Factory Radio Communications

rf indoor collllllunicafions ______

Noise and propagation measurements reveal/imitations for UHF/microwave i'ndoor radio communication systems

By Theodore S. Rappaport Virginia Polytechnic Institute and State University A glimpse into a typical factory re· any type of tether will require a radio In Japan, spectrum has been set aside veals a high degree of automation has system for control. Optical systems are for 300 mW, 4800 bps indoor radio .entered into the work place. Computer viable, but become inoperative when systems operating in the 400 MHz and driven automated test benches, wired obstructed. Furthermore, radio systems 2450 MHz bands (4). guided robots and PC-controlled drill will be us~ful for quickly and cheaply Accurate characterization of the oper presses are a few examples of the connecting often moved manufacturing ating channel is a mandatory prerequi proliferation of computer technology equipment and computer terminals. Ra site for the development of reliable and automation in manufacturing. The dio will also accommodate reconfig wideband indoor radio systems. Radio boom in automation has created a need urable voice/data communications for channel propagation data from factory for reliable real-time communications in other facets of factory and office building buildings have been· made available for factories. In 1985, the Manufacturing operation and may eventually be used the first time through a research pro Automation Protocol (MAP) networking in homes and offices to provide univer gram sponsored by NSF and Purdue standard was established by manufac- sal digital portable communications (1 ). University. As shown here, it is not . turing leaders to encourage commer Presently, communications· between environmental noise, but rather multi cialization of high data rate communica computers and automated machines are path propagation that limits the capacity tions hardware for use in computer conducted almost exclusively over ca of a radio link. The severity of multipath controlled manufacturing. MAP is capa ble. Narrowband radio systems (with is largely dependent upon factory inven ble of supporting 10 megabit per second data rates less than 10 kbps) are tory, and building structure and age. (Mbps) data rates for short periods. MAP currently used in some factories for networks (and TOP networks in office dedicated control of overhead cranes Factory Noise buildings) rely on coaxial cable or fiber and wire-guided vehicles and for inven Although much of the radio · noise optic cable to interconnect users. tory control. While narrowband systems encountered in factories arises from Twisted-pair interconnection of com are well-suited for human operation and weak emitting sources, measurements puter terminals is also commonly used. simple communications, it is anticipated have revealed that some types of indus This article discusses the use of radio that for a moderately sized AGV fleet trial equipment emit harmonic RF en links for communication in modern fac (greater than 25 vehicles), data rates of ergy and can radiate substantial noise tories. several hundred kbps will be needed to up to several hundred megahertz (8). accommodate real-time computer con Equipment such as RF-stabillzed arc he method for transporting parts-in trol and navigation of CIM-integrated welders, induction heaters and plastic Tprocess in a futuristic, but realistic, AGV systems employing multiple-ac bonders are acute sources of noise. just-in-time (Jin manufacturing environ cess radio networking. Although interference i~ significant at ment is one of the areas of research at In the United States, the FCC has HF and VHF, noise signatures of such the National Science Foundation (NSF) allocated. spectrum for narrowband in equipment fall off rapidly above 1 GHz Engineering Research Center (ERC) for dustrial radio communications in VHF (8). Recent empirical measurements Intelligent Manufacturing Systems. Analy (450 MHz) and UHF (900 MHz) bands. have confirmed that typical machine sis has shown that an inexpensive, agile More recently, the FCC has authorized generated noise levels in operational mobile robot fleet, capable of navigating the use of suitably designed spread factories are much less severe at higher I without any type· of pre-made track, spectrum systems for 900 MHz, 2400 frequencies (9). Figure 1 shows results could easily accommodate the type of MHz and 5725 MHz (3). If transmitters of peak noise power spectrum measure material flow required for a JIT manufac- meet with FCC approval, unlicensed 1 ments. made along an engine manufac . turing system. A truly autonomous W transmitter power levels may be used turing transfer line in full operation. guided vehicle (AGV) that does not use-- over bandwidths Qreater than 25 MHz; Worst-case noise levels are seen to be --- -~j- RF Design ...... ------~------~------..-

rl indoor co~nlilunicafions ______--+-

40 dB lower at UHF/microwave frequen cies and are only 5 dB above the thermal ·20.00 ·20.00 noise floor. These results are encourag ing and indicate that noise will not 'E severely hamper most factory radio -40.00 'E ·40.00 I :g.CD systems operating at UHF and above.

