in a Finite Domain.” M. V. Wickerhauser, “Computation with Adapted Time-Fre O. V.Vasilyev, Y.Y. Podladchikov and D. A. Yuen, “Modeling quency Atoms” of Compaction Driven Flow in Poro-Viscoelastic Medium 15 M. V. Wickerhauser, “Wavelet Approximations to Jacobians Using Adaptive Wavelet Collocation Method.” with fig and the Inversion of Complicated Maps' U.S. Wavelets and Statistics O. V.Vasilyev, D. A. Yuen and S. Paolucci, “The Solution of F. Abramovich, T. Sapatinas and B. W. Silverman, “Wavelet PDEs. Using Wavelets.” with figures. Thresholding via a Bayesian Approach.” O. V.Vasilyev, D. A. Yuen and Y.Y. Podladchikov, “Applica F. Abramovich, T. Sapatinas and B. W. Silverman, “Stochas bility of Wavelet Algorithm for Geophysical Viscoelastic tic Atomic Decompositions in a Wavelet Dictionary.” Flow.” with figures. F. Abramovich and B. W. Silverman, “The Vaguelette-Wave R. O. Wells, Jr. and X. Zhou, “Wavelet Solutions for the let Decomposition Approach to Statistical Inverse Prob Dirichlet Problem lems.” R. O. Wells, Jr. and X. Zhou, “Wavelet Interpolation and 25 E. H. Adelson and E. P. Simoncelli, “Noise Removal via Approximate Solution of Elliptic Partial Differential Equa Bayesian Wavelet Coring.” tions' A. Antoniadis, G. Gregoire and G. P. Nason, “Density and R. O. Wells, Jr. and X. Zhou, “Representing the Geometry of Hazard Rate Estimation for Right Censored Data using Domains by Wavelets with Applications to Partial Differ Wavelet Methods. ential Equations” 30 T. Bailey, T. Sapatinas, K. Powell and W. J. Krzanowski, “Signal Detection in Underwater Sounds using Wavelets.” R. O. Wells, Jr., “Multiscale Applications of Wavelets to A. G. Bruce, D. L. Donoho, H. Gao and R. D. Martin, Solutions of Partial Differential Equations' “Denoising and Robust Non-linear Wavelet Analysis.” Wavelets and Numerical Analysis A. G. Bruce and H. Gao, “WaveShrink: Shrinkage Functions G. Beylkin, R. Coifman and V. Rokhlin, “Fast Wavelet Trans 35 and Thresholds.” forms and Numerical Algorithms I.’” A. G. Bruce and H. Gao, “WaveShrink with Semisoft Shrink G. Beylkin, “On the Representation of Operators in Bases of age. Compactly Supported Wavelets.” A. G. Bruce and H. Gao, “Understanding WaveShrink: Vari G. Beylkin, “On the Fast Algorithm for Multiplication of ance and Bias Estimation.” Functions in the Wavelet Bases.” 40 A. G. Bruce, H. Gao and D. Ragozin, “S+WAVELETS: An G. Beylkin, “Wavelets and Fast Numerical Algorithms.” Lec Object-Oriented Toolkit for Wavelet Analysis.” ture notes for an AMS short course, 1993. A. G. Bruce and H. Gao, “S+WAVELETS: Algorithms and G. Beylkin, “Wavelets, Multiresolution Analysis and Fast Technical Details.” Numerical Algorithms. Draft of INRIA lectures, May J. Buckheit and D. Donoho, “WaveLab and Reproducible 1991. 45 Research.” G. Beylkin and M. E. Brewster, A Multiresolution Strategy J. F. Burn, A. M. Wilson and G. P. Nason, “Impact During for Numerical Homogenization.” Equine Locomotion: Techniques for Measurement and P. Charton and V. Perrier, “Products Rapides Matrices-Vect Analysis.” eur en Bases d’Ondelettes: Application a la Resolution R. Coifman and N. Saito, "Local Discriminant Bases.” Numericue d’Equation aux Derivees Partielles.” 50 R. Coifman and F. Majid, "Adapted Waveform Analysis and P. Charton, “Produits de Matrices Rapides en Bases Denoising.” dOndelettes: Application a la Resolution Numericue R. Coifman and D. Donoho, “Translation-Invariant De-Nois d’Equation aux Derivees Partielles.” ing.” N.H. Getz, “A Fast Discrete Periodic Wavelet Transform. An R. Dahlhaus, M. H. Neumann and R. von Sachs, “Non-linear associated toolbox of Matlab routines is also available. 55 Wavelet Estimation of Time-Varying Autoregressive Pro L. Jameson, “On the Spline-Based Wavelet Differentiation cesses.” Matrix. A. Davis, A. Marshak and W. Wiscombe, “Wavelet-Based L. Jameson, “On the Differention Matrix for Daubechies Multifractal Analysis of Non-Stationary and/or Intermit Based Wavelets on an Interval.” tent Geophysical Signals.” with figures. L. Jameson, “On the Daubechies-Based Wavelet Differentia 60 B. Deylon and A. Juditsky, “Wavelet Estimators. Global Error tion Matrix. Mesures Revisited. L. Jameson, “On the Wavelet Optimized Finite Difference D. Donoho, “Nonlinear Solution of Linear Inverse Problems Method. by Wavelet-Vaguelette Decomposition' E. Kolaczyk, “Wavelet Methods for the Inversion of Certain D. Donoho, “Smooth Wavelet Decompositions with Blocky Homogeneous Linear Operators in the Presence of Noisy 65 Coefficient Kernels' Data, with FIG.5.1, FIG. 5.2, FIG. 5.4, FIG. 5.5, FIG. 5.6, D. Donoho, “De-noising by Soft Thresholding FIG. 5.8, FIG. 5.10, FIG. 5.11, FIG. 5.13, and FIG. 5.14. D. Donoho, “Interpolating Wavelet Transforms US 8,874,477 B2 159 160 D. Donoho, “Unconditional Bases are Optimal Bases for B. Vidakovic, “Nonlinear Wavelet Shrinkage with Bayes Data Compression and for Statistical Estimation” Rules and Bayes Factors.” D. Donoho and I. Johnstone, Adapting to Unknown Smooth R. von Sachs, G. P. Nason and G. Kroisandt, Adaptive Esti ness by Wavelet Shrinkage” mation of the Evolutionary Wavelet Spectrum.” D. Donoho and I. Johnstone, “Ideal Spatial Adaptation via R. von Sachs and K. Schneider, “Smoothing of Evolutionary Wavelet Shrinkage” Spectra by Non-linear Thresholding.” Also available are D. Donoho and I. Johnstone, "Minimax Estimation via Wave the figures. let Shrinkage' R. von Sachs and M. H. Neumann, "A Wavelet-based Test for D. Donoho and I. Johnstone, “Minimax Risk over I p Balls' Stationarity.” D. Donoho, I. Johnstone, G. Kerkyacharian and D. Picard, 10 R. von Sachs and B. MacGibbon, “Non-parametric Curve “Density Estimation via Wavelet Shrinkage' Estimation by Wavelet Thresholding with Locally Station D. Donoho, I. Johnstone, G. Kerkyacharian and D. Picard, ary Errors.” “Wavelet Shrinkage: Asymptopia?” A. T. Walden, D. B. Percival and E. J. McCoy, “Spectrum D. Donoho and I. Johnstone, “Ideal Denoising in an Estimation by Wavelet Thresholding of Multitaper Estima Orthonormal Basis Chosen From a Library of Bases.” 15 tors. D. Donoho, S. Mallat and R. von Sachs, “Estimating Cova Yazhen Wang, “Jump and sharp cusp detection by wavelets.” riances of Locally Stationary Processes: Rates of Conver Yazhen Wang, “Function estimation via wavelet shrinkage gence of Best Basis Methods.” for long-memory data.” T. Downie and B. W. Silverman, “The Discrete Multiple Yazhen Wang, “Small ball problems via wavelets for Gauss Wavelet Transform and Thresholding Methods.” ian processes.” H. Y. Gao, “Choice of Thresholds for Wavelet Shrinkage Yazhen Wang, “Fractal function estimation via wavelet Estimate of the Spectrum.” shrinkage.” P. Goel and B. Vidakovic, “Wavelet Transformations as Yazhen Wang, “Minimax estimation via wavelets for indirect Diversity Enhancers' long-memory data.” P. Hall and G. P. Nason, “On Choosing a Non-integer Reso 25 Yazhen Wang, “Change curve estimation via wavelets” (with lution Level when Using Wavelet Methods.” an application to image processing); FIG. 4(a), FIG. 4(b). I. Johnstone, “Minimax Bayes, Asymptotic Minimax and Yazhen Wang, “Change-point analysis via wavelets for indi Sparse Wavelet Priors” rect data.” I. M. Johnstone and B. W. Silverman, “Wavelet Threshold Yazhen Wang, “Self-similarity index estimation via wavelets Estimators for Data with Correlated Noise.” 30 for locally self-similar processes” (with Cavanaugh and A. Juditsky, “Wavelet Estimators. Adapting To Unknown Song); FIG. 1, FIG. 2, FIGS. 3 and 4. Smoothness.” M. V. Wickerhauser, “Fast Approximate Factor Analysis.” A. Juditsky and F. Leblanc, “Computing Wavelet Density Wavelets and Econometrics Estimators for Stochastic Processes.” S. A. Greenblatt, “Wavelets in Economics: An Application to G. Katul and B. Vidakovic, “Partitioning eddy motion using 35 Outlier Testing.” Lorentz wavelet filtering.” M. J. Jensen, “Wavelet Analysis of Fractionally Integrated R. Morgan and G. P. Nason, “Wavelet Shrinkage of Itch Processes.’ Response Sata.” M. J. Jensen, “OLS Estimate of the Fractional Differencing P. Moulin, “Wavelet Thresholding Techniques for Power Parameter Using Wavelets Derived From Smoothing Ker Spectrum Estimation.” 