Copyright Warning & Restrictions
The copyright law of the United States (Title 17, United States Code) governs the making of photocopies or other reproductions of copyrighted material.
Under certain conditions specified in the law, libraries and archives are authorized to furnish a photocopy or other reproduction. One of these specified conditions is that the photocopy or reproduction is not to be “used for any purpose other than private study, scholarship, or research.” If a, user makes a request for, or later uses, a photocopy or reproduction for purposes in excess of “fair use” that user may be liable for copyright infringement,
This institution reserves the right to refuse to accept a copying order if, in its judgment, fulfillment of the order would involve violation of copyright law.
Please Note: The author retains the copyright while the New Jersey Institute of Technology reserves the right to distribute this thesis or dissertation
Printing note: If you do not wish to print this page, then select “Pages from: first page # to: last page #” on the print dialog screen
The Van Houten library has removed some of the personal information and all signatures from the approval page and biographical sketches of theses and dissertations in order to protect the identity of NJIT graduates and faculty.
TOKEN BUS LAN PERFORMANCE: MODELING AND SIMULATION
y AMES KU OO
esis su mi e o e acu y o e G a ua e sc oo o e ew e sey I s i u e O ec o ogy i a ia u i me o e equi eme o e eg ee o Mas e o Scie ce i E ec ica E gi ee i g a ua y 199 APPROVAL SHEET
Title of Thesis: Token Bus LAN Performance: Modeling and Simulation.
Name of the Candidate: Ramesh Kurnool Master of Science in Electrical Engineering, 1990.
Thesis Approved :
Dr. Anthony Robbi Date Associate Professor Electrical Engineering
Dr. C. N. Manikopoulos Date Associate Professor Electrical Engineering
Dr. Irving Wang Date Associate Professor Electrical Engineering I A
ame ames Ku oo
eg ee a a e o e Co e e M S E E 199
Seco a y E uca io Go e me u io Co ege Ku oo A I ia
Co egia e I s i u io s A e e a e eg ee
ew e sey I s i u e o ec o ogy 9/ 7-1 / 9 M S E E
Co ege o E gg U i e si y / -11/ E
Ma o E ec ica E gi ee i g
Mi o Com u e Sys ems ABSTRACT
i e o esis oke us A e o ma ce Mo e i g a Simu a io ames Ku oo Mas e o Scie ce 199 esis i ec e y o esso A o y a i
A simu a io mo e ase o CSIM a ocess o ie e simu a e a guage o a a y e e e o ma ce o e oke us o oco is e e o e e o ma ce measu es suc as oug u a e age e ay a ma imum e ay e acke a e ese e Sys em e o ma ce is a a y e o i e e oa s um e o s a io s e wo k e g s i e e ysica a ogica is i u io o e s a io s wi acke e g as a a ame e e ious s u ies we e ase o e e ay- oug u a a ysis wi o iscussio o e e ec o a ia io o e ogica a ysica is i u io o e s a io s o e e o ma ce o e mo e w ic is o e i e ese esis e oa is o e e o e e wo k i e o m o a s eam o a a acke s wi u i o m y is i u e i e -a i a imes A com a iso o e oke us mo e wi a o a CSMA/C mo e s ows a e ysica is i u io o e s a io s as a mi imum e ec o e e o ma ce o e mo e i e case o e oke us mo e u as a co si e a e e ec o a o e CSMA/C mo e ACKNOWLEDGMENTS
I wish to express my sincere gratitude to Dr. Anthony Robbi for his valuable guidance, inspiration and encouragement during the entire course of this thesis. I am deeply indebted to Mrs. Perkins, Miss. Sandra Huffman, Miss.
Susana Saldarriaga and Mr. Bill Bowman at Document Preparation Centre for all their willful help and support. Furthermore, I would like to acknowledge the
Computer Aided Engineering Laboratory and the staff for providing the resources vital for the thesis work. Last, but certainly not the least I am indebted to my room partners Mr. Pratap Pasupuleti and Mr.Nanda Kanuri and my parents for their constant moral support, valuable guidance and encouragement. TABLE OF CONTENTS
CHAPTER Page I INTRODUCTION 1 II PROTOCOL AND MODEL DESCRIPTION 11
III SIMULATION DESIGN AND EXPERIMENTS 26 IV SIMULATION RESULTS 36 V COMPARATIVE PERFORMANCE WITH CSMA/CD MODEL . . . . 54 VI COMPARATIVE RESULTS FROM ANALYTICAL AND SIMULATION STUDIES 62 VII CONCLUSIONS 66
APPENDIX A
CSIM AND SOURCE LISTING OF SIMULATOR 68 REFERENCES 81 1
CHAPTER I INTRODUCTION
The design, installation, and operation of computer networks is vital to the functioning of modem computerized organizations. Over the past decade, complex and diverse networks have been established, trying together mainframes, minicomputers, personal computers, terminals, and other devices, such as communications controllers and cluster controllers [1]. Many of today's networks use public telecommunications facilities to provide users with access to the processing capabilities and data storage facilities associated with the mainframes and to permit fast interchange of information among the users of the network. As the cost of micro electronic devices has dropped, the intelligence in the various devices that are attached to the network has increased. Intelligent terminals, minicomputers, personal computers, and other programmable devices are all part of these large networks. These networks
are called wide area networks (WANS). In parallel with the growth of wide area networks, there has been another area of expansion in the use of computing facilities. Personal computers have spread rapidly and widely throughout organizations. As the use of personal computers has grown, so has grown a need for these personal computers to communicate-with each other and with the larger, centralized data processing facilities of the organization. In order to make this feasible a type of networking technology known as local area network (LAN) has been developed. A local area network is a communications network that provides i e co ec io o a a ie y o e ices wi i a sma a ea ese a a commu ica i g e ices a e o e ca e as o es o S a io s [5] A e u o is eca e e i s a e ase wo wi e s ou e i e cess o wo mi io A s I is e e o e ecomi g mo e a mo e im o a o e wo k esig e s o e a ua e e a i i y o A s a use a ious commu ica io o oco s o a e e esi e message oa s um e o s a io s a esi e e g o e e wo k I is u e im o a o a e a ge e a sc eme o mo e i g a simu a io o ese o oco s w ic ca e use o assess e e o ma ce o ew o oco s
A. Local Area Network Considerations: e ese e owa s is i u e a a ocessi g cou e wi e i c ease em asis o o ice au oma io a e e osi e g ow a e i e use o au oma e o ice e mi a s is ays a com u e s as c ea e a ema o mo e e i i i y i e mi a i s a a io a i e co ec io sys ems[1] e so a wo ks a io s a e eme gi g as a mea s o i c easi g o e a o uc i i y y a owi g e i i i ua o ice wo ke s o c e ica a o essio a o s a e access o os sys ems commo a a ases e i e a i e s a emo e com u e e wo ks e wo ks a io i se may a ge a yw e e om a ow cos key oa / is ay e ice o a sma com u e [1] e i us y es o se o is ee as esu e i se e a o osa s a o e i gs o commu ica io sys ems s eci ica y esig e o oca i e a a a s e ese A s a e i e e o o i e a a s o mec a ism o igi a e co e a a a e as we as o o -co e a a( acsimi e oice a u - 3 mo io i eo A A ca o e a so u io a a esses ese ese a u u e ee s o i us y w e ase o a commu ica io s a c i ec u e a ysica wi i g s a egy a a ows o g ow a mig a io i o e 1s ce u y A A ca e e i e e as a i o ma io a s o sys em o a a a s e amo g o ice sys em e mi a s c us e e co o e s o os sys ems ia a i e co ec i g me ium wi i e ou s o a si g e o ice ui i g ui i g com e o cam us e geog a ica co s ai s e imi a e e ee o use commo -ca ie aci i ies us i c easi g e a a a s e ca aci y o e A y a owi g eco omica a a a smissio a es o ma y mi io i s e seco o ay s e wo k a smissio ec o ogy e mi s e a s e o a ge ocks o a a a ese a es wi sim e e o ma ageme oce u es a co o o oco s ese a smissio a es a so e mi a a ge um e o a a e mi a s o s a e e commo ysica i e co ec io i k wi mi imum e o ma ce eg a a io esu i g om esou ce co e io B. Standards for LANs e IEEE s A s a a is e key s a a o oca a ea e wo ks is o ec as a e a io s i o e ISO OSI (O e sys ems I e co ec io e e e ce mo e as s ow i ig 1 1 e a c i ec u a mo e is a se o i e aces e owe o io o e ISO OSI mo e as ee su s uc u e y e o ec [ ] as s ow i ig 1 I a icu a e u c io s e o me y e a a i k aye a e ee is i u e o e wo su aye s a " ogica i k Co o ( C " a u e su aye a "Me ium Access Co o (MAC " e owe su aye e C aye se es o ge e a e a i e e e i k co o comma s w i e e com eme a y owe aye e
ig 1 1 ISO OSI Mo e 5
Fig 1.2 IEEE 802 Project Sublayer Model 6