I ·80.00 ·60.00 D. ~0 D. Multiple-Access Networking '2 ... 1 & L....al ..&. -L Considerations ·80.00 .i!: ·80.00 I ~ Future factory. radio communication a: systems will rely on multiple-access ·100.0 ·100.0 27.00 87.00 147.0 207.0 287.0 327.0 1.2 1.3 1.4 1.5 1.6 1.7 techniques to accommodate many fixed Frequency (MHz) Frequency (GHz) and portable/mobile terminals. Multiple access techniques such as frequency a) 27 MHz to 327 MHz (VHF) bl 1200 MHz to 1700 MHz lUHFl division multiple-access (FOMA), time division multiple access (TOMA), code Figure 1. Noise power spectrum measured 4 m from engine cylinder division multiple-access (COMA, also machining line (9). called spread-spectrum) and carrier sense multiole-access (CSMA. also Uctlit'Y fJI::tliM1i rc:tUIUJ f,Jc:ti ll\ltln Ur.i:tilnei users Into non-overlapping signal spaces. Because of non-Ideal condi tions, ~owever, there Is Inevitable over lapping of signals and the resulting co-channel interference can appear as lengthy message delays or as degrada tion of the desired received signal. Random access (CSMA, COMA) radio local area networks are attractive be cause they have relatively. few synchro nization requirements. Unlike fixed as signment networks (FDMA, TDMA) which assume all users require their own channel, random access techniques rely a) Transmitter b) Receiver on "bursty users" and assume that the likelihood of many users· using the network at one time Is small. For AGVs using reliable on-board dead-reckoning Figure 2. Block diagram of fac.tory multipath. measurement system. systems, infrequent position updates

60~------~~~~ 99 ·.:·LOS light •••• •• n<=4 80 • OISS tight clutter, Site C x LOS heavy ..... ··· lOS heavy clutter, Site E 50 A LOS wall LOS light clutter, SiteD 50 * OBS light .. ···········• •' • OBS heavy ,•' ...... ······· 20

All Sites ····:· ...... ········ 40 ...... ~ 10 .. ·· ..... ·;..

% Prob111blllly Signal ... ······< ...... ~~~ .., ** ,... .·><····· Signal Level Attenuation 30 n= 2.18 .. ·· * ...... ' ' .!!"-·· n=2

" ... / ::::.:.· .. ··.~*:·~~~---~····t··t .....~-~! ...~·········;:;· 0.5 10 ..•...... ····ei .. ·· ····· ······ 0.2 ····· ·········

·20 ·10 10 10 20 50 100 Distance (m) SIIJnal Level (dB about median)

Figure 3. Large scale signal attenuation at all Figure 4. Cumulative distributions for three measure measurement sites. ments and their fit to various distributions. .

68 January 1989 ... fRE UENCY SYN HESIS

11 c ..nk Silo" Woolt Uno In Silo F 10

.. ~ lloftlll#

·10 0 •• ·11 ~!~0----2~1--~50--~7~1--~100- .•••• lloYioltil m -- fUc:IM K•t0d8 TIME (o)

Figure 5. Temporal fading meas Figure 6. Cumulative signal level urement in engine assembly area distributions for temporal fading across crankshaft wash line (Site data In various factory location$. F). T·R separation is 25 m.

