Deterministic protocols for real-time communication in multiple ...
Deterministic protocols for real-time communication in multiple ... Deterministic protocols for real-time communication in multiple ...
S. Norden et al. / Computer Communications 22 (1999) 128–136 135effect of the different input parameters, on SR, we canconclude the following:• SR vs Load: In Fig. 2(a), as load increases, there is apredictable fall in SR values for all the three protocolsas more number of messages are generated and morenumber of collisions, thus reducing SR.• SR vs a: In Fig. 3(a), as the laxity factor increases, thedeadline associated with every message increases,resulting in an increase in the number of messagesthat are transmitted successfully, hence increasing SR.• SR vs N: In Fig. 4(a), as the number of nodesincreases, the SR falls. This is the result of an increasein the number of messages, as well as a larger epochtime, which leads to more messages missing theirdeadlines.• SR vs P/Nt: In Fig. 5(a), as P/Nt is decreased, theperformance will fall expectedly, as a smaller ratioimplies a larger overhead as result of notifier transmission,compared to actual packet transmission.An interesting observation from the graphs is that theECU is not always influenced by the SR. Normally, onewould expect that as SR decreases, the ECU will alsodecrease since less number of slots are used for successfultransmission. Sometimes, this relation may not hold wheneven if less number of messages are transmitted, thecombined service times may exceed those of the droppedmessages, which will still increase the ECU (see Fig. 2(a)and (b)). This explanation is supported by the increasingNTL values (in Fig. 2(c)) with increasing load, for all theprotocols, implying that messages with larger service timesare transmitted.5. ConclusionsIn this article, we have identified the need for real-timeprotocols to support multipacket real-time message transmissionwhich is the form of communication in a typicalreal-time system. We have highlighted the inadequacy ofthe existing protocols to support this requirement. As a solutionto the problem, we have proposed protocols with besteffort service for multiple-access networks, namely, theLDCR and MDCR protocols. The MDCR is an extensionof DCR protocol while the LDCR protocol is based on theconcept of message deferment.We have demonstrated the effectiveness of our best effortprotocols through simulation, for a wide range of parametersfor multiple performance metrics, by comparingthem with a protocol (PBCSMA) recently proposed forthe same problem. Our simulation results reveal that theLDCR protocol performs better than the PBCSMA andMDCR for all the three performance metrics – successratio, effective channel utilization, and normalized transmissionlength. We are currently studying the effect of dynamicallyassigning more than one vertex in the CR tree to anode, allowing additional messages to be sent in the sameepoch.References[1] K. Arvind, K. Ramamritham, J.A. Stankovic, Window MAC protocolsfor real-time communication services, Technical Report, Universityof Massachusetts, USA, 1992.[2] J.I. Capetanakis, Tree algorithm for packet broadcast channels, IEEETrans. on Communications 27 (10) (1979) 1476–1484.[3] D.E. Comer, D.L. Stevens, Internet working with TCP/IP-Volume 2,Prentice-Hall International Inc, Englewood Cliffs, NJ, USA, 1994.[4] Intel Corp., CSMA/DCR deterministic access LANs, PresentationViewgraphs, Folsom Microcomputer Group, Feb. 1988.[5] K.H. Kim, C. Serro, Robustness of real-time local area network protocols,Computer Communications 20 (1997) 169–176.[6] L. Kleinrock, Packet switching in radio channels: Part 1 – Carriersense multiple access modes and their throughput-delay characteristics,IEEE Trans. on Communications 23 (12) (1975) 1400–1416.[7] J.F. Kurose, M. Schwartz, Y. Yemini, Multiple-access protocols andtime-constrained communication, ACM Computing Surveys 16 (1)(1984) 43–70.[8] G. Le Lann, N. Rivierre, Real-time communications over broadcastnetworks: The CSMA-DCR and the DOD-CSMA-CD protocols,Technical Report, Institute National De Recherche En InformatiqueEt En Automatique, France, Mar. 1993.[9] N. Malcolm, W. Zhao, Hard real-time communication in multipleaccessnetworks, Real-time Systems 8 (1995) 35–77.[10] N. Malcolm, W. Zhao, The timed-token protocol for real-timecommunications, IEEE Computer 27 (1) (1994) 35–41.[11] S.K. Oh, G.J. Macewen, Task behaviour monitoring for adaptive realtimecommunication, Real-time Systems 11 (1996) 173–195.[12] R. Yavatkar, P. Pai, R. Finkel, A Reservation-based CSMA protocolfor integrated manufacturing networks, IEEE Trans. on Systems, Manand Cybernetics 24 (8) (1994) 1247–1257.[13] K. Ramamritham, Channel characteristics in local-area hard real-timesystems, Computer Networks and ISDN Systems 13 (1987) 3–13.[14] J.A. Stankovic, K. Ramamritham, Hard real-time systems, ComputerSociety Press of IEEE, 1730, Massachussets Avenue, Washington,DC, 1993.[15] O. Ulusoy, Network access protocol for hard real-time communicationsystems, Computer Communications 18 (12) (1995) 943–948.[16] V. Upadhya, Design and Analysis of Real-time LAN protocols, M.S.Thesis, Department of Computer Science and Engineering, IndianInstitute of Technology, Madras, Jan. 1996.[17] W. Zhao, K. Ramamritham, Virtual time CSMA protocols for hardreal-time communication, IEEE Trans. on Software Engineering 13(8) (1987) 938–952.Samphel Norden obtained his B.Tech degree in Computer Science andEngineering from the Indian Institute of Technology, Madras, in 1998.Currently, he is a research student at the Computer Science Department,University of Washington at St. Louis, USA. His research areas ofinterests are real-time networks and active networks.S. Balaji obtained his B.Tech degree in Computer Science and Engineeringfrom the Indian Institute of Technology, Madras, in 1998.Currently, he is a research student at the Computer Science Department,University of Illinois at Urbana Champaign, USA. His researchareas of interest are parallel and distributed computing and computernetworks.