MAC se es o ame e a a u i s a acqui e e ig o access e me ium A o e e o om wo aye s e i e aye s s eci ie y e ISO mo e s ou o e a e i e e e y Wi i is s uc u e e o ec as e i e ou s a a s; a C s a a a ee s a a s o me ium access co o oge e wi ei associa e ysica aye s o use wi e C s a a ee suc s a a s a e (1 Ca ie se se mu i e access wi co isio e ec io (CSMA/C ( oke - assi g us access (3 oke i g access e " oke - assi g us" access me o e su ec o is esis is o e o e me ium access co o ec iques w ic gi es e ig o e s a io s co ec e o e me ium o a smi ike CSMA/C e oke assi g us access me o commu ica es y oa cas i g e sig a s ge e a e a eac s a io o a s a io s a ac e o e me ium e me ium se es as a commo us oug w ic e a ac e s a io s ca a smi A oug CSMA/C is wi e y use i o ices u i g e e e o me o s a a eo e om e Ge e a Mo o s a o e com a ies i e es e i ac o y au oma io a se ious ese a io s a ou i ue o e o a i is ic e a io o e MAC o oco e e is a s ig c a ce a a s a io mig a e o wai a i a i y o g o se a ame (i e e wo s case is u ou e I e oke us me o e e a e s a io s a eac s a io akes seco s o se a ame So o ame wi e e a e o wai mo e a seco s o ge a c a ce o e a smi e [5] e ac o y au oma io eo e i 7
ike e co ce ua i ea o a i g u i o ike i s ysica im eme a io ecause a eak i e i g ca e wi i g e w o e e wo k ow u e ey o e a a i g is a oo i o e i ea o o ogy o mos assem y i es As a esu a ew s a a was e e o e a i g e o us ess o e ca e use y e CSMA/C me o 3 u e k ow wo s case e a io o a i g . . Discussion of the Physical layer o e ysica aye e oke us uses a oa a coa ia ca e use o ca e e e isio ee i e e a a og mo u a io sc emes a e e mi e i e oke us me o (1 ase co i uous eque cy S i Keyi g ( ase co e e eque cy S i Keyi g (3 Mu i e e uo i a y Am i u e mo u a e ase S i Keyi g a a a es o 5 a 1 M s a e ossi e usi g e ase co e e eque cy s i keyi g o e mu i e e uo i a y AM/ SK wo im o a as ec s o e e wo k si e a e o e a e g o e ca e a o a um e o o s I y ica si ua io s ase-co i uous- SK y e o a a og mo u a io is use I suc ac ica e wo ks e ca e e g is usua y imi e y ca e a e ua io e ma imum e wo k e g s ac ie e wi comme cia y-a ai a e ca es a y om 1 me e s o 7 me e s [ ] A oug e ysica aye s eci ica io i c u es e use o e ea e s o e e e u k ca e eyo e e mi e e g is as a e y ee ee e i ac ice a suc use may com omise e sim ici y o e wise o e e y e ase-co i uous- SK y e o se ice e um e o o s i e a o e
e wo ks as y ica y a ge om o a o ima e y 3 is s eci ica io oes o co s ai e ici y e ma imum um e o o s o is a um e e ici y e i e ca e a ame e s e e e e e wo k o o ogy is a o g u a c e u k equi i g a si g e u k ca e o e ou e o e e y s a io si e i u is s eci ica io oes co e e use o ac i e ege e a i e e ea e s o a c i g e a a is a s e e e wee e s a io s i e o m o acke s Eac acke is a co ec io o a i i e um e o y es ( o e 51 1 e c e e icie cy o e e wo k is a ge y e e e o e si e o e acke s ei g se a is ig w e a ge acke s a e use is is ue o e a ge u i i a io o e us y e a a acke s a a ge acke e g s A IEEE S a a s Commi ee as u is e i s A SI/IEEE s oke - assi g us Access Me o i 19 5 I as see wi es ea im eme a io oug ou i us y C. Performance of a network e e o ma ce o a A measu e y e oug u ac ie e a e e ay su e e y e acke s is a u c io o se e a a ame e s o e o oco a e e wo k ese a ame e s i c u e e um e o s a io s e wo k o o ogy e wo k e g acke e g a e is i u io o e s a io s [1 ] e e o ma ce o a e owo k is e y muc i ue ce y e e g o e e wo k w ic is e is a ce e wee e wo a es s a io s I e ese oke us mo e as e s a io s o m a ogica i g e e o ma ce o e mo e is o se e o a ious ysica a ogica is i u io o e s a io s o e us 9
Simu a io s u ies a a e ee o e o e oke us e o ma ce so a we e co ce a e o e e ay- oug u e a io s i a e e ec o e o e e oa o e e o ma ce o e mo e [3] [11] Some s u ies we e co ce a e o e ack ow e geme a io i y mec a isms i oke us e wo ks [9] Some o e s u ies we e ase o e queuei g a a ysis o e a a acke s i e u e s o e s a io s A e a o e me io e s u ies a e assume o e a icu a case o e ysica a ogica is i u io o e s a io s o e us o e o as ee ma e so a o es e e ec o a ia io o e ysica a ogica is i u io o e s a io s o e e o ma ce o e mo e u e i mos o e a o e me io e simu a io s u ies e e y s a io is mo e e y a ai o ocesses e i s ei g a e o se a ecei e a a acke s usi g e us a e seco ei g a e o ac as a a ic ge e a o [1 ] u e a s a io queues e a a acke s o a smissio i e ou u u e u i i ecei es e oke e oke us mo e e e o e i is esis co si e s e a ic ge e a i g co i io s w ic a e i e e om ose assume i e a o e me io e s u ies e a ic is o e e o e mo e y a a ic ge e a o i e o m o a s eam o a a acke s Eac a a acke is assig e a sou ce a a es i a io s a io a a om [1 ] e sou ce ca o e same as e es i a io s a io E e y s a io as a u e wi a ca aci y o s o i g e ac y o e a a acke I a e s a io s a e ou usy i e ei u e s a e u e a ic ge e a o wai s o a a om ime i o e o e s a io s is ee o acce a a a acke I is ocess o wai i g e queui g e ay o e a a acke s a e a ic ge e a o is o co si e e 1
I e ese mo e i a i io o e e ay- oug u a a ysis i e e ysica a e ogica is i u io o e s a io s a e ee co si e e e mo e assumes wo i e e y es o ysica is i u io ame y (1 U i o m ( C us e e o m I e u i o m is i u io e s a io s a e s ace u i o m y a o g e e wo k e g w e eas i e c us e e o m e s a io s a e i i e i o a um e o equa o u equa si e c us e s wo cases o ogica is i u io a e co si e e i is mo e e i s ei g e es case i w ic e ysica a ogica eig o s coi ci e eac o e a e wo s case i w ic e ogica eig o s a e se a a e y a a ge is a ce e e ec a ia io o e ysica a e ogica is i u io o e s a io s o e e o ma ce o e mo e is o se e i a e e e ime s e e ai s o w ic a e e ai e a e i C a e I e e ai s o e ysica a ogica is i u io o e s a io s a e gi e i C a e III CSIM a ocess o ie e simu a io a guage w ic is im eme e as a su e se o e C og ammi g a guage is use as a simu a io oo CSIM as a um e o ea u es w ic ease e ask o ui i g simu a io mo e s ey i c u e mo e i g sys ems esou ces message assi g a a co ec io a e uggi g [ 1
CSIM is copyrighted by Microelectronics and Computer Technology Corporation, 1985 11
CHAPTER II PROTOCOL AND MODEL DESCRIPTION
is c a e ese s e esc i io o e mo e o w ic e o oco is es e a a o e iew o esc i io o e o oco a is es e u e is c a e a so e i es e a ame e s use i e simu a io o e oke - assi g us o oco
A. Protocol Description. e IEEE s a a s i c u e wo access co o me o s ase o e oke assi g ec ique O e o ese me o s is use o a e wo k wi us o o ogy a e o e o a e wo k wi i g o o ogy e oke assi g s a a em oyi g e us o o ogy k ow as oke us is e i e i IEEE s a a is is e asis o e Ge e a Mo o s MA a c i ec u e a is use e e si e y i ac o y au oma io [ ] Wi e oke us a oac co o o e a smissio me ium is e e mi e y ossessio o a oke w ic is asse om s a io o a s a io i a ogica i g
A.1. Overview of Token bus method: e ig o access e ysica me ium is gi e y e oke e oke is a co o ame a gi es mome a y co o o e e me ium e oke is asse om o e s a io o a o e e e y o mi g a ogica i g o ma s ea y s a e o e a io co sis s o wo ases [ ]; ame y 1
(1 a a a s e ase ( oke a s e ase I e i s ase e s a io w ic is e oke o e a s e s a a o e us A e a s e i g e a a e oke o e a s e s e oke o i s successo i i s oke a s e ase e e y e i quis i g i s co o o e e us a e ocess e ea s e mai e a ce u c io s o e ogica i g so o me wi i e s a io s o i e o e o owi g (1 i g i i ia i a io ( os oke eco e y (3 ew s a io s a i io o e i g ( Ge e a ousekee i g o e ogica i g e ysica o e i g o e s a io s o e us is i e e e o e ogica o e i g [ ] a oug i may e e a e o i as s ow i e e am e o ig 1 e oke me ium access me o is a ways seque ia i a ogica se se a is u i g o ma s ea y s a e o e a io e ig o access e me ium asses om s a io o s a io e me ium access co o (MAC su aye o i es seque ia access o e s a e us me ium i a ogica y ci cu a as io is MAC su aye e e mi es w e a s a io as e ig o access e s a e me ium y ecog i i g a acce i g e oke om i s e ecesso s a io
A MAC I E A S UC U E e MAC aye e o ms se e a u c io s w ic a e oose y cou e 13
Fig 2.1 Logical Ring on Physical Bus 1
e esc i io s a s eci ica io s o e MAC aye i is s a a a e o ga i e i e ms o o e o e se e a ossi e a i io i g o ese u c io s e a i io i g o e MAC aye is i us a e i e ig w ic s ows i e asy c o ous ogica "mac i es" eac o w ic a es some o e MAC u c io s ese a i io s a e as o ows (1 I e ace Mac i e(I M ( Access Co o Mac i e(ACM (3 ecei e Mac i e( CM ( a smi Mac i e( M (5 ege e a i e e ea e Mac i e( M e u ose o eac mac i e is e ai e e ow A.2.1. Interface Machine: is mac i e ac s as a u e a i e ace e wee e C a MAC aye s I i e e s a i comi g a a a o e se ice imi i es a ge e a es a o ia e ou goi g se ice imi i es u e i a es queui g o se ice eques s I a so e o ms e a ess ecog i io u c io o ecei e C ames acce i g o y ose a esse o i s s a io A.2.2. Access Control Machine: is mac i e coo e a es wi e ACMs o i s e ow s a io s i a i g e oke o co o e a smissio access o e s a e us e ACM is es o si e o e o owi g i I i ia i a io a mai e a ce o e i g ii aki g ca e o ew y a mi e s a io s iii e ec io a eco e y om au s a ai u es i e e wo k A.2.3. Receive Machine: is mac i e acce s i u s om e ysica aye assem es em i o ames w ic i a i a es a asses em o e I M a 15
Fig 2.2 MAC Layer Functional Partitioning 1