suggest the use of random access high resolution graphics, maps) are networks for AGV control. On the other accommodated best by a fixed assign hand, direct remote control of a fleet of ment network. Selection of networking vehicles by a central dispatching station strategies of radio depends heavily upon (such as might be warranted for cleaning the number of users, the duration of a contaminated nuclear reactor) would transmissions, the limit of sophistication warrant a fixed assignment scheme. at each terminal, and the importance of At the ERC for Intelligent Manufactur real-time control. ing Systems at Purdue University, con siderable progress has been made in Multlpath Propagation determining realistic limitations on the Due to the large metal content of a delay characteristics of packet radio factory, multipath interference is created networks (10). Also, a powerful product by multiple reflections of the transmitted · form solution model for packet radio signal from the building structure and systems has been developed .(11 ). surrounding inventory. The resultant CSMA and COMA strategies can be received waveform is a sum of time- and merged to enhance multiple-access com frequency-shifted versions· of the origi munication performance in a multipath ·nal transmission and, depending on environment while providing some rang parameters of the signal and channel, for ing capability (1'2, 13). In Reference 12, the received signal may be greatly fundamental expressions that permit the distorted. eEW calculation of bit error rates in packet Historically, multipath has been iden eATE spread-spectrum systems have been tified as the most limiting factor in eNMR provided; these expressions permit sys portable radio communication systems. e MSK, FSK tem designers to analyze throughput For narrowband factory radio systems, supports many and delay as functions of the number of multipath causes large fluctuations (fad simultaneous users. ing) In received signal levels due to . other acronyms As the number of users increases, the tt;tmporal variations of the channel and Specialized real-time communications capability of the receiver. Additional signal loss will Synthesizers random access techniques diminishes, occur when an AGV is shadowed by for -..-~ and a fixed assignment approach is inventory. and equipment. In wideband required. Furthermore, if central com systems, the scatterers create intersym the puters using parallel processing archi- bol interference and cause the channel OEM. . tectures are required to simultaneously to be frequency selective. Consequently, communicate, navigate and control the maximum data rate supported by a many simultaneous users on a virtually multipath channel is limited. Typical Sciteq Electronics, Inc. fading channels require 30 dB more 8401 Aero Drive continuous basis, TDMA or FDMA ap .. San Diego, California 92123 preaches will be desired. Portable/ transmitter power to achieve low bit (619) 292-0500 mobile users transmitting large blocks error rates (1 0-4) compared to non Fax: (619) 292-9120 of data (i.e., MIS, video transmissions, fading systems. INFO/CARD 52 RF Design- 69 rl indoor colllntunicafion•------JL-

Path Loss Exponents In Different Factories Shadowing Effects of Common Factory Equipment Factory 0 No. of Corr. Attenuation Site n (dB) Points Coef. Obstacle Description (dB) Site B 2.39 10.20 33 .94 2.5 m storage rack with small metal parts Site C 1.89 5.55 41 .98 (loosely packed) 4-6 SiteD 2.43 7.94 34 .96 4 m metal box storage 10-12 Site E 2.12 8.03 18 .96 Site F 1·.92 4.79 17 .98 5 m storage rack with paper products (loosely packed) 2·4 Table 1. Path loss exponent as a function of factory 5 m storage rack with paper products site. (tightly packed) 6 5 m storage rack with large metal parts Path Loss Exponents In Different Factory Geographies (tightly packed) 20 Factory o No. of Corr. Typical N/C machine 8-10 Geography n (dB) · Points Coef. Seml .. automated Assembly Line 5-7 LOS light clutter 1. 79 4.55 26 .98 I n~ hAAW r.luttAr 1. 79 4.42 43 .98 0.6 m square reinforced concrete pillar 12-14 LU~ along wau 1 .4~ ~.~ U .~d btainie~~ ~teei ~iping ¥or Cook-Cooi OBS light clutter 2.38 4.67 23 .99 Process 15 OBS heavy clutter 2.81 8.09- 43 .97 Concrete wall 8e15 All Geographies* 2.18 7.92 135 .96 *LOS measurements along wall not Included In computation. Concrete floor 10 Table 2. Path loss exponent. as function of factory Table 3. Shadowing effects of some common factory topography. equipment.