136S. Norden et al. / Computer Communications 22 (1999) 128–136G. Manimaran obtained his B.E. degree in Computer Science andEngineering from Bharathidasan University, Thiruchirappalli, in1989, his M.Tech. in Computer Technology from the Indian Instituteof Technology, Delhi, in 1993 and his Ph.D. in Computer Science andEngineering from the Indian Institute of Technology, Madras, in 1998.His research interests are resource management in parallel and distributedreal-time systems, real-time networks and fault-tolerantcomputing.C. Siva Ram Murthy obtained his B.Tech. degree in Electronics andCommunications Engineering from the Regional Engineering College,Warangal, in 1982, his M.Tech. in Computer Engineering from theIndian Institute of Technology (IIT), Kharagpur, in 1984 and hisPh.D. in Computer Science from the Indian Institute of Science(IISc), Bangalore, in 1988. From March to September, 1988, he workedas a Scientific Officer in the Supercomputer Education and ResearchCentre at IISc. He subsequently joined IIT, Madras, as a Lecturer ofComputer Science and Engineering. He became an Assistant Professorin August 1989, and is currently an Associate Professor at the sameplace. He has held visiting positions at the German National ResearchCentre for Information Technology (GMD), Sankt Augustin, Germany,the University of Washington, Seattle, USA, and the University of Stuttgart,Germany. He is a recipient of the Seshagiri Kaikini Medal for thebest Ph.D. thesis and also of the Indian National Science AcademyMedal for Young Scientists.
- Page 4 and 5: S. Norden et al. / Computer Communi
- Page 6 and 7: S. Norden et al. / Computer Communi
136S. Norden et al. / Computer Communications 22 (1999) 128–136G. Manimaran obta<strong>in</strong>ed his B.E. degree <strong>in</strong> Computer Science andEng<strong>in</strong>eer<strong>in</strong>g from Bharathidasan University, Thiruchirappalli, <strong>in</strong>1989, his M.Tech. <strong>in</strong> Computer Technology from the Indian Instituteof Technology, Delhi, <strong>in</strong> 1993 and his Ph.D. <strong>in</strong> Computer Science andEng<strong>in</strong>eer<strong>in</strong>g from the Indian Institute of Technology, Madras, <strong>in</strong> 1998.His research <strong>in</strong>terests are resource management <strong>in</strong> parallel and distributed<strong>real</strong>-<strong>time</strong> systems, <strong>real</strong>-<strong>time</strong> networks and fault-tolerantcomput<strong>in</strong>g.C. Siva Ram Murthy obta<strong>in</strong>ed his B.Tech. degree <strong>in</strong> Electronics andCommunications Eng<strong>in</strong>eer<strong>in</strong>g from the Regional Eng<strong>in</strong>eer<strong>in</strong>g College,Warangal, <strong>in</strong> 1982, his M.Tech. <strong>in</strong> Computer Eng<strong>in</strong>eer<strong>in</strong>g from theIndian Institute of Technology (IIT), Kharagpur, <strong>in</strong> 1984 and hisPh.D. <strong>in</strong> Computer Science from the Indian Institute of Science(IISc), Bangalore, <strong>in</strong> 1988. From March to September, 1988, he workedas a Scientific Officer <strong>in</strong> the Supercomputer Education and ResearchCentre at IISc. He subsequently jo<strong>in</strong>ed IIT, Madras, as a Lecturer ofComputer Science and Eng<strong>in</strong>eer<strong>in</strong>g. He became an Assistant Professor<strong>in</strong> August 1989, and is currently an Associate Professor at the sameplace. He has held visit<strong>in</strong>g positions at the German National ResearchCentre <strong>for</strong> In<strong>for</strong>mation Technology (GMD), Sankt August<strong>in</strong>, Germany,the University of Wash<strong>in</strong>gton, Seattle, USA, and the University of Stuttgart,Germany. He is a recipient of the Seshagiri Kaik<strong>in</strong>i Medal <strong>for</strong> thebest Ph.D. thesis and also of the Indian National Science AcademyMedal <strong>for</strong> Young Scientists.
Change language