ACM e M accom is es is y ecog i i g e ame s a a e e imi e s c ecki g e ame c eck seque ce a a i a i g e ame s s uc u e M is a so es o si e o ece io o oise u s a quie co i io s is oise u s co i io is e a e w e e s a io is is e i g o e us o some es o se A.2.4. Transmit Machine: is mac i e ge e a y acce s a a a ame om e ACM a a smi s i as a seque ce o a omic sym o s i a o e o ma o e ysica aye e M ui s a MAC o oco - a a-u i y e aci g eac ame wi e equi e eam e a s a i g e imi e a a e i g e ame co o seque ce a e e imi e A.2.5. Regenerative Repeater Machine: is mac i e is a o io a MAC com o e ese o y i s ecia " e ea e s a io s" e g i a oa a o a ea -e emo u a o I suc s a io s e M e ea s e i comi g a omic sym o s eam om e ysica aye ack o e ysica aye o e a smissio O a ese i e mac i es e ACM is o e mos c i ica a e mos com e I is e key co o mec a ism o e oke - us me o e I M a M a ici a e ea i y i e o e a io o e MAC aye o oco
A.3. MAC Definitions: C i ica MAC a ame e s w ic a e co s ai e y is s eci ica io a e e i e as o ows A.3.1. MAC-symbols: e sma es u i o i o ma io e c a ge e wee wo MAC su aye e i ies [ ] e si MAC- sym o s a e e i e i a e 1 17
AME A E IA IO
e o o e 1 o _ a a a _i e si e ce S a _sig a
a e 1 MAC sym o s Co e io a i a y a 1 a a i s a e se a ecei e as e o a o e MAC-sym o s es ec i e y A 3 MAC_sym o _ ime e ime equi e o se a si g e MAC_sym o i e ica o a sym o s is is e i e se o e A a a a e as s ow i a e
omi a omi a a a a e MAC_sym o _ ime
1 M /s 1 mic o seco 5 M /s mic o seco s 1 M /s 1 mic o seco s
a e MAC sym o imes 18
A.3.3. Octet time: Co es o s o e ime i e a equi e o a smi eig MAC-sym o s
A.4. Elements of MAC Sublayer Operation: e MAC o oco as e e o e is i e e o e o us i se se a i s ou o e a e a su i e mu i e co cu e e o s [3] ese e o s may e cause y commu ica io e o s o s a io ai u es ese e o s may i c u e mu i e oke s os oke oke - ass ai u e ea s a io (s a io wi i o e a i e ecei e a u ica e s a io a ess e ese mo e u e s u y is assume o e ee o e o s A.4.1. Basic Operation: S ea y s a e o e a io ( e e wo k co i io w e e a ogica i g as ee es a is e a o e o co i io s a e ese sim y equi es e se i g o e oke o a s eci ic successo s a io as eac s a io is i is e a smi i g A e s a io s a e co ec e o a commo me ium w ic is a us ma e o a oa a coa ia ca e ysica y e oke us is a i ea o ee-s a e ca e o o w ic e s a io s a e a ac e ogica y s a io s a e o ga i e i o a i g wi eac s a io k owi g e a ess o e s a io o i s " e " ( S a " ig " ( S a a om k owi g i s ow a ess ( S W e e ogica i g is i i ia i e e ig es um e e s a io o s e oke a may se e i s a a ame A e i is o e wi a smi i g i asses e e missio o i s imme ia e eig o y se i g e eig o a s ecia co o ame ca e a oke e oke o aga es a ou e ogica i g wi o y e oke o e ei g e mi e o a smi ames Si ce o y o e s a io is a owe o a smi 19
a a ames a a ime co isio s wi o occu A im o a oi o ea i e is a si ce e ca e is i e e y a oa cas me ium eac s a io ecei es e e y ame isca i g ose o a esse o i W e a s a io asses e oke i se s a oke ame s eci ica y a esse o i s ogica eig o i e i g i es ec i e o w e e e s a io is ysica y oca e o e ca e I is a so im o a o o e a w e s a io s a e i s owe e o ey wi o e i e i g so e MAC o oco as o isio s o a i g o a e e i g s a io s om e i g
A.5. Frame Format of IEEE 802.4 Std: e oke us ame o ma is s ow i e ig 3 e a ious com o e s o e ame a e esc i e e A.5.1. Preamble: e eam e a e ece es e e y a smi e ame eam e is se y a MAC aye as a a o ia e um e o a _i e sym o s is wi e eco e y e ecei e mo em as a i a y a a sym o s a occu ou si e e ame e imi e s eam e is ima i y use y e ecei i g mo em o acqui e sig a e e a ase ock y usi g a k ow a e ag ee u o y e a e s a io s A seco a y u ose o e eam e is o gua a ee a mi imum e e imi e (E o s a e imi e (S ime e io o a ow s a io s o ocess e ame e ious y ecei e e mi imum amou o eam e a smi e is e e e o e a a a e o e me ium a e mo u a io sc eme im eme e e s a a equi es a e u a io o e eam e s a e a eas mic o seco s ega ess o e a a a e [ ] Fig 2.3 Frame format of token bus 1
A.5.2. Start Delimiter(SD): e ame s uc u e equi es a s a e imi e w ic egi s e ame e s a e imi e co sis s o sig a i g a e s a a e a ways is i guis a e om a a A.5.3. Frame Control Field: e ame oc e ( C e e mi es w a c ass o ame is ei g se is ie is o o e oc e a ious y es o c asses i c u e MAC co o C a a S a io ma ageme a a a S ecia u ose a a( ese e o u u e use A.5.4. Destination Address Field(DA): e es i a io a ess i e i ies e s a io o w ic e ame is es i e is may e o oc e s e e i g o e um e o i s use o a essi g A esses s a e ei e 1 i s o i s i e g A a esses o a gi e A s a e o e same e g A.5.5. Source Address Field(SA): e sou ce a ess i e i ies e s a io o igi a i g e ame a as e same o ma a e g as e es i a io a ess i a gi e ame A.5.6. MAC Data Unit Field: e e i g o e i a e s eci ie i e ame s ame co o oc e e MAC a a u i ie ca co ai a C o oco a a u i w ic is use o e c a ge C i o ma io e wee C e i ies o a MAC ma ageme a a ame w ic is use o e c a ge ma ageme i o ma io e wee MAC ma ageme e i ies o a a ue s eci ic o o e o e MAC co o ames A.5.7. Frame Check Sequence (FCS) field: e CS is a 3 i ame c ecki g seque ce A.5.8. End Delimiter(ED): e ame s uc u e equi es a e e imi e w ic e s e ame a e e mi es e osi io o e ame c eck seque ce e
a a e wee e S a e E s a e a i eg a um e o oc e s A i s e wee e s a a e e imi e s a e co e e y e ame c eck seque ce A.5.9. Token frame: e oke ame is a s ecia MAC co o ame w ic gi es a s a io e ig o a smi is ame as a s uc u e s ow i e ig 3 is ame as e A = e a ess o e s a io s successo i e ogica i g e oke ame as a u a a_u i
A.6. The Token Bus MAC Sublayer Protocol. W e e i g is i i ia i e s a io s a e i se e i o i i e esce i g o e o ei a esses oke assi g is a so o e om ig o ow a esses Eac ime a s a io ossesses e oke i ca a smi ames o a ce ai amou o ime ca e e oke o ime e i mus ass e oke o I e ames a e s o e oug se e a co secu i e ames may e se I a s a io as o a a i asses e oke imme ia e y u o ecei i g i e oke us e i es ou io i y c asses a o a ic wi e owes a e ig es Eac s a io is i e a y ei g i i e i o ou su s aio s o e a eac io i y e e As i u comes i o e MAC su aye om a o e e a a a e c ecke o io i y a ou e o o e o e ou su s a io s us eac su s a io mai ai s i s ime wi s i u a e ime k ow as a ge o a io cou e ( C W e a s a io ecei es e oke i egi s a smi i g e ames o io i y c ass W e i is o e e oke is asse i e a y o e io i y su s a io w ic may e a smi u i i s ime e i es a w ic oi e oke is asse i e a y o io i y su s a io is ocess is e ea e u i a i s ames a e ee se o i s ime as e i e 3
Agai w e e s a io oes o a e a y ames o se i sim y asses e oke o i s successo e io i y sc eme is eyo e sco e o e ese mo e e mo e assumes a si g e io i y o a e a a acke s
A 7 Logical Ring Maintenance. om ime o ime s a io s a e owe e u a wa o oi e i g O e s a io s s a io s u e o a ea e e i g e MAC su aye o oco o i es a e ai e s eci ica io o e ac y ow is is o e w i e mai ai i g e k ow wo s case ou o oke o a io e ow a ie ske c o is mec a ism is gi e O ce e i g as ee es a is e eac s a io s i e ace mai ai s e a esses o e e ecesso a successo s a io s i e a y e io ica y e oke o e so ici s i s om s a io s o cu e y i e i g a wis o oi y se i g o e o e solicit successor ames e so ici a io ame gi es e se e s a ess a i s successo s a ess S a io s i si e a a ge may i o e e ( o kee e i g so e i esce i g o e o ei a esses A e e s a io se s e solicit successor ame i o e s a es o se wi ow w ic is a co o e i e a o ime (equa o o e s o ime I o s a io i s o e e wi i a s o ime( as i E e e e es o se wi ow is c ose a e oke o e co i ues wi i s o ma usi ess [4]. I e ac y o e s a io i s o e e i is i se e i o e i g a ecomes e oke o e s successo I mo e s a io s i o e e a e same ime e ames wi co i e a e ga e as i E e e e oke o e wi e u a a i a io a go i m s a i g wi e oa cas o a resolve contention ame 24
The solicitation of new stations will not interfere with the guaranteed worst case for the token rotation. Each station has a timer that is reset whenever it acquires the token. When the station receives the token the old value of this timer is inspected just before the timer is reset. If it exceeds a threshold value, there has been too much traffic recently, so not bids may be solicited this time around. No guarantee is provided for how long a station may have to wait to join the ring when traffic is heavy, but in practice it should not be more than a few seconds. Leaving the ring is easy. Consider a station A, with successor B, and predecessor C. To leave the ring, A sends C a set_successor frame telling it that henceforth its successor is B instead of A. Then A stops transmitting.
Due to transmission errors or hardware failures, problems can arise with the logical ring or the token. For example, if a station tries to pass the token to a station that has gone down, what happens? The solution is straight forward. After passing the token, a station listens to see if the successor either transmits a valid frame passes the token. If it does neither, the token is passed a second time.