To determine propagation character scope the attenuated, distorted and signal strengths relative to a reference Istics inside factory buildings, radio delayed versions of the pulse. The measurement (made at 1o wavelength wave propagation experiments at 1300 measurement apparatus consisted of a distance) over various topographies at MHz were made in five fully operational periodic pulse-modulated transmitter all five factories. Tables 1 and 2 Indicate factories (14). Over 30,000 narrowband with a peak output power of 1 W. The that in all cases, received power and T-R fading measurements and 950 wide receiver consisted of a low noise ampli separation are highly correlated, thus band impulse response measurements fier followed by a square law envelope confirming that equation 1 Is a valid were made in a diverse collection of detector and a 350 MHz digital storage model for factory radio systems .. Further industries and building structures. Fac oscilloscope. A directional coupler al more, although signals attenuate more tories that participated in the research lowed OW envelope. measurements to rapidly with distance in obstructed include l~ading engine and automobile be made simultaneously by a modified topographies, the worst case attenu parts manufacturers and dry-foods pro communications receiver. The measure ation Is not as severe as In partitioned ducers. An overview of the experiments ment system block diagram Is shown In homes and office buildings (18,.19). This . and the measurement equipment are Figure 2. Wldeband dlscone antennas is due to large ceiling expanses, wide given in (15). Briefly, four distinct to· (16) were used at both transmitter and aisles, and metal ceiling truss work and pographical settings were identified in receiver. inventory that readily facilitate multiple each of the five factories. Topographies CW measurements revealed that path paths. ranged from line-of-sight (LOS) trans loss falls off exponentially with distance Because accurate descriptions of path mission paths along lightly cluttered and is described by a log-normal distri obstacles were kept during the factory aisles to heavily cluttered, obstructed bution about a mean path loss law given measurements, it Is possible to extract paths between adjacent aisles. In each by: from the data the RF signal loss caused topography, three measurement loca by typical factory surroundings. By com .. tions were selected, with graduated Path loss(d)oc d" (1) paring the. received signal levels for transmitter·r~ceiver separations be shadowed locations with the ensemble tween 20 and 80 m. The transmitter was where d is the distance between trans average of the factory measurements, positioned in the clear in the center of mitter and receiver In meters and n Is shadowing losses have been computed. the particular topography while the re· the mean path loss exponent found by Table 3 indicates typical shadowing ceiver was moved along 1 m tracks at usjng a linear least squares fit to .a losses that can occur when a receiver each measurement location. · scatter plot of measured path loss Is placed directly behind an obstruction Wideband channel impulse responses against measured distance (17). This (deep shadowing). The data reveal that were measured in the time domain by model has also been found to hold at knife-edge diffraction theory is pessimls .. repeatedly transmitting a 10 ns pulse UHF for channels Inside and around tic;· deeply shadowed locations experl~ and receiving on a digital storage oscillo- houses (2). Figure 3 shows the received ence received signal levels. consistently

70 Januaty 1989 5 to 20 dB larger than predicted by the LOS path. Energy density a'ntennas mitter or receiver is shadowed by large knife-edge diffraction. For long paths that couple both electric an~ magnetic equipment or by stacks of inventory, the with obstructions located In the middle fields may also be useful in combatting predominant energy arrives 50 to 150 of the path, knife-edge diffraction is in multipath fading (20). ns after the first observable signal (2, 14). · better agreement with the empirical Factory channel impulse response To determine how individual signal com data. measurements (also called power delay ponents change with receiver motion, Measurements made with a moving profiles) reveal that for LOS paths there 19 equally spaced power impulse re CW receiver over many local areas typically exist only a few specular multi sponse measurements were made along reveal that fading is usually Rayleigh in path components, with the direct path 1 m tracks throughout five factories. heavily cluttered LOS and lightly clut having a larger signal level than the Figure 8 illustrates how specular reflec tered obstructed topographies, Rician latter components. Over obstructed tions from perimeter walls, etc. are for paths along perimeter walls and over paths, however, when either the trans- easily distinguishable. In Figure 9, typi- lightly cluttered LOS paths, and log normal for paths that traverse heavily cluttered obstructed topographies. Fig ure 4 illustrates some of these typical fading distributions and their fit to some of the observed fading data. · Additional CW measurements were made to determine the temporal vari ation of factory channels caused by motion of personnel and work in pro gress. The transmitter and receiver were :~1 positioned to traverse busy assembly lines and main aisles. Figure 5 shows a LPF1 - typical measurement and the corre !~~J~i~l~1~·1 sponding chan_ges in signal strength GB1 - over several seconds. Figure 6 illus trates that temporal fading between ······-·.:.·I)_·-~1 •.. fixed radio terminals is well described T081 by a Rician distribution having K 10 dB. Measurements made over identical paths with different receiver antenna heights reveal that received signal strengths are often highly correlated Ul~ (not independent) for separations of 2 F P1 p..;:i:;;"F wavelengths (0.5 m). As seen in Figure 'l 7, however, close-spaced antenna diver ,::~·:~:~.! sity may be useful when antennas are TFM1 located in a horizontal plane in line with INFO/CARD 53