40 nels. G. P. Nason and B. W. Silverman, “The Discrete Wavelet Wavelets and Fractals Transform in S.” A. Davis, A. Marshak and W. Wiscombe, “Wavelet-Based G. P. Nason and B. W. Silverman, “The Stationary Wavelet Multifractal Analysis of Non-Stationary and/or Intermit Transform and some Statistical Applications.” tent Geophysical Signals.” with figures. G. P. Nason and B. W. Silverman, “Wavelets for Regression 45 C. Jones, 2-D Wavelet Packet Analysis of Structural Self and other Statistical Problems.” Organization and Morphogenic Regulation in Filamentous G. P. Nason, T. Sapatinas and A. Sawczenko, “Statistical Fungal Colonies. Modelling of Time Series using Non-decimated Wavelet J. Lewalle, "Wavelet Transforms of the Navier-Stokes Equa Representations.” tions and the Generalized Dimensions of Turbulence” G. P. Nason, “Wavelet Regression by Cross-Validation” 50 W. Hwang and S. Mallat, “Characterization of Self-Similar G. P. Nason, “Functional Projection Pursuit.” Multifractals with Wavelet Maxima. M. H. Neumann and R. von Sachs, “Wavelet Thresholding in Wavelets and Communication Theory Anisotropic Function Classes and Application to Adaptive J. Dill and A. R. Lindsey, “Wavelet Packet Modulation: A Estimation of Evolutionary Spectra.” Generalized Method for Orthogonally Multiplexed Com A. B. Owen, “Monte Carlo Variance of Scrambled Equidis 55 munication.” tribution Quadrature.” R. Learned, H. Krim, B. Claus, A. S. Willsky, and W. C. Karl, D. B. Percival, “On the Estimation of the Wavelet Variance.” “Wavelet-Packet-Based Multiple Access Communica A. Pinheiro and B. Vidakovic, “Estimating the Square Root of tion. a Density Via Compactly Supported Wavelets.” A. R. Lindsey, “Multidimensional Signaling via Wavelet J. Raz, L. Dickerson and B. Turetsky, A Wavelet Packet 60 Packets. Model of Evoked Potentials. A. R. Lindsey, Generalized Orthogonally Multiplexed Com N. Saito, “Local Feature Extraction and Its Appplications munication via Wavelet Packet Bases, chapter 1, chapter 2, Using a Library of Bases.” chapter 3, chapter 4, chapter 5, chapter 6. Also with appen N. Saito, “Simultaneous Noise Supression and Signal Com dix and references. pression using a Library of Orthonormal Bases and the 65 Wavelets and Computer Graphics Minimum Description Length Criterion.” M. Cohen, S. Gortler, P. Hanrahan and P. Schröder, “Wavelet B. Vidakovic, “A Note on Random Densities via Wavelets” Radiosity.” With FIG. 12 and FIG. 14. US 8,874,477 B2 161 162 M. Cohen, S. Gortler, P. Hanrahan and P. Schröder, “Wavelet A. Fournier, "Atmospheric energetics in the wavelet domain Projections for Radiosity.” I: Governing equations and interpretation for idealized A. Dreger, M. H. Gross R. Koch and L. Lippert, “A New flows’. Submitted to J. Atmos. Sci., 2000 (revised). Method to Approximate the Volume Rendering Equation A. Fournier, "Atmospheric energetics in the wavelet domain using Wavelet Bases and Piecewise Polynomials, with II: Time-averaged observed atmospheric blocking'. Sub FIGS. 5-6, FIGS. 7-10, and FIGS. 11-13. Also abstract mitted to J. Atmos. Sci., 1999. available. Technical Report No. 220, Computer Science A. Fournier, "Atmospheric energetics in the wavelet domain Department, ETH Zürich, 1994. III: Instantaneous transfers between block and local A. Fournier, “Wavelets and their Applications in Computer eddies', submitted to J. Atmos. Sci., 1999. Graphics.” This is 2.5 MB compressed. 10 Hardware and Software Implementation of Wavelet Trans M. H. Gross and L. Lippert, “Fast Wavelet Based Volume forms Rendering by Accumulation of TranparentTexture Maps.” J. Fridman and E. S. Manolakos, “On Linear Space-Time With FIG. 6, FIG. 7, FIG. 8, FIG. 9, FIG. 10 and FIG. 11. Mapping for the 1-D Discrete Wavelet Transform.” Technical Report No. 228, Computer Science Department, J. Fridman and E. S. Manolakos, “Distributed Memory and ETH Zürich, 1995. 15 Control VLSI Architectures for the 1-D Discrete Wavelet M. H. Gross and R. Koch, “Visualization of Multidimen Transform.” sional Shape and Texture Features in Laser Range Data J. Lu, “Computation of 2-D Wavelet Transform on the Mas using Complex-Valued Gabor Wavelets.” sively Parallel Computer for Image Processing.” M. H. Gross and L. Lippert, “Ray-tracing of Multiresolution MathSoft Engineering & Education, Inc, “Wavelets Exten B-Spline Volumes.” Also abstract available. Technical sion Pack', 1999. Report No. 239, Computer Science Department, ETH Zür O. Nielsen and M. Hegland, “A Scalable Parallel 2D Wavelet ich, 1996. Transform Algorithm.” P. Hanrahan and P. Schröder, “Wavelet Methods for Radiance Game Theory References Computations.” With FIG. 8, FIG.9 left, FIG.9 right, FIG. Use of Game Theory to Control Ad Hoc Networks 10, FIG. 10 top left and FIG. 10 top right. 25 The resources relating to adhoc networks and game theory C. Herley, “Exact Interpolation and Iterative Subdivision listed in the Game Theory and Ad Hoc Networks References Schemes.” Appendix, each of which is expressly incorporated herein by P. Schröder, W. Sweldens and D. Zorin, “Interpolating Sub reference, provides a sound basis for understanding the impli division for Meshes with Arbitrary Topology’. An abstract cations of game theory for the design, control and analysis of is also available. 30 communications networks, and in particular, ad hoc net Wavelets and Physics works. A review of these references will assure a background J. C. van den Berg, ed., “Wavelets in Physics”, Cambridge in this field for an understanding of aspects of the invention University Press, 1999 (a survey of many applications). which rely on these topics. G. Kaiser, "A Detailed Introduction to Mathematical and Description of Games Physical Wavelets” 35 Roger Myerson, Nash Equilibrium and the History of Eco C. Best and A. Schäfer, “Variational Description of Statistical nomic Theory. JEL 1999 Field Theories using Daubechies' Wavelets.” Rationality, Dominance, Weak Dominance etc G. Beylkin, J. Dunn and D. Gines, “Order N Static and Quasi Douglas Bernheim, Rationalizable strategic behavior. Econo Static Computations in Electrodynamics using Wavelets.” metrica 1984 J. C. Cohen and T. Chen, “Fundamentals of the Discrete 40 David Pearce, Rationalizable strategic behavior and the prob Wavelet Transform for Seismic Data Processing.” lem of perfection. Econometrica 1984 A. Fournier, “Wavelet Analysis of Observed Geopotential Douglas Bernheim, Axiomatic characterization of rational and Wind: Blocking and Local Energy Coupling Across choice in strategic enviroments. Scand. J. of E. 1986 Scales. Shimoji and Watson, Conditional Dominance, rationalizabil A. Fournier, “Wavelet Multiresolution Analysis of Numeri 45 ity and game forms. Journal of Economic Theory 1998 cally Simulated 3d Radiative Convection.” David Roth, Rationalizable predatory pricing. Journal of A. Fournier, “Wavelet Representation of Lower-Atmospheric Economic Theory 1996 Long Nonlinear Wave Dynamics, Governed by the Ben Basu and Weibul, strategy subsets closed under rational jamin-Davis-Ono-Burgers Equation.” behavior. E. Letters 1991 F. Herrmann, 'A scaling medium representation, a discussion 50 Larry Samuelson, Dominated Strategies and common knowl on well-logs, fractals and waves.” An abstract is also avail edge. Games and Economic Behavior 1992 able. Marx and Swinkels, Order independence for iterated weak I. Pierce and L. Watkins, “Modelling optical pulse propaga dominance. Games and Economic Behavior 1997 tion in nonlinear media using wavelets.” Equilibrium: Nash, Refinements, Correlated W. C. Shann, “Finite Element Methods for Maxwells Equa 55 Selten, Reexamination of the Perfectness concept for equi tions with Stationary Magnetic Fields and Galerkin-wave librium points in extensive form games. International Jour let Methods for Two Point Boundary Value Problems.” nal of Game Theory 1975. with separate abstract and table of contents. Myerson, Refinements of the Nash equilibrium concept. L. R. Watkins and Y. R. Zhou, “Modelling Propagation in International Journal of Game Theory 1975. Optical Fibres using Wavelets.” 