Even if that fails the station transmits a who_followsframe specifying the address of its successor. When the failed station's successor sees a who_follows frame,it responds by sending a set_successor frame to the the station whose successor failed, naming itself as the new successor. In this way the failed station is removed from the ring. Thus the logical ring is maintained. The model assumes a stable steady-state as far as the membership of the logical ring is concerned. Thus the performance figures computed are the upper bounds. 5
B. Model Description. We a e i e es e i e a ua i g e e o ma ce o oke - us e wo k w ic im eme s e IEEE s a a a o e s a u ack ow e ge a ag am se ice o e a a- i k use s is mea s e a ag am se ice o e e e e is o es o si e o ma agi g a y ack ow e geme mec a isms e wee emo e s a io s e o owi g ge e a assum io s a e ee ma e o e mo e e e wo k a c i ec u e i c u es a um e o s a io s co ec e o a s a e commu ica io c a e ase o e IEEE s eci ica io s Eac s a io is mo e e y a ai o ocesses e i s is o se a ecei e e acke s a e seco o ass e oke o i s successo a e i is o e wi e a smissio o acke s A i io a y eac s a io ca acce a a a acke om a se e w ic ge e a es e simu a e e wo k oa e oke is asse y e s a io s esi i g o e me ium acco i g o a ogica i g; e oke gi es co o o e a smissio c a e Eac s a io as a u e o s o e e acke s o e ese mo e e ca aci y o e u e is imi e o o e acke e ogica i g is assume o e s a ic i e o wi ows a e o e e o a ow o assi e s a io s o e i c u e i e ogica i g o s a a y s a io s a ea e e ogica i g [11] A s a io is ac i e i i as a acke o a smi i i s u e a is i ac i e i as o acke s e s a io o i g e oke a smi s e ac y o e acke i i is ac i e a imme ia e y asses e oke i i is o
CHAPTER HI SIMULATION DESIGN AND EXPERIMENTS
CSIM a ocess o ie e simu a io a guage (im eme e o o o e C og ammi g a guage is use o im eme a io o e mo e a e simu a io I is a oac ocesses a e e i e w ic a e i e e e og ams o oce u es w ic ca e ecu e i a a e ( seu o wi o e ocesses [3] e ocesses a ea o e ecu e i a a e I e CSIM mo e e commu ica io s me ium is ec a e as facility , a a smissio s a e ec a e as events. Simu a e ime is asse y usi g a hold u c io e esc i io o a ese u c io s is gi e i A e i A
A. Description of the software model e so wa e mo e com ises s a io s cou e oge e i a ogica i g I a ogica i g e s a io s a e assig e ogica osi io s i a o e e seque ce wi e as mem e o e seque ce o owe y e i s e ysica o e i g o e s a io s o e us is i e e e o e ogica o e i g I o e wo s e ogica successo o a s a io may o e i s imme ia e ysica eig o Eac s a io as a u e o s o e a si g e ou goi g a a acke o i i e e g ( i s Eac s a io k ows e a ess o i s e ecesso ( S a i s successo ( S a a om k owi g i s ow a ess ( S I eac e e ime e um e o s a io s e e wo k e g e s a io is i u io a e acke e g a e use as a ame e s Eac e e ime co sis s o u i g e mo e o 1 acke s w ic 27 are given to the overall network as load. These packets are offered to the model by a traffic generator which generates a stream of packets. For every packet, a source station and destination station are selected randomly [12]. Care was taken in the random selection of the source station, that it is neither in the transmission process nor its buffer is full. Obviously, the source station cannot be the destination. If the buffers of all the stations are found full, then the traffic generator waits for a random time until one of the stations is found ready to receive a packet. During this process the delay experienced by a data packet at the traffic generator before being assigned to any station is not considered. Further the time taken by traffic generator to assign a data packet to a station is assumed to be zero. Any station can transmit a data packet if it has the token and pass the token to its successor station after transmitting the packet. A station holds the token for a certain time given by the "token hold time" if it has a data packet to transmit. This token hold time is the maximum time that a station can hold the token to transmit the data packet. In the present model the token hold time is set to a time large enough for a station to transmit one packet of variable length. The token hold time is changed according to the packet length and is same for all the stations in the model. The length of the token is same for all the experiments. Until a station receives the token, a data packet waits in the buffer. Thereby the token enables the station to transmit the packet to its destination. A station passes the token to its successor immediately if it has no packet to transmit.
A.1.Time-domain discussion This section gives the time domain features from
e iew oi o e a smi i g s a io e a ious imi gs i o e u i g e a a a oke a smissio s a e s ow i e ig 3 A e imi gs e cou e e u i g e simu a io a e e i e i e ig 3 1 e ge e a ogic im eme e a a y s a io is e ai e c ea y i e ow c a gi e i e ig 3 e a o e ow c a a o e e ci c e ma ke i e ig 3 is o simu a e e owe a o e ow c a w ic is simu a e e ai s e s ea y s a e o e a io o e mo e o s a wi e ig es a esse s a io w ic is e mas e s a io i e ogica i g ge e a es e oke a s a s a smi i g e a a acke i i s u e as o e O e wise i us asses e oke o i s successo a so o is ese mo e oes o ea wi e ece io si e o commu ica io a a i g o a smissio e o s ese issues a e ea e a e ig e e e s o e o oco e o e e oa o e e wo k is co o e y e ime ga e wee acke a smissio s o e a ic ge e a o A e e e y a smissio i i ia io e ime ga o e e a smissio is se ec e a om y wi i a a ge gi e y wo a ame e s 1 a [1 ] 1 a a e wo i ege a ues y a yi g e a ues o 1 a e i e a i a a e o e acke s a e s a io s is c a ge usi g a a om u c io w ic ge e a es a i ege e wee 1 a e a ues o 1 a co o e o e e oa i eac e e ime e e ai s o ca cu a io o e o e e oa is gi e a e i is c a e o eac e e ime e acke e g a e oke e g a e ke co s a IEEE S s a a s a e use o e mo e As e e s a a s c a e ca aci ies o 1 5 a 1 M s(mega i s e seco [ ] a e ossi e o e ese mo e e c a e ca aci y is assume o e 1 M s Fig 3.1 Time domain picture from the view point of the transmitting station, 3
ig 3 ow c a o e ogic a e s a io s 31
B. Physical Distribution of the stations e s a io is i u io is a ie o ce ai e e ime s i e o owi g co igu a io s B.1.Uniform: S a io s a e s ace u i o m y a o g e e g o e e wo k( wi s aci g e wee e e y ai o a ace s a io s ei g / -1 as s ow i e ig 3 3
ig 3 3 S a io is i u io U i o m s aci g
B.2.Equal-sized Clusters: e s a io s a e i i e i o c c us e s wi eac c us e a i g / c um e o s a io s e ce e s o e c us e s a e u i o m y s ace a o g e e wo k e s a io s i eac c us e a e s ace u i o m y ( ig 3
ig 3 S a io is i u io Equa si e c us e s
B.3.Unequal-sized clusters: e s a io s a e i i e i o c c us e s wi eac c us e a i g a i e e um e o s a io s ( ig 3 5 32
Fig 3.5
Station distribution. Unequal-sized clusters
C. Logical Distribution of the stations In the token bus accessing method the stations form a logical ring as already discussed. The logical neighbor of any station may not be its physical neighbor. In the present model we co si e two different logical distributions. In the first distribution, which is the best case, the logical neighbors are same as the physical neighbors. In the other distribution, the worst case, the logical neighbours are physically apart by a maximum possible distance.
D. Statistical Measurements The statistical measurements, collected during the simulation runs are offered load, throughput, and average packet delay. Further the maximum delay of a data packet is also recorded. D.1. Offered Load (G): Offered load is the total number of bits offered to the network. It includes the bits due to the control packets, tokens. The offered load
(normalized) is calculated by dividing the total packet bit rate, in bits per second 33
( o a um e o acke s o e e o e e wo k i c u i g e oke s y e c a e ca aci y 1 M s e o mu a use o com u a io o e o e e oa i e ese mo e is gi e as
G = ( ( O_O - ACKE S ACKE _ E G + O A _ OKE S OKE _ E G sim_ ime ca aci y (1
w e e O_O _ ACKE S is e o a um e o a a acke s o e e o e e wo k ACKE _ E G is e e g o e a a acke i i s (51 1 e c O A _ OKE S is e o a um e o oke a smissio s y e s a io s u i g e sim_ ime OKE _ E G is e e g o e oke ame w ic is assume o e 11 i s sim_ ime is e o a ime simu a e y e mo e ca aci y is e c a e ca aci y w ic is assume o e1 M s I e oke us access me o w e a s a io as o acke o a smi i as o ass e oke o i s eig o E e i e e a e o acke s o e e o e e wo k e s a io s w ic a e i e a e o ass e oke amo g emse es Si ce e o e e oa i o es e oke s e e a ways e is s a o - e o o e e oa e e i e a se ce o a a acke s D.2. Input Load (G'): We e i e a o e oa ca e as "i u oa " (G e i e as e um e 3
o a a i s o e e o e e wo k e seco o ma i e o sys em ca aci y e i u oa is gi e y e o mu a G = ( O_O _ ACKE S ACKE _ E G /sim_ ime ca aci y ( e i u oa oes o ake e oke a smissio s i o co si e a io D.3. Throughput (S): e oug u o e mo e is ca cu a e y e o mu a
W e e i is e ime ake o a smi a o aga e a a a acke om a s a io i o a s a io ooki g ack o e ig 3 i i c u es e o aga io e ay i w ic is e ime ake o o aga e e a a acke om e s a io i o e s a io sim_ ime is e o a simu a e ime a ca aci y is e ca aci y o e e wo k (1 M s D.4. Average delay (D): e a e age e ay is e a e age acke e ay ca cu a e y e o mu a
w e e is e o a um e o acke s o e e o e mo e i is e ime ake o a a a acke o a smi success u y om e s a io i o e s a io
us e ime s e y e acke i e u e ( u om ig 3 u i e s a io ecei es e oke e a i g a smissio us i e i es e o a e ay o o e acke 35
D.5. Maximum Delay (D max ) This is ma imum e ay e e ie ce y a a a acke O e 1 a a acke s o e e o e e wo k as e oa o e a a acke wi e e ie ce e ma imum e ay a o e a a acke s max wi e
e ma imum o a ig s
us e simu a io a ame e s a e ee e ai e a e esu s o ese e e ime s a e gi e i e o owi g c a e 36
CHAPTER IV
SIMULATION RESULTS
A. Overview of the experiments:
The performance of the LAN is modeled with various parameters like packet length, station distribution, logical distribution of the stations, number of stations and offered load which is controlled by the values of the packet arrival rates R1 and R2. Each experiment consists of running the model for 1000 packets. In each experiment the load offered to the network is varied by varying the arrival rate of the packets at the stations by varying the values of R1 and R2. In some experiments the packet length is varied for all other constant parameters. In some experiments the physical station distribution is varied from uniform to clusters for all other constant parameters. In some other experiments the number of stations is varied from 5 to 50 for all other constant parameters. Finally in some experiments the performance of the model is tested for different network lengths with all other parameters constant. In most of the experiments the network length is assumed to be of 1000 meters with 20 stations distributed uniformly. In all the experiments the token hold time is same for all the stations, time enough for a station to transmit one data packet and pass the token to its successor. B. Discussion of the results The results of various experiments are discussed in this section. B.1. Throughput Versus Offered load: Figure Fig 4.1 shows the variation of the throughput with the offered load for a packet length of 512 bytes. It is observed that at lower loads the throughput increases nearly linearly with the offered load. As the offered load becomes higher 000 t r 20 t t n 2 b t
4. 38 the throughput is stabilized at a maximum value. It can also be observed that the offered load will not be reduced to zero. This is due to the token passing between the stations even when none of the stations have data packets to transmit. This prevents the offered load from reducing to zero in the absence of data packet transmission. The throughput of an ideal network (a network that has no overhead or token passing) increases to accommodate the load up to an offered load equal to the full capacity of the system; then remains at 1.0, which means that the network is fully utilized [3]. In the present experiment, the overhead due to the token passing always causes the performance to fall short of this ideal. Fig 4.2 shows the offered load-throughput relation for a shorter packet length. The slope of this curve is less than that of the curve with a higher packet length. This is because there is relatively more token overhead due to the token passing at smaller packet lengths. The difference of the slopes of both these curves can be better seen from Fig 4.3, where they are plotted together. It can also be observed that the maximum throughput value is higher for longer packets. This is because,the channel capacity being the same, the proportion of the number of useful data bits carried by the medium at any instance decreases as the packet length decreases. B.2. Effect of the total length of the network on the throughput The total length of the network is the distance between the farthest stations . The following Table 4.1 gives the values of the throughput achieved for various lengths with a packet length of 4096 bits ,token length of 112 bits for 20 stations. 28 b t 000 t r 20 t t n
4.2 1 me e s; s a io s;u i o m y is i u e
4. 1
1 oug u o a e g (me e s
1 1 9 1 1 1 97 5 1 1 971 1 1 1 9 1 1 957 1 1 9 1 1 919 1 1 3 9 1 1 3 9 3 5 1 3 959 1 1 3 95 1 3 937 1 3 9 9 1 3 95 1 3 9 3 1 3 959 5 3 955 1 3 9 3 933 3 9 3 9 1
a e 1 oug u o i e e e wo k e g s 42
From Table 4.1 it can be observed that for a given packet length (4096 bits) as the length of the network is increased, the throughput of the network is decreased.