Finally, a synthesized signal generator An intelligent microprocessor-controlled/ that's designed to help you generate more programmable memory, for example, than just signals. Introducing the new remembers up to 10 of your most com Ramsey RSG-10. The signal generator you monly used test set-ups. And unlike other can afford, the signal generator that's units, just one touch of the memory designed to do what a signal generator is exchange button is all it takes to quickly meant to do-provide a very stable, accu shift from one test set-up to another. You rate, easy to control signal. gef more work done, easier and faster. For the price, you wouldn't expect any At $2495, the new Ramsey RSG-10 is more. But, there is more to the RSG-10. your profit generator. I===-- =·-=====I RAMSEY ELECTRONICS . :: ..li !\:,;::-:I{ y 2575 BAIRD ROAD, PENFIELD, NY 14526 =-=::!!!i":i:::!i::=!!!!:F= = FAX: (716) 586·4754 TELEX: 466735 RAMSEY Cl

RF Design INFO/CARD 54 71 rl indoor collllftunicafions______----l-_

a situation arises In an indoor radio - 2 m •ntenn~~ height communication system. In Reference 6, ...... 1.5 m •nten111 height 20 empirical measurements were pre sented that indicate that channel delay spread can change by as much as 180 . 10 Re'-tlve ns over a few centimeters of receiver ..• Received movement. .n SIQNII 0 I Strength Conclusion Extensive measurement, characteri I .M (dB) ·10 zation and modeling of Indoor factory .H radio channels have been carried out. The work here r~veals that manmade noise Is not a serious problem at -Dotoy("'l frequencies greater than 1 GHz, and 0 0.5 Dlaplacement (m) that fading characteristics are highly dependent upon local topography in the Figure 8. Closely spaced power workplace. Shadowing data and large Impulse response measurements In a factory building. Note the ".-..-.-.1"•--' ..-.1~--l -~"'"'- scale path loss models have been UtiV'IoiiVt-'""''-' diiU IVIIII llllloi UCIIOIO lVI Ug~ -...... ~www IW'VWWWt"""'••"""u"'"'"' ~""'"" ..,...; 'w.-..~~ lope using height diversity (1.5 m signing reliable narrowband indoor radio reflections are easily identifiable and 2.0 m antenna heights at systems for portable communications (14). receiver). and AGV control. Wideband measure ments reveal that current technology ~ER Performance of High Speed Digital cal multipath power delay profiles are limits data rates to about 50 to 100 kbps; Mobile Transmission," 1988 IEEE Veh. presented. While this accommodates current needs, Tech. Conf. Proc., June 15, 1988,. pp. One. measure of multi path conditions it is anticipated that greater capacity will 119-126. . is the; RMS delay spread a which is be required for the highly automated · 8. E. Skomal, Man-Made Radio Noise, inversely proportional ~o the maximum and flexible factories of the future. Van Nostrand Reinhold Co., NY, 1978, usable data rate of a channel. RMS Ongoing work at Virginia Tech Is aimed chapters 3 and 4. delay spread Is computed as the square at developing robust wideband commu 9. L. Duran, "Characteristics of Elec root of the second central moment of the nication system designs for Indoor radio tromagnetic Interference In Industrial power delay profile. A food processing communications. · liiJ Environments," Masters