60 Kalai and Samet, Persistent equilibria in strategic games. R. O. Wells, “Adaptive Wave Propogation Modelling.” International Journal of Game Theory 1984. L. Zubair, “Studies in Turbulence using Wavelet Transforms Kohlberg and Mertens. On the strategic stability of Equilib for Data Compression and Scale-Separation.” ria. Econometrica, 1986. A. Fournier, “An introduction to orthonormal wavelet analy Aumann, Correlated equilibria as an expression of baysian sis with shift invariance: Application to observed atmo 65 rationality. Econometrica, 1987. spheric-blocking spatial structure', to appear in J. Atmos. Aumann and Brandenberger, Espitemic conditions for equi Sci., 2000 Dec. 1. librium. Econometrica 1995 US 8,874,477 B2 163 164 Hal Varian, A model of Sales. American Economic Review Dilip Abreu. On the theory of infinitely repeated games with 198O Discounting. Econometrica 1988 The Extensive Form Games with Perfect Information Benoit and Krishna. Finitely Repeated games. Econometrica, Rubinstein, On the interpretation of game theory. Economet 1985. rica 1991 5 James Friedman. A noncooperative equilibrium for Super Reny, Common beliefs and the theory of games with perfect games. Review of Economic Studies 1971. information. Journal of Economic Theory 1993. James Friedman. Cooperative equilibria in finite horizon Aumann Backward induction and common knowledge of supergames. Journal of Economic Theory 1985. rationality. Games and Economic Behavior 1995 Fudenberg and Maskins. The Folk Theorem in repeated Binmore. A note on backward induction: Aumann, Reply to 10 games with discounting or with incomplete information. Binmore. Games and Economic Behavior 1996 Econometrica 1986. Selten, A Reexamination of the perfectness. Roy Radner. Collusive Behavior in non-cooperative epsilon Hyperbolic Discounting equilibria in oligopolies with long but finite lives. Journal O'Donoghue and Rabin, Doing it now or doing it later. 15 of Economic Theory 1980. American Economic Review 1999 Ariel Rubinstein. Equilibrium in supergames with the over David Laibson, Golden Eggs and Hyperbolic Discounting. taking criterion. Journal of Economic Theory 1977. Quarterly Journal of Economics 1997 Renegotiation The Economics of Altruism Benoit and Krisna, Renegotiation in finitely repeated games. Gary Becker, Atheory of social interactions. Journal of Politi Econometrica 1993 cal Economy 1974 Bergin and MacCleod, Efficiency and renegotiation in Ted Bergstrom, A fresh look at the rotten kid theorem and repeated games. Journal of Economic Theory 1993 other household mysteries. Journal of Political Economy Andreas Blume, Interplay communication in repeated games. 1989 Games and Economic Behavior 1994 Bernheim and Stark, Altruism within the family reconsid 25 Geir Asheim, Extending renegotiation proofness to infinite ered: do nice guys finish last. American Economic Review horizon games. Games and Economic Behavior 1991 1988 Bernheim and Ray, Collective dynamic consistency in Lindbeck and Weibull. Altruism and time consistency: the repeated games. Games and Economic Behavior 1989 economics of fait accompli. Journal of Political Economy Farrel and Maskin, Renegotiation in Repeated Games. 1988 30 Games and Economic Behavior 1989 Bruce and Waldman, Transfers in kind: why they can be Cooperative Game Theory efficient and non-paternalistic. American Economic Freidman, Game theory with applications to economics chap Review 1991 ter 6 and 7 Jack Robles, Paternal altruism or smart parent altruism? CU Nash, The Bargaining problem. Econometrica 1950 WP 98-10 35 Kalai and Smordinski. Other Solutions to Nash's problem. Mathew Rabin, Incorperating fairness into economics and Econometrica 1975 game theory. American Economic Review 1993 Noncooperative Bargaining Ray and Ueda, Egalitarianism and incentives. Journal of Eco Rubinstein, Perfect equilibrium in a bargaining model. nomic Theory 1996 Econometrica 1982 Bernheim, Shleifer and Summers. The strategic bequest 40 Joel Watson, Alternating offer bargaining with two sided motive. Journal of Political Economy 1985 incomplete information. Review of Economic Studies Extensive Form Games without Perfect Information 1999 Kreps and Wilson, Sequential Equilibrium. Econometrica, Reputation 1983. Kreps, Milgrom, Roberts and Wilson, Reputation and imper Van Damme, Stable Equilibria and forward induction. Jour 45 fect information: predation, reputation and entry deter nal of Economic Theory 1989. ence: rational cooperation in the finitely repeated prison Strategic Information Transmission er's dilemma. Journal of Economic Theory 1981 Crawford and Sobel, Strategic information transmission. Aumann and Sorin, Cooperation and Bounded recal. Games Econometrica 1982. and Economic Behavior 1989 Cho and Kreps, Signalling games and stable equilibria. Quar 50 Klaus Schmidt, Reputation and equilibrium characterization terly Journal of Economics 1987 in repeated games with conflicting interests, Econometrica Mailath, Okuno-Fujiwara and Postlewaite, Belief based 1993, 325-352 refinements in signalling games. Journal of Economic Cripps and Thomas, Reputation and Commitment in Two Theory 1993 person games without discounting, Econometrica, 1995, Milgrom and Roberts, Limit pricing and entry under incom 55 1401-1420 plete information: an equilibrium analysis. Econometrica Joel Watson. A reputation refinement withough equilibrium, 1982 (pages 443-459) Econometrica 1993, 199-206 Cho and Sobel, Strategic Stability and uniqueness in signal Celentani, Fudenberg and Levine, Maintaining a reputation ling games. Journal of Economic Theory 1990. against along lived opponent Econometrica 1996, 691-704 Farrell, Meaning and credibility in cheap talk games. Games 60 Evolutionary Game Theory and Economic Behavior Vince Crawford, An Evolutionary interpretation of VHBB's Milgrom and Roberts, Limit pricing and entry under incom experimental results on coordination. Games and Eco plete information, an equilibrium analysis, Econometrica nomic Behavior 1991 1982 Gilboa and Matsui, Social Stability and Equilibrium, Econo Milgrom, Good news and bad news, representation and appli 65 metrica 1991 cations, Rand. Kandori, Mailath and Rob, Learning, Mutation, and Long Folk Theorems for Repeated Games Run Equilibria in games, Econometrica 1993. US 8,874,477 B2 165 166 Peyton Young. An Evolutionary Model of Bargaining, Jour Brams S.J. & J. M. Togman, Cooperation through threats: the nal of Economic Theory 1993 Northern Ireland case, PS: Political Science & Politics, Peyton Young. The Evolution of Conventions, Econometrica March 1998. 1993 Gardner, Chan-person games, Ch5 Non-cooperative games. Larry Samuelson, Stochastic Stability with alternative best Colman A. M., Ch8 Multi-person games: Social dilemmas, in replies. Journal of Economic Theory his Game Theory and Experimental Games, Oxford: Per Noldeke and Samuelson, The Evolution of Backwards and gamon, 1982. Forwards Induction, Games and Economic Behavior 1993 Kay J., Ch5 Co-operation and Co-ordination, in his Founda Jack Robles, An Evolutionary Folk Theorem For Finitely tions of Corporate Success: How Business Strategies Add 10 Value, Oxford: OUP, 1993. Repeated Games CUWP 99 Brams S. J., Chl Int. relations games, in Game Theory and Kim and Sobel, An Evolutionary Approach to Preplay Com Politics, NY: Macmillan, 1975. munication Econometrica 1995 Credible Commitment General Game Theory Dixit & Nalebuff, Chó Credible commitments. Bierman H. S. & Fernandez, L., GameTheory with Economic 15 Bierman & Fernandez, Ch23 Subgame-perfect equilibrium Applications. Addison-Wesley, 1993. Rasmusen, Cha.1 Subgame perfection. Dixit A., & Nalebuff B. Thinking Strategically: the Competi Gardner, Ché Credibility and subgame perfection. tive Edge in Business, Politics, and Everyday Life, New Ghemawat, Ch3 Preemptive capacity expansion in the tita York: Norton, 1991. nium dioxide industry. McMillan J. Games, Strategies, and Managers, Oxford: Repetition and Reputation OUP, 1992. Dixit & Nalebuff, Cha Resolving the Prisoner's Dilemma: Baird D. G., Gertner R. H., and Picker R. C., Game Theory Ch9 Cooperation and coordination. and the Law, Harvard U.P., 1994. Nowak, M., R. May, & K. Sigmund, The arithmetic of mutual Rasmusen E. Games and Information: An Introduction to help, Scientific American, 1995 Game Theory, Oxford: B. Blackwell, 2nd edition, 1994. 