As the length of the network is increased, the distance travelled by the token is increased. This increases the token delay, which is part of the time that a packet has to wait before it is transmitted by the source station to its destination.
The increase in the token delay has the effect of increasing the total simulated time, which brings down the throughput. B.3. Throughput Versus Input load:
Fig 4.4 shows the variation of the throughput with input load defined in
the last chapter. It can be seen from the figure that the throughput increases
linearly with the input load and stabilizes to a maximum throughput at higher
loads. A striking difference between Fig 4.3 and 4.4 is that the the curve in the
Fig 4.3 passes through the zero origin where as the curve in Fig 4.4 does not pass
through the origin. This implies that the input load is zero when no data packets
are offered to the network, where as the offered load will never be zero, owing to
the token passing between the idle stations when they have no data packets to
transmit. Further the offered load to the network is greater than the input load.
This is due to the fact that the offered load includes not only the data packets
offered to the network but also the token passing between the stations ,whereas
the input load includes only the data packets.
B.4. Average Delay Versus Throughput
Fig 4.5 and Fig 4.6 show the variation of the average packet delay with
the throughput for two different packet lengths. As the arrival rate of the packets
at the stations increases, the offered load increases. As a result of the increased 2 b t 20 t t n 000 t r
4.4 2 b t 20 t t n 000 t r
4.f 28 b t 20 t t n 000 t r
46 offered load, the throughput of the model increases. This has an effect of increasing the average token rotation period. Average token rotation period is the average time taken by the token to complete one round of passing through all the stations in the logical ring. A station has to wait for the token before it transmits a packet. So, as the average token rotation period increases the average packet delay increases. It can be observed from the figures that the average packet delay increases gradually with the throughput and attains a maximum at the highest throughput. Delay-throughput analysis of two different packet lengths (in bytes) is shown in the figure Fig 4.7. The use of longer packets yields better throughput , but at the cost of increase in the time delay. The sensitivity of the model to varying packet lengths can be seen from the figure Fig 4.7. In this delay analysis the additional overhead to the data packet transmission caused by the token transmission is considered. The maximum delay experienced by a data packet is recorded for each experiment. A situation arises where the traffic generator supplies a data packet to a station that has just passed the token to its successor. In such a case the data packet has to wait in the buffer until the station receives the token in the next round. Thus the data packet experiences the maximum delay under the above mentioned conditions. The maximum delay experienced by a data packet is recorded as 8,945.55 micro seconds. B.5. Throughput Versus Number of stations Results of experiments with four different packet lengths (64, 128, 256 and 512 bytes) as a parameter, uniformly distributed in 1000 meters are shown in the Fig 4.8. It can be seen that as the packet length increases the maximum throughput increases. It can also be seen that as the number of stations increases for any packet length the performance of the LAN is enhanced. This is because as the number of stations is increased within the same length, the stations will be located close to one another . Thus, less time is spent in token passing and the 1000 meters,20 stations;uniformly distributed
4. 1 me e s u i o m y is i u e
ig 49 overhead due to the token passing is decreased. B.6. Throughput Versus Station Distribution
Results of experiments with equal sized clusters and unequal sized clusters, for the best case with different intercluster spacing are as shown in the Table 4.2, Table 4.3. The model has a 1000 meters length and a packet length of 512 bytes. All equal sized clusters are symmetrical about the physical center of the network. Table 4.2 shows the maximum throughput achieved when the 20 stations are in equal sized clusters with intracluster spacing being 10 meters. It can be seen that there is only a slight change in the throughput for a change in the number of clusters, all else being equal
Configuration Inter-cluster Maximum separation throughput
2 clusters 410.0 meters 0.976
4 clusters 280.0 meters 0.973
10 clusters 100.0 meters 0.974
Uniform 55.0 meters 0.973
Table 4.2 Equal sized clusters Table 4.3 shows the maximum throughput achieved when the 20 stations 50 are in unequal clusters with the intercluster space being 10 meters. Even in this case there is a negligible change in the throughput for the change in the number of clusters
Configuration Inter-cluster Maximum separation throughput
3 clusters (10,5,5) 415.0 meters 0.977 2 clusters (15,5) 820.0 meters 0.976 2 clusters (19,1) 820.0 meters 0.975
Table 4.3 Unequal sized clusters These results confirm that physical configuration of the workstaions has a minimal effect on the throughput. This is due to the fact that the total distances travelled by the token in all the above mentioned physical configurations is the
same. Therefore the token overhead is the same. C. Effect of logical distribution of stations on the performance of the model As explained in the previous chapter the model is tested for two different logical distributions of the stations. The first case of distribution being the best
case where the logical neighbours are same as the physical neighbours and the other case being the worst case where logical neighbours are physically apart by a maximum distance. In this experiment the stations are uniformly distributed on a length of 1000 meters. Results of this experiment are as shown in the Table 4.4. 51
1 oug u oug u o a e g ( es case (wo s case (me e s
1 1 9 977 1 1 1 97 971 5 1 1 971 9 9 1 1 1 9 9 5 1 1 957 95 1 1 9 9 5 1 1 919 99 1 1 3 9 9 5 1 1 3 9 9 1 5 1 3 9 959 1 1 3 95 95 1 3 937 93 1 3 9 9 9 1 3 75 1 3 9 1 959 1 3 957 95 5 3 9 9 9 5 1 3 939 935 3 9 7 9 5 3 9 9 3 9 79 1
a e Com a iso o oug u o e es a wo s cases (U i o m is i u io 5
I ca e o se e om e esu s a e e is o a a ecia e i e e ce i e oug u o e A e wee e wo cases We ca i e om is a e ogica is i u io o e s a io s oes o a e a e ec o e e o ma ce o e mo e w e e s a io s a e is i u e u i o m y is is ima i y ecause e oke o e ea is i e e e o e ogica is i u io o e s a io s e oke as o ass oug a e s a io s i es ec i e o ei oca io s o e me ium as suc e oke o e ea wi e e same O e o e a i e s a io s a e is i u e i e o m o c us e s as e ai e i e e ious c a e e oug u o e wo s case o e ogica is i u io is ess a a o e es case is is ecause w e e s a io s a e is i u e i e o m o c us e s e is a ce a e e y e oke e wee e s a io s i e wo s case is i c ease is i c eases e oke e ay w ic is a o e ime a a acke as o wai e o e i ge s a smi e y e sou ce s a io o i s es i a io is i c ease i e oke e ay as e e ec o i c easi g e o a simu a e ime w ic as e e ec o ec easi g e oug u A c ose ook a a e 5 s ows a e oug u i e case o e wo s case ec eases as e i e -c us e is a ce is i c ease is is ecause as e i e -c us e is a ce is i c ease e is a ce a e e y e oke i o e o a io wi e i c ease i c easi g e oke e ay e e y ec easi g e oug u 53
Configuration Inter-cluster Throughput Throughput
separation (best case) (worst case)
2 clusters 810.0 meters 0.976 0.912
4 clusters 280.0 meters 0.973 0.934 10 clusters 100.0 meters 0.974 0.945
Table 4.5 Comparison of throughput for best and worst cases
(non-uniform distribution)
An implication of the results from Table 4.4 and Table 4.5 is that when the stations are uniformly distributed, the logical distribution of the stations does not have an appreciable effect on the throughput of the network. This is because, the token overhead will be the same in both the cases as the token has to pass through all the stations irrespective of their locations on the medium. But, when the stations are distributed in the form of clusters, there will be a fall in the throughput. This is due to the fact that the distance travelled by the token will be large wdhen compared to the best case, which decreases the throughput. 5
CHAPTER V COMPARATIVE PERFORMANCE WITH CSMA/CD MODEL