Thesis In Elec .. factory that manufactures dry goods and trlcal Engineering, Purdue· University, has considerably less metal than the References December 1987. · · other factories had multlpath delay 1. D. Cox, ~~universal Portable Radio 10. S. Beuerman, E. Coyle, "The Delay spreads consistently half of those ob Communications," IEEE Trans. Veh. Characteristlcs.of CSMAICD Networks," served in metalworking factories. The Tech., vol. VT-34, no. 3, pp. 117-121. IEEE Trans. Comm., vol. 36, no. 5, May worst case RMS delay spread was 300 2. D. Cox, H. Arnold, P. Porter, 1988, pp. 553-563. ns in a modern engine plant; typical "Universal Digital Portable Communica ! 11. R.L Hamilton, Jr., E. Coyle, "A values are between 100 ns to 200 ns. tions, A System Perspective," IEEE Two-Hop Packet Radio Network with Unlike In office buildings (19), delay Trans. J. Sel. Areas Comm., vol. SAC-5, Product Form Solution," Proc. 20th spread values in factories do not depend no. 5, June 1987, pp. 764-773. Annual Cont. on Info. Sciences and on whether or not there exists an LOS 3. M. Marcus, "Regulatory Policy Systems, Princeton Univ., March 1986, path. Considerations for Radio Local Area pp. 871-876. In Reference 21, It was shown that the Networks," IEEE Communications Maga 12. R. Morrow, Jr., J. Lehnert, "Bit-to coherence (flat fading) bandwidth over zine, vol. 25, no. 7, July 1987, pp. 95-99. Bit Error Dependences in Slotted DS/ Rayleigh fading channels using DPSK 4. H. Nogami, M. Wakao, "Premises SSMA Packet Systems with Random modulation is approximately 1/50th of Radio Systems in Japan," IEEE Com Signature Sequences," accepted to the reciprocal of RMS delay spread. For munications Magazine, vol. 25, no .. 10, IEEE Trans. Comm., to appear early factory radio channels, this suggests October 1987, pp. 49-55 .. 1989: . worst-case maximum data rates of 33 5. J.C-1 Chuang, N. Sollenberger, 13. J. Fischer, J. Cafarella, c. Bouman, kbps using simple receivers, with higher ••Burst Coherent Detection with Robust G. Flynn, V. Dolat, R. Boisvert, "Wide capacity coming about from distributed Frequency and Timing Estimation for' Band Packet Radio for Multipath Environ· antennas, adaptive equalizers and di Portable Radio Communications," to ments," IEEE Trans. Comm., vol. 36,· versity techniques. Recently, rapid appear in 1988 GLOBECOM Proc. no. 5, May 1988, pp. 564-573. changes in channel group delay have 6. T. Rappaport, ••Delay Spread and 14. T. Rappaport, "Characterizing the been found to cause burst error in digital Time Delay Jitter in Manufacturing Envi UHF Factory Multlpath Channel," Ph.D. communications systems due to shifts ronments,'' 19881EEE Veh. Tech. Conf. Dissertation, Purdue University, Decem In eye pattern timing (22,23). This phe Record, Philadelphia, PA, June 15, ber 1987. Also available as Tech. Report nomenon, known as jitter, becomes . 1988, pp. 186-189. TR-ERC88-12, NSF Engineering Re increasingly important as operating band 7. T. Takeuchi, F. lkegaml, S. search Center for Intelligent Manufac widths are increased while the fading Yoshida, N. Kikuma, "Comparison of turing Systems, Purdue University, June rate of the channel remains small. Such Multipath Delay Characteristics with 1988.