25 Hofstadter D., Ch29 The Prisoner's Dilemma computer tour Ghemawat P. Games Businesses Play: Cases and Models, naments and the evolution of cooperation, in his New York: Wiley, 1995. Metamagical Themas, Penguin, 1985. Gardner R., Games for Business and Economics, New York: Marks R. E., Midgley F D. F., & Cooper L. G., Adaptive Wiley, 1995. behaviour in an oligopoly, in Evolutionary Algorithms in Strategic Decision Making 30 Management Applications, ed. by J. Biethahn & V. Nissen, Dixit & Nalebuff, Intro; Ch2 Anticipating your rivals (Berlin: Springer-Verlag), 1995. response; Ch3 Seeing through your rivals response. Baird Gertner & Picker, Ch2 Dynamic interaction and the Barnett, F. W. Making game theory work in practice, Wall extensive-form game, Ch5 Reputation and repeated games. Street Journal, 1995. Gardner, Ch7 Repeated games, Ch8 Evolutionary stability Bierman & Fernandez, Ch5 Nash equilibrium I, Ch11 Nash 35 and bounded rationality. equilibrium II Rasmusen, Cha Dynamic games and symmetric information, O'Neill B., International escalation and the dollar auction, Ch5 Reputation and repeated games with symmetric infor Journal of Conflict Resolution, 1986. mation. Schelling T. C., Ch7 Hockey helmets, daylight saving, and Unpredictability other binary choices, in his Micromotives and Macrobe 40 Dixit & Nalebuff, Ch7 Unpredictability; Ch8 Brinkmanship. havior, NY: Norton, 1978. Bierman & Fernandez, Ch11.9 Marks R. E. Competition and common property, 1998. Gardner, Ch3 Mixed strategies. McMillan J., Ch3 Understanding cooperation and conflict. Rasmusen, Ch3 Mixed and continuous strategies. McAfee R. P. & J. McMillan, Competition and game theory, Bargaining Journal of Marketing Research, 1996. 45 Dixit & Nalebuff, Ch10 The voting strategy; Ch11 Bargain Baird, Gertner, & Picker, Ch1 Simultaneous decision-making 1ng. and the normal form game. McMillan, Ch5 Gaining bargaining power; Chó Using infor Gardner, Ch1 Introduction, Ch2 Two-person games, Ch16 mation strategically. Voting games. Elster J., Ch14 Bargaining, in Nuts and Bolts for the Social Rasmusen, Chl The rules of the game. 50 Sciences, Camb.: CUP, 1989 Schelling T. C., What is game theory? in his Choice and Murnighan J. K. Game's End, Chapter 15 in his: Bargaining Consequence: Perspectives of an Errant Economist, Games: A New Approach to Strategic Thinking in Nego Camb.: Harvard UP, 1980. tiations, NY: William Morrow, 1992. Decision Analysis—Games Against Nature Bierman & Fernandez, Chó Bargaining. Apocalpse maybe, and An insurer's worst nightmare. The 55 Schelling T. C., Ch2 Essay on bargaining, in The Strategy of Economist, 1995/96 Conflict, Camb.: Harvard UP, 1980. Bierman & Fernandez, Chs 1-3. Baird Gertner & Picker, Ch7 Noncooperative bargaining Ulvila J. W. & R. Brown, Decision analysis comes of age, Gardner, Ch12 Two-person bargains. Ch14 n-person bar Harvard Buisness Review 1982. gaining and the core. Howard R. A., Decision analysis: practice and promise, Man 60 Rasmusen, Ch11 Bargaining. agement Science, 1988. Brams S. J., Negotiation Games: Applying Game Theory to Clemen R. T., Making Hard Decisions: An Introduction to Bargaining and Arbitration, NY: Routledge, 1990. Decision Analysis, Belmont, Calif.: Duxbury, 1996. Using Information Strategically Samson D., Chs 2-6, 11, Managerial Decision Analysis, Chi McMillan, Ché Using information strategically cago: R. D. Irwin, 1988. 65 Bierman & Fernandez, Ch17 Bayesian equilibrium, Ch19 Strategic Moves Adverse selection and credit rationing Dixit & Nalebuff, Ch5 Strategic moves. Rasmusen, Ch2 Information P-13 US 8,874,477 B2 167 168 Baird Gertner & Picker, Cha Signalling, screening, and non For a History of Game Theory Since Old Testament Times, verifiable information Point your Browser at the Following URL: www.canter Gardner, Ch9 Signaling games. bury.ac.nz/econ/hist.htm Bidding in Competition For further surfing on the Net about game theory, start at the Revenge of the nerds. It’s only a game, and Learning to play 5 following URLs: www.pitt.edu/~alroth/alroth.html the game. The Economist, 1994 E. Dekel, D. Fudenberg and D. K. Levine, Learning to Play Landsburg S. E. Cursed winners and glum losers, Ch18 of his Bayesian Games (Jun. 20, 2001). www.gametheory.net/ The Armchair Economist: Economics and Everyday Life, html/lectures.html New York: The Free Press, 1993. Drew Fudenberg and David K. Levine, The Nash Threats 10 Folk Theorem With Communication and Approximate Norton, R., Winning the game of business, Fortune, 1995, Common Knowledge in Two Player Games (Jun. 10, Koselka, R., Playing poker with Craig McCaw, Forbes, 1995, 2002). Dixit & Nalebuff, Ch12 Incentives. Game Theory and Ad Hoc Networks References McMillan, Ch11 Bidding in competition E. Altman, R. El Azouzi, and T. Jimenez. Slotted aloha as a McAfee R. P. & J. McMillan, Analyzing the airwaves auction, 15 stochastic game with partial information. In Proceedings Journal of Ec. Perspectives, 1996 of WiOptO3, 2003. R. Marks, Closed tender vs. open bidding auctions, 22 Dec. A Archer and E Tardos. Truthful mechanisms for one-param 1994. eter agents. In Proc. 42nd IEEE Symp. On Foundations of The Economist, Secrets and the prize, 12 Oct. 1996, p. 98. Computer Science, 2001. Scientific American, Making honesty pay, January 1997, p. N. Ben Salem, L. Buttyan, J.P. Hubaux, and Jakobsson M. A 13. charging and rewarding scheme for packet forwarding. In Gardner, Ch11 Auctions. Proceeding of Mobihoc, June 2003. Brams S.J. & A. D. Taylor, Fair division by auctions, Ch9 of L. Blazevic, L. Buttyan, S. Capkun, S. Giordiano, J.-P. their Fair Division: From Cake-Cutting to Dispute Reso Hubaux, and J.-Y. Le Boudec. Self-organization in mobile lution, Cambridge: CUP, 1996. 25 ad-hoc networks: the approach of terminodes. IEEE Com Rasmusen, Ch12 Auctions. munications Magazine, 39(6):166-174, June 2001. Contracting, or the Rules of the Game G. E. Bolton and A. Ockenfels. ERC: A theory of equity, Kay, Cha. Relationships and contracts. reciprocity and competition. The American Economic Dixit & Nalebuff, Ch12 Incentives. Review, 90:166-193, March 2000. McMillan, Ch8 Creating incentives; Ch9 Designing con 30 A. Bovopoulos and A. Lazar. Asynchronous iterative algo tracts, Ch10 Setting executives’ salaries. rithms for optimal load balancing. In Proceedings of the Williamson O. E., Strategizing, economizing, and economic 22nd Annual Conference on Information Sciences and organization, Strategic Management Journal, 1991. Systems, pages 1051-1057, 1988. Bierman & Fernandez, Ch7 Involuntary unemployment. Felix Brandt and Gerhard We?. Antisocial Agents and Vick Gardner, Ch10 Games between a principal and an agent. 35 rey Auctions. In Pre-proceedings of the Eighth Interna Milgrom P. & Roberts.J., Ch5 Bounded rationality and private tional Workshop on Agent Theories, Architectures, and information; Chó Moral hazard and performance incen Languages (ATAL-2001), pages 120-132, 2001. tives. Felix Brandt. A Verifiable, Bidder-Resolved Auction Proto Economics, Organization and Management, Englewood col. In Proceedings of the 5th International Workshop on Cliffs: Prentice-Hall, 1992. 40 Deception, Fraud and Trust in Agent Societies, pages Choosing the Right Game: Co-Opetition 18-25, 2002. Brandenburger A. M. & B.J. Nalebuff. The right game: using S. Buchegger and J. Le Boudec. Performance analysis of the GameTheory to shape strategy, Harvard Business Review, CONFIDANT protocol. In Proceedings of IEEE/ACM 1995 mayet. Som.yale.edu/coopetition/index2.html Workshop on Mobile Ad Hoc Networking and Computing Koselka R., Businessman's dilemma, and Evolutionary eco 45 (MobiFIOC), June 2002. nomics: nice guys don’t finish last, Forbes, Oct. 11, 1993. R. Buyya, H. Stockinger, J. Giddy, and D. Abramson. Eco Brandenburger A. M. & B. J. Nalebuff, Co-opetition: 1. A nomic Models for Management of Resources in Peer-to revolutionary mindset that combines competition and Peer and Grid Computing. In Proceedings of the SPIE cooperation; 2. The Game Theory Strategy that’s changing International Symposium on The Convergence of Informa the game of business. New York: Currency Doubleday, 50 tion Technologies and Communications (ITCOM), August 1996. 