e u ose o is c a e is o gi e some i sig i o e e a i e e o ma ce o e ese mo e em oyi g oke us access me o wi a o a o e mo e em oyi g CSMA/C access me o Ca ie Se se Mu i e Access wi Co isio e ec io (CSMA/C me ia access me o is a mea s y w ic wo o mo e s a io s s a e a commo us a smissio me ium I is a W ( is e w i e a k e wo k [5] I CSMA/C e e is o ce a s a io co o i g e a smissio ac i i ies o a e s a io s I is access me o w e a s a io as a a a acke o se i i s is e s o e me ium o see i a yo e e se is a smi i g I e me ium is ou usy e s a io wai s u i e me ium ecomes i e W e a s a io e ec s a i e me ium i a smi s a a a ame e e is a sma c a ce a us a e a s a io egi s se i g a o e s a io wi ecome ea y o se a se se e c a e I e i s s a io s sig a as o ye eac e e seco o e e a e wi se se a i e c a e a wi a so egi se i g esu i g i a co isio I a co isio occu s e s a io wai s a a om amou o ime a e a smi s e a a ame a is co i e o com a iso we co si e e wo mo e s wi i e ica oa ge e a i g co i io s o e mo e s a e s a io s i a e wo k e g o 1 me e s e c a e ca aci y i o e mo e s is 1 M s Comparison of throughput o o mo e s e oug u is e e e o e acke e g a e 55
um e o s a io s o a i e um e o s a io s a i e ica a ic co i io s e oug u o o mo e s i c ease wi e i c ease i e acke e g u e ma imum oug u ac ie e i e CSMA/C mo e is ess a a o e oke us mo e is is ue e ig co isio a e a ig oa s i CSMA/C mo e I e oke us e e is o c a ce o e acke s o co i e e e o ma ce o e oke us im o es e y muc a ig e oa s e i e e ce i e oug u o o mo e s ca e see om e a e 5 1
acke e g i y es o o s a io s 1 5 51 CS CS CS CS
77 55 91 7 95 77 5 7 5 7 1 9 71 9 7 1 7 5 7 93 9 75 79 3 5 93 3 97 7 5 1 9 5 9 9
a e 5 1 Ma imum oug u oke us CS CSMA/C om a c ose ook a e esu s i e a o e a e i ca e see a e 5
oug u o a CS MA/C mo e ec eases as e um e o s a io s wi i a gi e e g is i c ease is is ue o g ea e ike i oo o co isio s u e e o ma ce o e oke us mo e im o es as e i e s a io is a ce ec eases ecause ess ime is s e i oke assi g e ysica co igu a io o e s a io s o e us as a ma ke e ec o e oug u o e CSMA/C mo e As see ea ie e oke us mo e is qui e i se si i e o e c a ge i e ysica co igu a io o e mo e ecause e o e ea ue o e oke assi g wi e e same i es ec i e o e ysica co igu a io o e s a io s as a ea y iscusse i e e ious c a e e c a ge o e oug u wi a c a ge i e ysica co igu a io ca e o se e om e a e 5 a 5 3
A e age Ma imum oug u Co igu a io se a a io m oke us CSMA/C
c us e s 51 3 97 57 c us e s 35 97 5 1 c us e s 3 5 1 97 U i o m 3 1 97 1
a e 5 Equa si e c us e s 57
Average Maximum Throughput Configuration separation, m Token Bus CSMA/CD
3 clusters(10,5,5) 455.98 0.97 0.57 2 clusters(15,5) 393.48 0.97 0.59 2 clusters(19,1) 112.75 0.97 0.69
Table 5.3 Unequal sized clusters Comparison of average packet delay The average packet delays for various values of the throughput can be observed from the Table 5.4. It can be seen that at lower loads the CSMA/CD method offers the shortest delay (milli seconds). This is because the stations do not have to wait for a token; they can transmit data immediately. In the token passing method the delay suffered by a packet at lower loads is higher than that of the CSMA/CD. This is because, the overhead due to token-passing is relatively severe at lower loads. But at higher loads the delay suffered by a data packet in the token bus model is very much less than the delay in CSMA/CD model. The higher delay in the CSMA/CD model is due to an increase in frequency of collisions. The variation of the average packet delay with the throughput for
both the models can be seen from Fig 5.1. 5
oug u A e age e ay A e age e ay (mi i sec (mi i sec (CSMA/C ( OKE US
9 7 1 1 5 5 97 3 9 5 9 5 7 5 7 5 3 55 7 3 5 1 5 1
a e 5 A e age e ay e sus oug u
Com a i g e o e e oa o e wo mo e s i is see a e o e e oa i e case o e CSMA/C mo e i c u es e o igi a y a smi e a a acke s a e e a smissio s w e o io s o some o e a a acke s a e os i co isio e o e e oa wi e e o w e o acke s a e o e e o e 1 me e s; s a io s; u i o m y is i u e ; 51 y es
Fig 5.1 60 network. But in the case of the token bus model, since the access to the medium is controlled by the token, the offered load includes both the tokens and the data packets. There are no collisions or retransmissions in the case of the token bus. In the token bus model the offered load will never reduce to zero because the token will always be passing among the stations. Finally ,the CSMA/CD method is more nondeterministic [5], which is often inappropriate for real time work. But the token bus accessing method is more deterministic than the CSMA/CD. The token passing bus method has excellent throughput and efficiency at high load. Discussion of the MAP protocol:
A group of people working under the auspices of IEEE was set up to write official standards for the LANs. Many people were invited to join the committee and contribute to the work. Some of these people came from General Motors, which was carefully looking at LANs. GM wanted to set up a network for factories. An important application of the GM network was factory automation, in which all the robots working on the assembly lines would all be interconnected by a LAN [5]. Since the cars on the assembly lines move by at a fixed rate, whether the robots are ready or not, GM felt it essential to have a LAN in which the worst case transmission time had an upper bound that was known in advance. CSMA/CD does not have this property. Essentially, CSMA/CD works by having all the machines listen to the cable. If the cable is idle, any machine can transmit. If two stations transmit at the same time, there is a collision, in which case they all stop, wait a random period of time, and try again later. In theory, there is no upper bound on the time 1 a s a io mig a e o wai o se a message GM came u wi e A ca e token bus i w ic s a io s ook u s ou - o i us gi i g a e e mi is ic wo s -case e o ma ce a e a a s a is ica o e [5] GM a o e com a ies co ce e wi ac o y au oma io c ea y saw e ee o a o s eci ic o oco s i eac OSI aye is wo k e o e MAP (Manufacturing Automation Protocol) usi g e oke us MA o oco o ows e y c ose y o e OSI mo e MA uses e oke us o e ysica me ium I uses e IEEE a a i k o oco C ( ogica i k Co o i e a a i k aye i co ec io ess mo e as e se ice a ai a e o e e wo k aye
C A E I
COM A A I E ESU S OM A A Y ICA A SIMU A IO S U IES
I e ese c a e a com a iso o e simu a io esu s o ai e i e e ious c a e s wi e o e wi ose o e a a y ica esu s
o e u ose o a a ysis we co si e a g ou o ac i e s a io s o mi g a ogica i g We co si e o e ou o oke assi g assumi g a e e y s a io ge s e oke o y o ce e um e o s a io s a smi i g i o e oke o a io e e s u o w e e o o a s a io as acke s o a smi w ic e e s o e a i a a e o e acke s a eac s a io A i ac i e s a io us asses e oke o i s ogica eig o I ge e a e oug u o a e wo k [1 ] ca e e i e as
w e e is e ime s e o e e wo k a smi i g messages a Y is e o e ea associa e wi e message a s e s e o e ea ca cu a io e e s u o e um e o acke s a smi e ecause o eac acke a s e e oke a o e associa e o e ea s a e a so a smi e I e e a e o acke s wai i g o a smissio a a y s a io e oke is asse wi ou ei g use o e u ose o a a ysis we assume a a e s a io s a e a a 63 packets to transmit. This is the case for the maximum throughput. We consider the best case of the token bus for our analytical model in which the logical neighbors coincide with the physical neighbors. The following assumptions are made in the evaluation of the throughput for the token bus:
(a) An explicit token scheme is considered; and (b) The token is not an integral part of the packet, but is transmitted as a separate packet. (c) The logical ring is assumed to be static, i.e., no addition or deletion of stations is possible.
(d) The total length of the network is assumed to be 1,000 meters. The throughput for this model can be derived to the form as shown in the equation (2)
where a is the normalized propagation delay defined as
where R is the data rate(channel capacity), d is the length of the total length of the network, V is the propagation velocity, which is about two- thirds of the speed 64 of light, and L is the length of the data frame transmitted. and q is the normalized token overhead delay defined as
q = token transmission time / message transmission time (4)
Consider the model with 20 stations and with the following parameters:
Packet length = 4096 bits Token length = 112 bits Speed of the channel = 2 * 10 m/sec. Capacity of the channel = 10 Mbps. Length of the network = 1,000 meters.
Substituting these parameters in the above formula for the throughput , we get the value of the throughput as S = 0.969. For the same packet length of 4096 bits, the simulated model yields a throughput of 0.972.
