摘要
Aprincipalchallengeinsupportingreal-timevideoservicesoverATMistheneedtoprovidesynchronousplay-outinthefaceofstochasticend-to-endnetworkdelays.Inthispaper,anintelligenttrafficsmoothmechanism(ITAM)isproposedtomeetthecontinuityrequirementwhichiscomposedofaback-propagationneuralnetwork(BPNN)trafficpredictoraplay-outbuffer,andafuzzyneuralnetwork(FNN)basedplay-outratedeterminator.TheBPNNtrafficpredictoronlinepredictsthemeanpacketrateofthetrafficinthefutureinterval(FI)andtheFNNisdesignedtoadaptivelydeterminatetheplay-outtimeaccordingtothenumberofpacketsinthebufferandthetrafficcharacterpredicted.Simulationresultsshothatcomparedtothewindowmechanism,ITSMachieveshighcontinuitywithaccepteddelay.Furthermore,ITSMcanbeadaptivelymodifiedtomeettheQoSofdifferentkindsofservicesbyFNNparametertraining.
出版日期
2000年01月11日(中国期刊网平台首次上网日期,不代表论文的发表时间)