2001. Brandenburger A.M. & Harborne W. S. Jr., Value-based busi M. Castro, P. Druschel, A. Ganesh, A. Rowstron, and D. S. ness strategy, Journal of Economics and Management Wallach. Security for structured peer-to-peer overlay net Strategy, 5(1), 1996. works. In Proc. 5th Symposium on Operating Systems Baird Gertner & Picker, Chó Collective action, embedded 55 Design and Implementation, Boston, Mass., December games, and the limits of simple models. 2002. Morrow J. D., Game Theory for Political Scientists, Princ Clarke, E. H. (1971). Multipart pricing of public goods, Pub eton: P.U.P., 1994. lic Choice 11, 17-33. Casson M., The Economics of Business Culture: Game T. Clausen and P. Jacquet. Optimized link state routing pro Theory, Transaction Costs and Economic Performance, 60 tocol. October 2003.hipercom.inria.fr/olsr/ Oxford: OUP, 1991. S. Corson and J. Macker. Mobile ad hoc networking (MA Schelling T. C. Altruism, meanness, and other potentially NET): Routing protocol performance issues and evalua strategic behaviors, American Economic Review, 68(2): tion considerations. Request for comments 2501, Internet 229-231, May 1978. Engineering Task Force, 1999. www.rfc-editor.org/ Crawford V.P., Thomas Schelling and the analysis of strategic 65 J. Crowcroft, R. Gibbens, F. Kelly, and S. Ostring. Modelling behavior, in Strategy and Choice, ed. by R.J. Zeckhauser, incentives for collaboration in mobile ad hoc networks. In MIT Press, 1991. Proceedings of WiOpt 03, 2003. US 8,874,477 B2 169 170 Dasgupta, P., P. Hammond, and E. Maskin (1979). The imple Michael Harkavy, J. D. Tygar, and Hiroaki Kikuchi. Elec mentation of social choice rules, Review of Economic tronic Auctions with Private Bids. In 3rd USENIX Work Studies 46, 185-216. shop on Electronic Commerce, pages 61-74, September Deering, S., D. Estrin, D. Farinacci, V. Jacobson, C. Liu, and 1998. L. Wei (1996). The PIM architecture for wide-area multi 5 X. Hong, K. Xu, and M. Gerla. Scalable routing protocols for cast routing, ACM/IEEE Transactions on Networking 54, mobile ad hoc networks. IEEE Networks, 16(4): 11-21, 153-162. July 2002. A. Demers, S. Keshav, and S. Shenker. Analysis and simula M.-T. Hsiao and A. Lazar. Optimal decentralized flow control tion of a fair queueing algorithm. In SIGCOMM 89. Pro of markovian queueing networks with multiple controllers. 10 Performance Evaluation, 13(3):181-204, 1991. ceedings of the ACM Symposium on Communications Jackson, Matthew O., and Asher Wolinsky (1996). “A Stra Architectures & Protocols, pages 1-12, 1989. tegic Model of Social and Economic Networks.” Journal of C. Doulligeris and R. Mazumdar. User optimal flow control in Economic Theory 71, 44-74. an integrated environment. In Proceedings of the Indo-US M. Jakobsson, J.P. Hubaux, and L. Huttyan. A micro-payment Workshop on Signals and Systems, 1988. 15 scheme encouraging collaboration in multi-hop cellular Joan Feigenbaum, Christos HPapadimitriou, and Scott Shen networks. In Proceedings of Financial Crypto 2003, Janu ker. Sharing the cost of multicast transmissions. Journal of ary 2003. Computer and System Sciences, 63:21-41, 2001. Y. Jin and G. Kesidis. Equilibria of a noncooperative game for Feigenbaum, Joan, Christos Papadimitriou, Rahul Sami, and heterogenous users of an ALOHA network. IEEE Commu Scott Shenker (2002). “A BGP-based Mechanism for Low nications Letters, 6(7):282-284, July 2002. est-Cost Routing.” In Proc. 21st Symposium on Principles D. Johnson, D. Maltz, and Y.-C. Hu. The dynamic source of Distributed Computing, ACM Press, 173-182. routing protocol for mobile ad hoc networks, April 2003. J. Feigenbaum and S. Shenker. Distributed algorithmic www.ietforg/internet-drafts/draft-ietf-manet-dsr-09.txt mechanism design: Recent results and future directions. In Johnson, D. S., M. Minko , and S. Phillips (2000). The prize Proc. 6th Int’l Workshop on Discrete Algorithms and 25 collecting Steiner tree problem: theory and practice, in Methods for Mobile Computing and Communications, “Proceedings of the 11th Symposium on Discrete Algo pages 1-13, Atlanta, Ga., September 2002. rithms.” pp. 760-769, ACM Press/SIAM, New York/Phila T. Fent, G. Feichtinger, and G. Tragler. A dynamic game of delphia. offending and law enforcement. International Game Y. Korilis and A. Lazar. On the existence of equilibria in Theory Review, 4(1):71-89, 2002. 30 noncooperative optimal flow control. Journal of the ACM, Ferguson, D., C. Nikolaou, and Y. Yemini (1989). An 42(3):584-613, 1995. economy for flow control in computer networks, in "Pro Korilis, Y., A. A. Lazar, and A. Orda (1995). Architecting ceedings of the 8th Infocom. pp. 100-118, IEEE Com noncooperative networks, Journal on Selected Areas in puter Society Press, Los Alamitos. Communications 13, 1241-1251. S. Floyd and K. Fall. Promoting the use of end-to-end con 35 R. La and V. Anantharam. Charge-sensitive tcp and rate con gestion control. IEEE/ACM Transactions on Networking, trol in the internet. In Proceedings of INFOCOM 2000, 7(4):458-472, August 1999. 2OOO. E. Friedman and D. Parkes, “Pricing WiFi at Starbucks— R. La and V. Anantharam. Optimal routing control: Repeated Issues in Online Mechanism Design”. In Proc. Fourth game approach. IEEE Transactions on Automatic Control, ACM Conf. on Elec. Commerce (EC03), 2003. Extended 40 47(3):437-450, 2002. version at www.eecs.harvard.edu/econcs/pubs/online.pdf. Ron Lavi and Noam Nisan. Competitive analysis of incentive citeseer.nj.nec.com/article/friedman03pricing.html compatible on-line auctions. In Proc. 2nd ACM Conf. on Friedman, E., and S. Shenker (1997). “Learning and Imple Electronic Commerce (EC-00), 2000. mentation in the Internet, preprint. www.aciri.org/shen S.-J. Lee, W. Su, J. Hsu, M. Gerla, and R. Bagrodia. A per ker/decent.ps 45 formance comparison study of ad hoc wireless multicast A. C. Fuqua, T-W Ngan, and D. S. Wallach, “Economic protocols. In Proceedings of IEEE INFOCOM 2000, pages Behavior of Peer-to-Peer Storage Networks”, Workshop 565-574, March 2000. on Economics of Peer-to-Peer Systems, Berkeley, Calif., A. Legout and E. W. Biersack. Revisiting the fair queueing 2003, citeseer.nj.nec.com/fuquaC)3economic.html paradigm for end-to-end congestion control. IEEE Net Goldberg, A.V., J. D. Hartline, and A. Wright (1999). “Com 50 work Magazine, 16(5):38-46, September 2002. petitive Auctions and Digital Goods.” InterTrust Technical A. B. MacKenzie and S. B. Wicker. Selfish users in aloha: A Report 99-01. Available at www.intertrust.com/star/tr/tr game-theoretic approach. In Vehicular Technology Con 99-01.html ference, 2001. VTC 2001 Fall. IEEE VTS 54th, volume 3, A. Goldsmith and S. Wicker. Design challenges for energy October 2001. constrained ad hoc wireless networks. IEEE Wireless 55 J. MacKie-Mason and H. Varian. Pricing congestible network Communications, 9(4):8-27, 2002. resources. IEEE Journal on Selected Areas in Communi P. Golle, K. Leyton-Brown, I. Mironov, and M. Lillibridge. cations, 13(7): 1141-1149, 1995. Incentives for sharing in peer-to-peer networks. In Proc. P. Marbach. Priority service and max-min fairness. In Pro 3rd ACM Conf. on Electronic Commerce, Tampa, Fla., ceedings of IEEE INFOCOM 2002, volume 1, pages 266 October 2001. 60 275, 2002. Z. Haas. A new routing protocol for reconfigurable wireless S. Marti, T. J. Giuli, K. Lai, and M. Baker. Mitigating routing networks. In IEEE 6th International Conference on Uni misbehavior in mobile ad hoc networks. In Proceedings of versal Communications Record, volume 2, pages 562-566, MOBICOM 2000, August 2000. October 1997. K. Mase, Y. Wada, N. Mori, K. Nakano, M. Sengoku, and S. Garrett Hardin. The Tragedy of the Commons. Science, 162: 65 Shinoda. Flooding schemes for a universal adhoc network. 1243-1248, 1968. Alternate Location: diedff.com/ In Industrial Electronics Society, 2000. IECON 2000., vol page95.htm. ume 2, pages 1129-1134, 2000. US 8,874,477 B2 171 172 R. Mazumdar, L. Mason, and C. Doulligeris. Fairness in net C. Papadimitriou. Algorithms, games, and the internet. In work optimal flow control: Optimality of product forms. Proc. 33rd ACM Symposium on Theory of Computing, IEEE Transactions on Communications, 39(5):775-782. pages 1-5, Hersonissos, Crete, Greece, July 2001. 