Decreasing the packet length to 1024 bits, with the same token length of 112 bits the analytical model gives a throughput of 0.88, where as the simulated model yields a throughput of 0.87. Delay analysis: A system with a known worst case is a ring with the stations take turns sending data packets. We consider a logical ring with N stations in which the token is rotating. Each data packet has to wait in the buffer until the station gets 5
e oke o a smi e a a acke We e i e e wo s case oke o a io ime as e ime ake y e oke o go a ou e ogica i g o ce is i c u es e o i g ime o e oke a eac s a io e a e age acke e ay is e a e age ime ake y e a a acke o a smi om e sou ce s a io o e es i a io us e ime s e y e a a acke i e sou ce u e u i e s a io ecei es e oke us i ca e see a e a e age acke e ay e e s o e oke o a io ime e ge e a e essio o e wo s case oke o a io ime o e es case o e ogica is i u io o e s a io s is gi e y e equa io (5
w e e is e a a acke a smissio ime o e mo e wi s a io s a a acke e g o 9 i s com u a io o e wo s case oke o a io ime gi es a a ue o mi i seco s o e same a ame e s e simu a e mo e gi es a wo s case oke o a io ime o 37 mi i seco s o a acke e g o 1 i s e a a y ica mo e gi es e wo s case oke o a io as mi i seco s e simu a e mo e gi es a wo s case oke o a io o 5 mi i seco s
us i is o se e a e simu a e mo e ga e e esu s e y c ose o ose o a a y ica esu s
C A E II CO C USIO S
Simu a io s u y o se e a as ec s o e oke us e o ma ce is ese e A owe oa s e o e ea ue o e oke - assi g amo g e s a io s causes ow c a e u i i a io yie i g ow oug u u a ig e oa s oke us yie s a e ce e oug u wi ou s owi g a y eg a a io e use o o ge acke s yie s e e u i i a io o e c a e u wi a i c ease i e a e age acke e ay e o e e oa o e e wo k is e e e uce o e o ue o e oke assi gs e e i o e o e s a io s a e a a acke s o a smi e e o ma ce o e oke us im o es as e um e o s a io s wi i a gi e e g o e e wo k is i c ease o a gi e um e o s a io s a a gi e acke e g e oug u o e e wo k ec eases as e e wo k e g i c eases e ysica is i u io o e s a io s o e us as a eas e ec o e oug u o e e wo k W e e s a io s a e is i u e u i o m y e i e e ce o e oug u e wee e es case a e wo s case o e ogica is i u io o e s a io s is eg igi e O e o e a w e e s a io s a e is i u e i e o m o c us e s e e is a s ig ec ease i e oug u i e case o e wo s case i a y a com a iso o e e o ma ce o e oke us mo e is com a e wi a o a o e mo e usi g CSMA/C access me o e simu a e mo e i is esis oes o co si e e i mec a ism o e e y e oke us o oco e oke us o oco e a e 7
io i y mec a ism i w ic e a a acke s a e i e e io i ies o u u e wo k e e sio o e ese mo e is ossi e so as o i c u e e io i y mec a ism A e e si e s u y ca e ma e o see ow e a wi wi e a oca e o a ious io i i e a a acke s u e e e a ic ge e a i g e i o me ca e c a ge a e e o ma ce o e mo e ca e s u ie u e ese co i io s
A E I A
CSIM A SOU CE IS I G O SIMU A O
CSIM is a ocess o ie e simu a io a guage w ic is im eme e as a su e se o e C og ammi g a guage I CSIM a ocess is a i e e e og am o oce u e w ic ca e ecu e i a a e wi o e ocesses ( e ocesses o y a ea o e ecu e i a a e CSIM simu a es a a e o simu a eous seque ces o ac i i ies CSIM o i es a e e e se o ea u es w ic aci i a e e im eme a io o ocess o ie e simu a io mo e s ese a e im eme e as a se o e e sio s o e C og ammi g a guage A CSIM og am accesses ese ea u es ia u c io o oce u e ca s om a C og am co s uc e y e mo e e CSIM applied to communication system: I e CSIM mo e e commu ica io me ium is ec a e as a aci i y , w ic is a a a o ec a ca e ese e a e ease y ocesses A aci i y ca e i o e o e wo s a es;
USY o EE I a ocess ese es a EE aci i y e aci i y is assig e o a ocess a e ocess co i ues If a ocess ese es a USY aci i y e ocess is sus e e u i a o e ocess e eases a aci i y W e is a e s e wai i g ocess is gi e co o o e aci i y a e ecu io esumes Summary of the CSIM statements used: CSIM is em e e i C og ammi g a guage I a i io o C e o owi g s a eme s o CSIM a e use i e simu a io 9
Declarations and Initializers: I i ia i e s a e ea y e ecu a e CSIM u c io s w ic e u oi e s o CSIM a a s uc u es I i ia i e s a eme s mus e e ecu e a e e create s a eme i e es ec i e ocesses
1 E E e ; e = e e (" ame" ; e a o e ec a a io ec a es e (a a ia e o y e E E o e a oi e o a e e o ame " ame" E e s ec a e i e sim( i s ocess a e g o a y accessi e E e s ec a e i a o e ocesses a e oca o e ec a i g ocess ese oca e e s ca e asse as a ame e s o o e ocesses oca e e s a e e e e w e e ec a i g ocess e mi a es e a smissio s a e ec a e as events i e CSIM mo e A event is a o e ki o a a o ec as wo s a es; occured a not-occured. W e a ocess waits o a e e w ic is i theoccured s a e e e e is au oma ica y ace i e not-occured s a e a e ocess co i ues W e a ocess waits o a e e w ic is i e not-occured s a e e ecu io o e ocess is sus e e W e some o e ocess sets a e e e e e is ace i e occured s a e ACI I Y ; = aci i y (" ame" ; e a o e ec a a io ec a es (a a ia e o y e ACI I Y o e a aci i y o ame " ame" aci i ies s ou e ec a e wi g o a a ia es a i i ia i e i e sim(mai ocess io o e egi i g o e simu a io a o e mo e 70
Execution Statements: 1. clear(ev); This statement resets event ev to a not-occured state. The event must have been declared to be of type EVENT and initialized using an event statement.
2. hold(t); This statement is used to suspend a process for a simulated time interval of length t. The variable t should be of type float. 3. release(f); This statement is used to release a facility f. f must have been declared to be of type FACILITY and initialized with afacility statement. 4. reserve(f); This statement is used to reserve a facility f. If the facility is already reserved by other processes, then the reserving process is suspended until the facility is released so that it can gain access to the facility. 5. set(ev); This statement sets an event ev to the occured state. 6. t = simtime();
This statement sets t to the current simulated time. This is a function of type float.
7. terminate();
This function is used to end of terminate a process that is executing. A terminat statement is not required if the process exits normally. 8. rerun();
The rerun statement causes the simulation support system to be completely 7 1 reinitialized (with the exception of the random number function). This feature is used to obtain several independent executions of the models.
Operation: The CSIM command file, csim should be copied to the present working directory or a path is to be set properly so as to access it. The CSIM programs are compiled by the command csim filename.c and executable module is placed in the file a.out. All the standard options to the C compiler can be used, including the -o option which specifies a name for the executable module. The command a.out -T will cause a CSIM debugging event trace to be created on the standard output file. The first executable procedure must be named sim. Sim must contain a create function. The header file csim.h must be included with proper path at the beginning of the program. The create function causes a procedure to be set up as a process, capable of executing in a pseudo-parallel manner with other processes. 72
DESCRIPTION OF THE CSIM CODE FOLLOWS a 1 1 3 199 e o age 1 , A A OKE US MO E WI WE Y S A IO S I A E G O ME E S
/ ea e/ i es /
i c u e s io i c u e ma i c u e "/us /mesu a/csim/ i /csim " i c u e u/us /mesu /use s/ k 7 /e e / oke s " / co ai s e a ame e s ike acke e g oke e g ca aci y o e e wo k o aga io s ee e c /
/ G o a a ia e ec a a io s / i o a _ is ; s uc k i _ oke ( ; i s i i ia i e( ; s a ic i s is a ce( = { 5 5 5 5 5( 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 ; / i i ia i i g e i e s a io is a ces / s a ic i o e []-{ 1 3 5 7 9 1 11 1 13 1 15 1 17 1 19 }; / i i ia i i g e ogica o e i g o e s aio s / E E e e 1 e e 3 e e 5 e e 7 e e 9 e 1 e 1 e 11 e 1 e 13 e i e 15 e 1 e 17 e 1 e 19; / ec a i g e e e s / i i _a _s ( o _ y_ e ( ; e e oa o ( ; e e i a om( ; oa OKE _ O _ IME; ou o o a io s; oa e a_ im•; s uc s / s uc u e a o s o e ai s o e cu e a a acke ou mess_a _ ime; oa e _ ime; i i ; c i ic [ _O IC 5i A _s ( O _O A I S];
s uc k /" s uc u e o i g e e ai s o e s a io cu e y o i g e oke /
i s o; / um e (A.- - u ie o i g e oke cu e y A / i oke is ; /A is a ce - u•o e y e oke A/ i e _ i /A is a ce aye i e y e oke a e i c osse e 11 a io i e ogica i g / oa ime_ I e ; /A ime s e y e oke a e s a io A / } o k e ; s a ic i c eck- ; / egi i g- o e im ocess A/
sim(
i i s - ( e - i p, t L017•11!": - 'Tit e 7 ;
ou C oke e Mc y age e ay - acke _ e ay - 4
a 1 1 3 199 e o age