January 1989 B-D CRYSTAL subsidiary of H.• -Q lntt·rnatimml

Precision Quality I Fast Delivery Quartz Crystal Products For: l•k] ..L •cRYSTAL · • COMMUNICATION • flO ... • 100 ... L_ • CHANNEL ELEMENTS -DWI't(ftol ...... • PAGERS • TEST INSTRUMENTS • OSCILLATORS • CRYSTAL CLOCK ~L • TCXO ...... • vcxo ..lliCIIUDI1MC•I -...... FILTERS • MONOLITHIC Figure 9. Typical multipath power impulse responses computed as a • DOSCRETE spatial average of 19 measurements over a 1 m track. 15. T. Rappaport, C. McGillem, .. Char- .....------, 48 HR. SERVICE J . acterizing the UHF Factory Multi path About the Author Availabl···e For Crystals Channel,'' Electronics Letters, Sept. 10, Dr. Ted Rappaport is assistant 6Q6-283-5000 1987, 23(19), pp. 1015-1 017. professor at the Bradley Depart- Fax 606-283-0883 16. T. Rappaport, 11Simple-to-Build . ment of Electrical Engineering, Vir- Wide-Band Antennas," RF Design, April ginia Polytechnic Institute and State 8-D CRYSTAL ENTERPRISES 1988 35-41 University, Blacksburg, VA 24061; Tel: 1438 Cox Avenue 17. !f.~appa~rt, c. McGillem, ~~uHF (703) 231-6834. He is also president Erlanger, Kentucky 4101.8 Fading in Factory. Buildings,'' 1988 of TSR Technologies. INFO/CARD 56 URSI/National Radio Science Meeting, January 5, 1988. 18. S. Alexander, .. Radio ·Propagation Analog Circuit Simulation Within Buildings at 900 MHz," Elec NEW IS_SPICE/386 On 386 PC's, $386 tronics· Letters, vol. 19, no. 20, Sep Outperforms Workstations tember 1983, p. 860. Increases Spel:ld by 200 -· 600o/o 19. D. Devasirvatham, .. Time Delay Circuit Size nearly Unlimited Spread and Signal Level Measurements Supports 80287,80387, Weitek 1167/3167 RF SWITCI:t of 850 MHz Radio Waves in Building SPICE, a world wide standard for analog circuit simula tion runs on all 80766 PC's In real mode as IS_SPICE, Environments," IEEE Tr"ans. Ant. Prop., for only $95.00: Performs AC, DC and Transient, Noise, vol. AP-34, no. 11, November 1986, pp. Distortion, Fourier and Sensitivity Analysis. 1300-08. 20. K. Ito, S. Sasaki, 11A SmElll Printed Antenna Composed of Slot and Wide Strip for Indoor Communication. Sys tems," 1988 IEEE Antennas and Propa gation Symposium, June 12, .1988, pp. 716-719. 21. R. Bultitude, ~~Measurement, Char acterization and Modelling of 800/900 · Model A64 fs an ultra wide band MHz Mobile Radio Channels for Digital PIN diode solid state switch for trans fering both low and high level signals .Communications, Ph.D. Thesis, Carle with negligible distortion, high isola ton University, Ottawa, February 1987. SPICE_NET, $295: Schematic entry for ANY SPICE tion, and 111ini111Um loss. 111 simulator. Automatically makes a Complete SPICE 22. J.C-1 Chuang, The Effects of netlist and places output waveforms on your Switch Type: SPOT schematic. Time Delay Spread on Portable Radio Freq. Range: 1-500 MHz Communications Channels with Digital PRE_SPICE, $200: Adds Monte Carlo Analysis, Parameter Sweeping, Optimization and Algebraic loss: .5 dB 111ax. Modulation," IEEE Tr"ans. Veh. Tech., Parameter Evaluation. An easy to use Menu Drive vol. VT-36, no. 3, August 1987, pp.135- Program and Extensive Model Ubraries are Included • Isolation: -60 dB min ~ 300 MHz . 140. lntu_Scope, $250: A graphics post processor that VSWR: 1.3:1 max. (1.1:1 max. 5-500 MHz) works like a digital oscilloscope. Easy to ~ 23. S. Ariyavisitakul, et.al., ~~Fractional use with all the waveform operations you -- Quantity and O.E.M. pricing Bit Differential Detection of MSK: A will ever need. intusoft Scheme to Avoid Outages Due to Fre quency-Selective Fading," IEEE Tr"ans .. . For Information, (213) 833-0710 ' Please Write or Call WIDE BAND ENGINEERING .COMPANY, INC. Veh. Tech., vol. VT-36, no. 1, February P.O. Box 6607 All Programs come P.O, Box 21652, Phoenix, AZ 115036 U.S.A. San Pedro, CA with a 30 Day Money Telephone (602) 2511-1570 1987, pp. 36-42. 90734-6607 Back Guarantw - INFO/CARD 49 INFO/CARD 57 RF Design ' 73