1991. D. C. Parkes. Iterative Combinatorial Auctions: Achieving P. Michiardi and R. Molva. Core: A collaborative reputation 5 Economic and Computational Efficiency (Chapter 2). PhD mechanism to enforce node cooperation in mobile ad hoc thesis, University of Pennsylvania, May 2001. www.eec networks. In Communication and Multimedia Security s.harvard.edu/parkes/pubs/ch2.ps. 2002 Conference, 2002. W. Peng, X.-C. Lu. On the reduction of broadcast redundancy P. Michiardi and R. Molva. Game theoretic analysis of secu in mobile ad hoc networks. In Mobile and Ad Hoc Net 10 working and Computing, 2000. MobiFHOC., pages 129 rity in mobile ad hoc networks. Technical Report RR-02 130, 2000. 070, Institut Eurécom, 2002. C. Perkins, E. Belding-Royer, and S. Das. Ad hoc on-demand P. Michiardi and R. Molva. A game theoretical approach to distance vector (AODV) routing. Request for comments evaluate cooperation enforcement mechanisms in mobile 3561, Internet Engineering Task Force, 2003. ad hoc networks. In Proceedings of WiOpt 03, March 15 C. E. Perkins, editor. Ad Hoc Networking. Addison-Wesley, 2003. Boston, 2001. Dov Monderer and Moshe Tennenholtz. Distributed Games: Adrian Perrig, Sean Smith, Dawn Song, and J. Doug Tygar. From Mechanisms to Protocols. In Proceedings of the 16th SAM: A Flexible and Secure Auction Architecture Using National Conference on Artificial Intelligence (AAAI), Trusted Hardware, 1991. citeseer.nj.nec.com/ pages 32-37, 1999. perrig91 Sam.html Moulin, H. (1999). Incremental cost sharing; characterization V. Rodoplu and H.-Y. Meng. Minimum energy mobile wire by strategyproofness, Social Choice and Welfare 16, 279 less networks. IEEE Journal on Selected Areas in Commu 32O. nications, 17(8): 1333-1344, August 1999. Moulin, H. and S. Shenker (1997). Strategyproof Sharing of A. Rowstron and P. Druschel. Pastry: Scalable, distributed Submodular Costs: Budget Balance Versus Efficiency, to 25 object address and routing for large-scale peer-to-peer sys appear in Economic Theory. www.aciri.org/shenker/ tems. In Proc. IFIP/ACM Int’l Conf. on Distributed Sys cost.ps tems Platforms, pages 329-350, Heidelberg, Germany, Moulin, Hervé, and Scott Shenker (2001). "Strategyproof November 2001. Sharing of Submodular Costs: Budget Balance versus Effi A. Rowstron and P. Druschel. Storage management and cach 30 ing in PAST, a large-scale, persistent peer-to-peer storage ciency.” Economic Theory 18, 511-533. utility. In Proc. 18th ACM Symposium on Operating Sys J. B. Nagle. On packet switches with infinite storage. IEEE tems Principles, pages 188-201, Chateau Lake Louise, Transactions on Communications, 35(4):435-438, April Banff, Canada, October 2001. 1987. E. Royer and C.-K. Toh. A review of current routing protocols J. F. Nash. Non-cooperative games. Annals of Mathematics, 35 for adhoc mobile wireless networks. IEEE Personal Com 54(2):286-295, September 1951. munications, 6(2):46-55, April 1999. T.-W. J. Ngan, D. S. Wallach, and P. Druschel. Enforcing fair T. Sandholm. Distributed rational decision making. In G. sharing of peer-to-peer resources. In Proc. 2nd Int’l Work Wei?, editor, Multiagent Systems: A Modern Approach to shop on Peer-to-Peer Systems, Berkeley, Calif., February Distributed Artificial Intelligence, chapter 5. The MIT 2003. 40 Press, 1999. Noam Nisan and Amir Ronen. Algorithmic mechanism Tuomas Sandholm. Limitations of the Vickrey Auction in design. Games and Economic Behavior,35:166-196, 2001. Computational Multiagent Systems. In Proceedings of the Nisan, N. (1999). Algorithms for selfish agents, in “Proceed 2nd International Conference on Multi-Agent Systems ings of the 16th Symposium on Theoretical Aspects of (ICMAS). AAAI Press, 1996. Menlo Park, Calif. Computer Science. pp. 1-17, Springer-Verlag, Berlin. 45 S. Shenker. Making greed work in networks: A game-theo Lecture Notes in Computer Science, Vol. 1563. retic analysis of switch service disciplines. IEEE/ACM Nisan, N. and A. Ronen (2000). Computationally Feasible Transactions on Networking, 3 (6):819-831, December VCG Mechanisms, to be presented at “Games 2000.’ http:/ 1995. www.cs.huji.ac.il/-noam/vcgbased.ps S. Singh, M. Woo, and C. S. Raghavendra. Power-aware Noam Nisan and Amir Ronen. Algorithmic Mechanism 50 routing in mobile ad hoc networks. In Proceeding of Design. In Proceedings of the 31st ACM Symposium on MOBICOM 1998, pages 181-190, 1998. Theory of Computing, pages 129-140, 1999. Shapley, L. S. (1953). A value for n-person games, in “Con Osborne, M.-J. and A. Rubinstein (1994). “A Course in Game tributions to the Theory of Games. pp. 31-40, Princeton Theory,” MIT Press, Cambridge Mass. University Press, Princeton. R. Ogier, F. Templin, and M. Lewis. Topology dissemination 55 Sharman Networks. Kazaa Guide: Supernode FAQ, 2003. based on reversepath forwarding, October 2003. www.kazaa.com/us/helpffaq/Supernodes.htm. Vesuvio.ipv6.cselt.it/internet-drafts/draft-ietf-manet-tbrpf J. Shneidman and D. Parkes, “Rationality and Self-Interest in 11.txt Peer to Peer Networks'. In Proc. 2nd Int. Workshop on E. Ogston and S. Vassiliadis. A Peer-to-Peer Agent Auction. Peer-to-Peer Systems (IPTPS '03), 2003, citeseer.nj.nec. In Proceedings of the First International Joint Conference 60 com/shneidman03rationality.html on Autonomous Agents and Multi-Agent Systems (AA Jeffrey Shneidman and David C. Parkes. Using Redundancy MAS), 2002. to Improve Robustness of Distributed Mechanism Imple A. Orda, R. Rom, and N. Shimkin. Competitive routing in mentations, 2003. Working Paper. Poster version to appear multi-user communication networks. IEEE/ACM Transac at ACM Conference on Electronic Commerce EC 03. tions on Networking, 1(5):510-521, October 1993. 65 V. Srinivasan, P. Nuggehalli, C. Chiasserini, and R. Rao. M. J. Osborne and A. Rubinstein. A Course in Game Theory. Cooperation in wireless ad hoc networks. In Proceedings MIT Press, Cambridge, 1994. of INFCCOM 2003, volume 2, pages 808–817, 2003. US 8,874,477 B2 173 174 A. Urpi, M. Bonuccelli, and S. Giordano. Modelling coop ware in Wearable Computing Personal and Ubiquitous eration in mobile ad hoc networks: a formal description of Computing, Springer-Verlag, Vol. 7, No. 5, pages 299-308, selfishness. In Proceedings of WiOpt 03, March 2003. October 2003 A. van den Nouweland, P. Borm, W. van Golstein Brouwers, L. Buttyan, J. P. Hubaux ButtyanI:03a Stimulating Coop R. Groot Bruinderink, and S. Tijs. A game theoretic eration in Self-Organizing Mobile Ad Hoc Networks approach to problems in telecommunication. Management ACM/Kluwer Mobile Networks and Applications, Vol. 8, Science, 42(2):294-303, February 1996. No. 5, October 2003 H. R. Varian. Mechanism design for computerized agents. In Karl Aberer, Philippe Cudré-Mauroux, Anwitaman Datta, Proc. First Usenix Workshop on Electronic Commerce, Zoran Despotovic, Manfred Hauswirth, Magdalena 10 Punceva, Roman Schmidt AbererCDDHPS:03 P-Grid: A New York, N.Y. 1995 Self-organizing Structured P2P System ACM SIGMOD Vickrey, W. (1961). Counterspeculation, auctions, and com Record, 32(3), September 2003 petitive sealed tenders, Journal of Finance 16, 8-37. Olivier Dousse, Patrick Thiran DousseT:03 Physical con Walsh, W. and M. Wellman (1998). A market protocol for nectivity of self-organized adhoc wireless networks IEEE decentralized task allocation, in “Proceedings of the Third 15 Intelligent systems, Vol. 18, Iss. 4, July-August 2003 International Conference on Multi-Agent Systems. pp. Frank Siegemund, Michael Rohs SiegemundR:03 Rendez 325-332, IEEE Computer Society Press, Los Alamitos. vous Layer Protocols for Bluetooth-Enabled Smart J. E. Wieselthier, G. D. Nguyen, and A. Ephremides. Algo Devices (extended version) Personal and Ubiquitous Com rithms for energy-efficient multicasting in static ad hoc puting Journal (PUC), vol. 7, nr. 2, July 2003 wireless networks. Mobile Networks and Applications, Benoit Garbinato, Philippe Rupp GarbinatoR:03b) From Ad 6:251-263, 2001. Hoc Networks to Ad Hoc Applications ERCIM News, no C. Jason Woodard and David C. Parkes, 1st Workshop on the 54, July 2003 Economics of P2P systems, Strategyproof Mechanisms for M. Gastpar, B. Rimoldi and M. Vetterli GastparrV:03 To Ad Hoc Network Formation, 2003, citeseer.nj.nec.com/ code or not to code: Lossy source-channel communication woodard03strategyproof.html, www.sims.berkeley.edu/ 25 revisited IEEE Transactions on Information Theory, 49(5): research/conferences/p2pecon/paperS/S6-Woodard.pdf 1147-1158, May 2003 Z. Zhang and C. Douligeris. Convergence of synchronous and Karl Aberer, Philippe Cudre-Mauroux, Anwitaman Datta, asynchronous greedy algorithms in a multiclass telecom Zoran Despotovic, Manfred Hauswirth, Magdalena munications environment. IEEE Transactions on Commu Punceva, Roman Schmidt, Jie Wu AbererCDDHPSW:03 nications, 40(8): 1277-1281, 1992. 