oke _ e ay - message e ay- a e age_ e ay- o a _ e ay- a _ oke e ay- a acke _ e ay- sum_ya ke _ e ay- S- - a _ k_ e ay = o _ k e ay - ; i o _ y_ is a _sou ce_ o a _ es _ o o a _ is cou ; i e _s a a oke _ os;
c ea e ("si e" / i i ia i i g a e e e s / e e e ("e " ; e = e e ("e 1" ; e e e ("e " ; e 3 e e ("e 3" e e e ("e " ; e 5 = e e ("e 5" ; e e e ("e " ; e 7 = e e ("e 7" ; e e e e ("e " ; e 9 e e ("e 9" e 1 - e e ("e 1 " ; e i - e e ("e 11" ; e - e e ("e 1 " ; e 13 e e ("e 13" ; e 1 - e e ("e 1 " ; e 15 = e e ("e 15" ; e 1 e e ("e 1 " ; e 17 - e e ("e 17" ; e i - e e ("e 1 " ; e I9 - o e ("e 19" ; oke - aci i y (" oke " ;
- e (e ; / e i q i e e% / e ( i ;
se _ so (e 5 ; ; e (e ; se (e 7 ; se (e ; se (e 9 ; se (e 1 ; se (e 11 ; se (e 1 ; se (e 13 ; se (e i ; se (e 15 ; se (e 1 ; se (e 17 ; se (e 1 ; se (e 19 OKE _ O _ IME = OKE _ IME +( o _ y_ is / OkE _ O _S EE ; o _ oke _ ime = ( O_O _S A IO S OKE _ O _ IME; o a _ is = s _i i ia i e( ; w i e(cou O O ACKE S a 1 1 3 199 e o age 3 75
o_ oke s = ; cu e _ ime = o ( ; i e _ga = a om( 1 ; i ( s1=-
i e _ga - ; s1=1;} cu e _ ime - - i e g M; a _sou ce_ o a om( O_O S A IG S_ -1 ;/ a om u c io o ick e sou ce s a io /
a _ es _ o - I om( O_O _S A IO S - 1 ; / a om u c io o ick u e es i a io / sim_ ime cu e _ ime; / sim ime kee s ack o e cu e simu a e ime / i (s a io ( a _sou ce_ o]- mess e ime sim_ ime
e _s - i _a _s ( a _sou ce_ o sim_ ime ; i ( e _s -= -1
oke - i _ oke ( oke cu e _ ime ; co i ue; 1 e se a _sou ce_ o- e s ;
w iie( a sou ce o-- a es e
a _ es o- a om( OO S A IOUS-1 ; cou ++; i ( i;9 3 " cou k sim ime ; / use o e uggi g / i i " a _sou ce o a _ us _ o ; / use o e uggi g / i (sim_ ime cu e _ ime { oke i o nrder I 0 ]; t 01:on t it [o [ -- ; i oke - 0; oke e _ i - ;
ei e
oke - i ok( ok cu e ime ; } / is i is o e i / i 3 3 " okc s o oke oko is oke ims e ; oke os - oke oe isk cke e i ; i ( oke o - - a ie ( a ;ou ce o - _; is oke e ay --(s a io ( a sou ce o]- s _ is oke oke is / E u i; 1 e se oke _ e ay = (s a io ( a _sou ce o]- s _ iu + o a is - oke e i - ( okrr, oko I /TOKIMLPPA':::PEEli: i ( oke e ay oke _ e ay1 ( o a _ is -koke koke isk + a 1 1 3 199 e o age 6 s a io [ a _sou ce_ o]- s _ is / OKE _ O _S EE ;
oke e ay = ( a _sou ce_ o - oke s _ o - 1 OKE _ O i ( oke e ay oke _ e ay = ( O_O _S A IO S oke s _ o-1 + a _sou ce_ o OKE O E o oke s += oke _ e ay / OKE _ O _ IME; i (sim_ ime--eu e _ ime oko _ e ay += OKE _ O _ IME; i ( oke ime_s e
oke e ay += OKE _ O _ IME oke ime_s e ; o_ oiCe s++; 1 i ( oke s _ o-- a _sou ce_ o ( oke oke _ is == II oke ime_s e ( oke _ e ay ( o _ oke _ ime- oke ime_s e ; oke e ay ( o a _ is / ; o oke s = O_O _S A IO S-1; 1 o _ k e ay += ( oke e ay + oke _ e ay ; i ( a a ese == 1 / i e u e o e s a io as e acke o se w e e oke is ecei e /
ese e( oke ; o ( OKE _ O _ IME ; / o i g e oke o e o ime w e e s a io as a a a acke o a s e / e ease( oke ; / e ease e oke a ass i o e successo /
e se / i e u e o e s aio oes o a e a a a acke o a smi /
e ease ( oke ; / ass e oke o i s successo / 1 / message_ e ay gi es e o aga io e ay o e a a acke om e sou ce es i a io s a io /
message_ eiay -( y_s [ a _ es _ o]- s _ is - y_s [ a _sou ce_ o]- s _ i MESSAGE O S EE ;
i (message e ay message e ay
/ Message_ e ay is e sum o e a smissio a o aga io e ays o e a a acke /
Message_ e ay = message_ e ay +( oa PACKET_LENGTH/CAPACITY;
/ acke e ay ac ua e ay e e ie ce y a a a acke w ic i c u es e wai i g ime o e a a acke i e u e o e sou ce s a io /
acke e ay Message_ e ay + oke _ e ay + oke _ e ay ;
i ( acke _ e ay ma e ay
ma e ay = acke e ay; / ca cu a es e ma imum e ay e e ie ce a 1 1 3 199 e o age 5 y a a a acke / sum acke _ e ay += acke _ e ay; / o a e ay o a e a a acke s / i (sim_ ime oke _ e ay + oke _ e ay +Message_ e ay ma _ ime ma _ ime sim_ ime+ oke _ e ay - oke _ e ay +Message_ e ay; / ma _ ime is e o a simu a e ime / s a io [ a _sou ce_ o]- mess_ e _ ime im ime + oke _ e ay -1- oke _ e ay ;
o a _ e ay += Message e ay;
/ o a _ e ay is e sum o e a smissio a o aga io e ays o a a a acke s / oke s += o oke s; / kee s ack o e o a oke a smissio s u i g e e i e simu a io / }/ o w i e oo / a _ acke _ e ay sum_ acke _ e ay/ O_O _ ACKE S;/ com u es e a e age acke e ay /
S (( o a e ay/ma ime /1 ; / S com u es e o ma i e oug u o e mo e /
=((( ACKE _ E G O_O _ ACKE S + ( oke s OYE E G /ma _ ime /1 ; / com u es e o ma i e o e e oa /
G = ( ( ACKE E G OO ACKE S /ma _ ime /1 ; / G com u es e o ma i e i u oa /
i (" 1 31 3 3 1 31 % 3 " ( G S a _ acke _ e ay ma e ay ; e u ( ; 1 / e s uc u e e ow is use o i e osi io o e oke a a gi e i s a ce o ime /
s uc k i _ oke ( ese ese ime s rr c k ese ; oa ese ime; { oa ca _ ime = o e ay; i e _s oke _s _ is co -O s- ; c a c; i ( ese 1 e i
o e ay ( o a _ is -s a io [ OO S A IO -1]- s is - ese e i / OKE _ O _S EE ; i ( o e ay - ese ime
ca ime = o e ay; c e k- ; ese s o=o e [O]; ese oke _ is -s a io [o e ( ]]- s _ is ; ese e _ i = ; s-1;
e se a 1 1 3 199 e o age 78 1 ca ime = 1 ; ese e _ i += ese ime OKE O S EE ;
1 i ( ese ime_s e
i ( OKE _ O _ IME ese ime_s e - ese ime
ca _ ime = OKE _ O _ IME - ese ime_s e ; ese ime_s e = ;
1
e se
ca _ ime - 1 ; ese ime_s e += ese ime;
1
i (ca _ ime = 1 1 w e(ca _ ime = ese _ ime { i (co ==1 II c eck==
c eck=1; i (ca _ ime + OKE _ O _ IME ese ime
ese ime_s e = ese _ ime ca _ ime; ca ime-1 ;
e se ca ime += OKE O IME;
i -1
e s = ( ese s o + 1; i (( e _s == O_O _S A IO S s==
oke _s _ is = o a _ is y_s [ e _s 1 1- s is ; o _ e ay = oke _s _ is / OKE _ O _S EE ; i ( o _ e ay ese _ ime - ca _ ime
ese e _ i += ( ese _ ime - ca _ ime OKE _ O S EE ; ca ime=1 ;
C se
ese s _ o = e s ; i ( e _s — O_O _S A IO S e _s =o e ( ]; ese oke _ is = s a io [ e _s ]- s _ is ; ca _ ime += o _ e ay; a 1 1 3 199 e o age 7 79
e _s =o e [ ]; } i ( e _s O_O _S A IO S { e _s =o e [ ]; i (ca _ ime = 1
oke _s is =s a io ( e _s ]- s _ is - ese oke _ is ; i ( oke _s _ is
oke _s _ is =s a io [ e _s ]- s _ is ; e _s =1; 1 o _ e ay oke _s _ is / OKE _ O _S EE ; i ( o _ e ay ese ime ca _ ime ( ese oke is += ( ese ime - ca _ ime OKE _ O _S E i ( ese oe is o a is ese oke is o a _ is ; ca _ ime 1 ; ] e se
e s ; ese s _ o ese oke _ is s a io [ e _s ]- s _ is ; ca _ ime += o _ e ay;
}
co =1; 1
e u ( ese ;
/ e u c io e ow is use o i i ia i e e is i u io o e s aio s gi e i e s a io is a ces a e ogica o e i g o e s aio s / s _i i ia i e(
i o a is a ce; i i is = ; o (i= ;i O_O _S A IO S;i++
y_s [i] (s uc s ma oc(si eo (s uc s ; s a io [i] (s uc s ma oc(si eo (s uc s ; i += s _ is a ce[i]; y_s [i]- s _ is = is ; y_s [i]- mess_a _ ime = ; o a _ is += s is a ce(o e [i]]; s a io [i]- s i o a _ is ; s a io [i]- mess a c ime = ;
o a _ is += y_s [i-1]- s _ is ; e u ( o a is ; o _ y_ e ( / is u c io is use o ca cu a e e o a e g o e e wo k w i e s a io is a ces / ( i i e - ; o (i- ;i O_O _S A IO S;i++ e += s _ is a ce[i]; e u ( e ;
i _a _s ( usy_s a _ ime / is u c io is use o ick u e sou ce s a io ge e a o w e a e s aio s a e usy / i usy s ; o g oa a _ ime;
i i s ; o (i-1;i - OO S A IO S;i++
s i+ usy_s ; i (s O_O _S A IO S-1 s O_O _S A IO S; i (s a io [s ]- mess_ e _ ime a _ ime e u (s ; } e u (-1 ; 81
E E E CES
[1] James Martin Local Area Networks Architecture & Implementation, Arben Group, Prentice Hall, 1984. [2] WEE Standards for Local Area Networks: Token Bus Accessing method ANSI/IEEE Std 802.4-1985 ISO/DIS 8802/4. [3] Stallings, William, "Local Network Performance", IEEE Communications Magazine, vol 22, pp. 3-41, Mr '84. [4] Herb Schwetman, "CSIM: A C-Based, Process-oriented simulation Language," Research Report, Microelectronics and Computer Technology Corporation, Austin, Texas. [5] Tanenbaum A.S. , Computer Networks. Englewood Cliffs, NJ: Prentice- Hall, 1988. [6] Mischa Schwartz., Telecommunication NETWORKS Protocols, Modeling and Analysis. Addison-Wesley Publishing company, 1987. [7] Kernighan, B.W. and D.M. Ritchie, The C Programming Language. Engle-Wood Cliffs, NJ: Prentice Hall, 1978. [8] Herb Schwetman, CSIM Reference Manual(Revision 12). Austin, Texas: Mcc,1987. [9] Anura P. Jayasumana , "Throughput Analysis of the IEEE 802.4 Priority Scheme", IEEE Transactions on Communications, Vol 37, No.6, June 1989. pp.565-571. [10] A.K.Sood, S.Akhtar and K.Y. Srinivasan, "An extended Token bus Protocol for embedded networks", Computers & Elect Engg Vol.14 No.3/4, pp.105-123, 1988. [11] Peter MARTINI, Otto SPANIOL, "Token-passing in high speed backbone networks for campus-wide environments," "Modeling Techniques and Performance Evaluation, Elsevieer Science Publishers , 1987. [12] Nanda K. Kanuri "CSMA/CD LAN Performance: Modeling and simulation in CSIM", Vol 20 Part 3, Proceedings of the Twentieth Annual Pittsburgh Conference held May 4-5, 1989. pp 1003-1007.