30 Advanced Peer-to-Peer Networking: The P-Grid System Urs Anliker, Jan Beutel, Matthias Dyer, Rolf Enzler, Paul and its Applications PIK Journal February 2003: Special Lukowicz, Lothar Thiele, Gerhard Troester AnlikerB Issue on Peer-to-Peer Systems, April-June 2003 DELTT:03 A Systematic Approach to the Design of Dis Samarjit Chakraborty, Simon Künzli, Lothar Thiele, Andreas tributed Wearable Systems IEEE Transactions on Comput Herkersdorf, Patricia Sagmeister ChakrabortyKTHS:03 CS 35 Performance Evaluation of Network Processor Architec Samuel Bendahan, Giovanni Camponovo, Yves Pigneur tures: Combining Simulation with Analytical Estimation BendahanCP:03 Multi-issue actor analysis: tools and Computer Networks, special issue on Network Processors, models for assessing technology environments Journal of Volume 41, Issue 5, pages 641-665, Elsevier Science, April Decision Systems 2003 Aslan Tchamkerten Tchamkerten:03. On the Discreteness 40 Srdjan Capkun, Levente Buttyan and Jean-Pierre Hubaux of Capacity-Achieving Distributions IEEE Transactions CapkunBH:03. Self-Organized Public-Key Management on Information Theory for Mobile Ad Hoc Networks IEEE Transactions on Stephan ten Brink, Aslan Tchamkerten BrinkT:03 Capacity Mobile Computing, Vol. 2, No. 1, January-March 2003 of the Binary Input Rayleigh Channel with Perfect Side D. Tuninetti and G. Caire TuninettiG:03. The Long-term Information IEEE Transactions on Communications 45 average capacity region per unit cost with application to L. Buttyan, J.-P. Hubaux and S. Capkun ButtyanHC:03 A protocol for sensor networks European Transactions on Formal Model of Rational Exchange and its Application to Telecommunications. Special Issue on Selected Papers the Analysis of Syverson’s Protocol Journal of Computer from the Conference European Wireless 2002. ETTVol 14, Security No. 1, January-February 2003 Marc Heissenbüttel, Torsten Braun, Thomas Bernoulli, and 50 L. Buttyan and J.-P. Hubaux (eds.), G. Avoine, S. Buchegger, Markus Waelchli HeissenbuettelB:04a BLR: Beacon S. Capkun, JY. Le Boudec, S. Vaudenay et al. Buttyan H: Less Routing Algorithm for Mobile Ad-Hoc Networks 03b Report on a Working Session on Security in Wireless Elsevier's Computer Communications Journal Olivier Ad Hoc Networks ACM Mobile Computing and Commu Dousse, Francois Baccelli, Patrick Thiran DousseBT:03a nications Review. Vol. 7, No. 1, January 2003 Impact of Interferences on Connectivity in Ad Hoc Net 55 Karl Aberer, Anwitaman Datta, Zoran Despotovic, Andreas works IEEE/ACM Transactions on Networking Wombacher AbererDDW:03 Separating Business Pro Mario Cagal, Jean-Pierre Hubaux and Christian Enz Cagal cess from User Interaction in Web-Based Information HE:03 Energy-efficient Broadcasting in All-wireless Commerce Electronic Commerce Research, 3 (1-2): Networks ACM Mobile Networks and Applications (MO 83-111, January–April, 2003, Kluwer Academic Publish NET) 60 CS Irena Maravic, Martin Vetterli Maravic V:04a Exact Sam G. Camponovo, Y. Pigneur CamponovoP:02a Analyzing the pling Results for Some Classes of Parametric Non-Band m-business landscape Annals of Telecommunications, limited 2-D Signals IEEE Transactions on Signal Process Hermes, January-February 2003, vol. 58, no. 1-2 ing, Volume 52, Issue 1 January 2004 Jeremy Elson, Kay Römer ElsonR:03 Wireless Sensor Net Christian Plessl, Rolf Enzler, Herbert Walder, Jan Beutel, 65 works: A New Regime for Time Synchronization ACM Marco Platzner, Lothar Thiele and Gerhard Troester SIGCOMM Computer Communication Review (CCR), PlesslEWBPTT:03. The Case for Reconfigurable Hard January 2003 US 8,874,477 B2 175 176 Karl Aberer, Philippe Cudré-Mauroux, Manfred Hauswirth Immune Networking Systems Prof. J.-Y. Le Boudec www AbererCH:02a) A Framework for Semantic Gossiping ...terminodes.org/MV2003-Present/Ma14/ImmuneLeBou ACM SIGMOD Record, December 2002 dec.pdf G. Avoine and S. Vaudenay AvoineV:03a Cryptography Fair exchange with guardian angels G. Avoine www.termin with Guardian Angels: Bringing civilization to pirates— 5 odes.org/MV2003-Present/Ma14/FairExchange Abstract In L. Buttyan and J.-P. Hubaux (Eds.), Report on Avoine.pdf a Working Session on Security in Wireless Ad Hoc Net Spatial analysis of mobile ad-hoc networks under the Signal works, ACM Mobile Computing and Communications to Interference Ratio connectivity model Prof. F. Baccelli, Review (MC2R), Vol. 7. No. 1., 2003, pp. 74-94 Ecole Normale Supérieure and INRIA, Paris www.termi 10 nodes.org/MV2003-Present/Me15/Spacial-Baccelli.pdf Kay Römer, Oliver Kasten, Friedemann Mattern RomerKM: Connectivity and interferences in wireless ad-hoc networks, a 02 Middleware Challenges for Wireless Sensor Networks percolation approach O. Dousse www.terminodes.org/ ACM SIGMOBILE Mobile Computing and Communica MV2003-Present/Me15/OlivierDousseMonteverita.pdf tion Review (MC2R), Fall 2002 Ad-hoc networks: the worst and the average case Prof. R. P. L. Dragotti, S. Servetto and M. Vetterli DragottiSV:02 15 Wattenhofer www.terminodes.org/MV2003-Present/ Optimal filterbanks formultiple description coding: analy Me15/MV-Wattenhofer.pdf sis and synthesis IEEE Transactions on Information www.terminodes.org Theory, 48(7):2036-2052, July 2002 www.google.com/search?&q terminodes R. Karrer and T. Gross KarrerG:02 Location Selection for www.ietforg/html.charters/manet-charter.html. Active Services Cluster Computing 5(3): 265-275: July citeseer.nj.nec.com/cs?cs=1&q manet&Submit Documents 2002 www.google.com/search?&q=%22ad+hoc '%22+and+game M. Vetterli, P. Marziliano, T. Blu VetterliMB:02 Sampling citeseer.nj.nec.com/cs?cs=1&q game--and--%22ad+hoc Signals with Finite Rate of Innovation IEEE Transactions %22 on Signal Processing, 50(6), 1417-1428, June 2002 carmen.cselt.it/idXwg/manet.html Q. Li, B. Rimoldi, and M. K. Simon LiRS:02 Bandwidth 25 Combinatorial Optimization and Auctions Efficient Constant-Energy Trellis-Coded Modulation Buckley, F. and Harary, F. Distances in Graphs. Redwood Schemes with Prescribed Decoding Delay IEEE Trans. on City, Calif.: Addison-Wesley, 1990. Information Theory, Vol. 48, Number 5, May 2002 Garey, M. R. and Johnson, D. S. Computers and Intractabil L. Blazevic, S. Giordano, J.-Y. Le Boudec BlazevicGL:02a ity: A Guide to the Theory of NP-Completeness. NewYork: Self Organized Terminode Routing Cluster Computing 30 W. H. Freeman, 1983. Journal, Vol. 5, No. 2, April 2002 Karp, R.M. “Reducibility Among Combinatorial Problems.” Mark Heitmann, Peter Aschmoneit HeitmannA:02a Cus In Complexity of Computer Computations, (Proc. Sym tomer Centred Community Application Design Interna pos. IBM Thomas J. Watson Res. Center, Yorktown tional Journal on Media Management (4:1. Spring 2002) Heights, N.Y., 1972). New York: Plenum, pp. 85-103, S. Capkun, M. Hamdi, J. P. Hubaux CapkunHH:02 GPS 35 1972. free Positioning in Mobile Ad-Hoc Networks In Cluster Levin, L. A. “Universal Searching Problems.” Prob. Info. Computing Journal, April 2002, Vol. 5, No. 2 Transm. 9, 265-266, 1973. Magaly Dubosson, Alexander Osterwalder, Yves Pigneur Papadimitriou, C. H. and Steiglitz, K. Combinatorial Optimi DubossonOP:02 ebusiness Model Design, Classification zation: Algorithms and Complexity. New York: Dover, and Measurements Thunderbird International Business 40 1998. Review, January 2002, Vol. 44, no. 1: 5-23 Algorithms and Theory of Computation Handbook, CRC K. Aberer, M. Punceva, M. Hauswirth, R. Schmid Aberer Press LLC, 1999, “NP-hard', from Dictionary of Algo PHS:02 Improving Data Access in P2P Systems IEEE rithms and Data Structures, Paul E. Black, ed., NIST. Internet Computing, January/February 2002, pp. 58-67 www.nist.gov/dads/HTML/nphard.html L. Blazevic, L. Buttyan, S. Capkun, S. Giordano, J. P. 45 db.uwaterloo.ca/~alopez-Ofcomp-faq/faq.html Hubaux, J. Y. Le Boudec BlazevicBCGHL:01 Self-Or S. Cook. “The complexity of theorem-proving procedures'. ganization in Mobile Ad-Hoc Networks: the Approach of Proceedings of the 3rd Annual Symposium on the Theory Terminodes IEEE Communications Magazine, June 2001. of Computing, ACM, New York, 151-158. S. Servetto, K. Nahrstedt ServettoN:01 Broadcast-Quality J. Edmonds. Paths, trees, and flowers. Canadian Journal of Video over IPIEEE Transactions on Multimedia, 3(1):162 50 Mathematics, 17, pp. 449-467. 173, March 2001 M. R. Garey, D. S. Johnson. Computers and Intractability, J. P. Hubaux, Th. Gross, J. Y. Le Boudec, M. Vetterli Hu W.H. Freeman & Co, 1979. bauxGLV:01 Towards self-organized mobile ad hoc net L. Levin. Universal Search Problems, Probl. Pered. Inf. 9(3), works: the Terminodes project IEEE Communications 1973. A translation appears in B. A. Trakhtenbrot. A survey Magazine, January 2001. 55 of Russian approaches to Perebor (brute-force search) Evaluation of the Zaurus SL5600 PDA as a platform for algorithms, Annals of the History of Computing, 6(4):384 ad-hoc networking L. Previtali www.terminodes.org/ 400, 1984 www.nada.kth.se/-Viggo/problemlist/compen MV2003-Present/Lu13/Zaurus Previtali.pdf dium.html Simulating large ad-hoc networks with ns-2 V. Naoumov, Algorithm for Optimal Winner Determination in Combinato www.terminodes.org/MV2003-Present/Lu13/Simulating 60 rial.—Sandholm (2001) wireless.fcc.gov/auctions/confer Naoumov.pdf ences/combin2001/papers/sandholmé.pdf Security of emergent properties in ad-hoc networks Prof. V. Bidding and Allocation in Combinatorial Auctions Nisan Gligor, Univ. of Maryland www.terminodes.org/MV2003 (2000) www.cs.berkeley.edu/-christos/games/readings/ Present/Ma14/Gligor-Monteverita.pdf nisanauctions.ps On the self-organization of security in multi-hop networks 65 Integer Programming for Combinatorial Auction Winner— Prof. JP. Hubaux www.terminodes.org/MV2003-Present/ Andersson, Tenhunen, Ygge (2000) www.csd.uu.se/-ar Ma14/Multi-Hop-JPH.pdf nea/abs/.../ps/icmas(0.ps US 8,874,477 B2 177 178 Iterative Combinatorial Auctions: Theory and Practice— Robust Combinatorial Auction Protocol against False-name Parkes, Ungar (2000) www.eecs.harvard.edu/econcs/pubs/ Bids Yokoo, Sakurai, Matsubara (2000) www.keclint ibundle(00.pdf t.co.jp/csl/ccrg/members/yokoo/PDF/aaai2000.pdf CABOB: A Fast Optimal Algorithm for Combinatorial Auc Incentive Compatible Multi Unit Combinatorial Auctions— tions—Sandholm, Suri, Gilpin (2001) www.cs.ucsb.edu/~ Bartal, Gonen (2002) www.cs.huji.ac.il/-noam/icmucaps suri?psdir/cabob.ps Secure Multi-agent Dynamic Programming based on Homo Some Tractable Combinatorial Auctions—Tennenholtz morphic.—Yokoo, al. (2002) www.kecl.ntt.co.jp/csl/ccrg/ (2000) iew3.technion.ac.il/-moshet/bmatchaaai.ps members/yokoo/PDF/aamas2002-secure-wd.pdf Truth Revelation in Approximately Efficient.—Lehmann, Exploring Bidding Strategies for Market-Based Schedul O’Callaghan. (1999) www.cs.huji.ac.il/%7Elehmann/pa 10 ing Reeves, Wellman. (2003) ai.eecs.umich.edu/people/ pers/mechanisms/LCS.ps wellman/pubs/dexter.pdf Preference Elicitation and Query Learning Blum, Jackson, Optimal Solutions for Multi-Unit Combinatorial Auctions.— Sandholm, Zinkevich (2003) www-2.cs.cmu.edu/People/ Gonen, Lehmann (2000) www.cs.huji.ac.il/~lehmann/ avrim/Papers/prefelicitCOLT ps Leibniz/EC00subps 15 An Efficient Approximate Algorithm for Winner. Sakurai, A Combinatorial Auction for Collaborative Planning Yokoo, Kamei (2000) www.keclintt.co.jp/csl/ccrg/mem Hunsberger, Grosz (2000) www.eecs.harvard.edu/grosz/ bers/yukofec()0.ps papers/HunsbergerGro2000-ICMAS.ps On the Sensitivity of Incremental Algorithms for. Kastner, eMediator: A Next Generation Electronic Commerce Hsieh. (2002) www.ece.ucsb.edu/~kastner/papers/incre Server—Sandholm (2002) www-2.cs.cmu.edu/~sand mental alg CA.pdf holm/eMediator.ci.pdf Effectiveness of Preference Elicitation in Combinatorial.— eMediator: A Next Generation Electronic Commerce Hudson, Sandholm (2002) www-2.cs.cmu.edu/~sand Server—Sandholm (1999) siesta.cs. Wustl.edu/~sand holm/elicitation. CMU-CS-02-124.pdf holm/eMediator.trips Constructing and Clearing Combinatorial Exchanges Towards a Universal Test Suite for Combinatorial. Leyton 25 Using.—Smith, Sandholm, Simmons (2002) www-2.cs.c- Brown, Pearson. (2000) robotics.stanford.edu/CATS/ mu.edu/~sandholm/exchange elicitation.pdf CATSps Opportunity Cost Algorithms for Combinatorial Auctions— Combinatorial Auctions for Supply Chain Formation— Akcoglu, Aspines, DasGupta. (2000) www.cs.yale.edu/ Walsh, Wellman, Ygge (2000) www-personal.engin.u- homes/aspnes/auctions-full-ps mich.edu/-wew/Papers/eco0.ps 30 SAM: A Flexible and Secure Auction Architecture Using.— Mutually Destructive Bidding: The FCC Auction Design Perrig, Smith, Song. (1991) paris.cs.berkeley.edu/-perrig/ Problem Mark Bykowsky Robert (1995) www.hss. projects/SAM/sam-eeta.ps.gZ caltech.edu/-ledyard/MDBpap d1.pdf On the Sensitivity of Incremental Algorithms for. Kastner, Combinatorial Auctions with Decreasing Marginal Utili Hsieh. (2002) beijing.cs.ucla.edu/~kastner/papers/PRE ties—Lehmann, Lehmann (2001) www.cs.huji.ac.il/% 35 LIM-eco1-incremental CA.ps 7Elehmann/papers/mechanisms/LLN.ps Using Value Queries in Combinatorial Auctions—Beno Hud Combinatorial Auctions with Decreasing Marginal Utili son Tuomas (2003) www-2.cs.cmu.edu/sandholm/ ties.—Lehmann, Lehmann, Nisan (2001) www.cs.huji. using value queries.acmecposterO3.pdf ac.il/~lehmann/papers/mechanisms/EC01.ps Characterization of Strategy/False-name Proof Combinato Winner Determination in Combinatorial Auction.—Sand 40 rial. Yokoo (2003) www.eecs.harvard.edu/-pesha/Pub holm, Suri, Gilpin. (2002) www.cs.ucsb.edu/~suri?psdir/ lic/111.pdf general-ps On Polynomial-Time Preference Elicitation with Value Que Preventing Strategic Manipulation in Iterative Auctions: ries—Zinkevich, Blum, Sandholm (2003) www-2.cs.cm Proxy. Parkes, Ungar (2000) www.eecs.harvard.edu/ u.edu/People/avrim/Papers/prefelicitACMEC.ps econcs/pubs/iGVA00.pdf 45 Exploring Bidding Strategies for Market-Based Schedul Towards a Characterization of Truthful Combinatorial Auc ing Wellman, MacKie-Mason. (2003) ai.eecs.um tions—Lavi, Mu'alem, Nisan (2003) www.cs.huji.ac.il/- ich.edu/people/wellman/pubs/dexterO2.pdf noam/partial-ps The Effect of False-name Bids in Combinatorial Auctions.— Preference Elicitation in Combinatorial Auctions (Ex Yokoo, Sakurai. (2003) www.keclintt.co.jp/csl/ccrg/mem tended. Conen, Sandholm (2001) www.cs.cmu.edu/~ 50 bers/yokoo/PDF/geb-false-name-2003.pdf sandholm/elicitation.acmec01.pdf Solving Concisely Expressed Combinatorial Auction Prob Reducing Truth-telling Online Mechanisms to Online.— lems—Boutilier (2002) www.cs.toronto.edu/~cebly/Pa Baruch AwerbuchYossi (2003) www.math.tau.ac.il/-azar/ pers/bidcomp.pdf priceps Mechanism Design with Incomplete Languages—Ronen A Combinatorial Auction with Multiple Winners for Univer 55 (2001) robotics.stanford.edu/-amirrfincomlete ang7.pdf sal. Kelly, Steinberg (1998) www.statslab.cam.ac.uk/~ Extremal Behaviour in Multiagent Contract Negotiation— frank/AUCTION/ks-16mar00ps Dunne (2004) www.csc.liv.ac.uk/research/techreports/ A Combinatorial Auction with Multiple Winners for Univer tr2004/ulcs-04-012.pdf sal. Frank Kelly Richard (1998) www.statslab.ca An Iterative Generalized Vickrey Auction: Strategy-Proof m.ac.uk/-frank/AUCTION/auction.pdf 60 ness. Parkes (2001) www.eecs.harvard.edu/econcs/ An Approximate Truthful Mechanism for. Archer. (2003) pubs/ivickrey01.pdf www.cs.cornell.edu/People/eva/papers/soda03.pdf Preference Elicitation in Proxied Multiattribute Auctions— Multiple-Issue Auction and Market Algorithms for the.— Sunderam, Parkes (2002) www.eecs.harvard.edu/econcs/ Teich, Wallenius. (1999) www-ec.njit.edu/-bartel/Nego pubs/proxysunderamps Pap/MultipleIssueAuctions.pdf 65 Price-Based Information Certificates for Minimal-Revela Costly Valuation Computation in Auctions—Larson, Sand tion. Parkes (2002) www.eecs.harvard.edu/-parkes/ holm (2001) www.cs.cmu.edu/~sandholm/tark01.pdf pubs/certificates.pdf