摘要
AfluidbuffermodelwithMarkovmodulatedinput-outputratesisconsidered.Whentrafficintensityisnearitscriticalvalue,thesystemisknownasinheavytraffic.Itisshownthatasuitablyscaledsequenceoftheequilibriumbuffercontentshasaweakordistributionallimitunderheavytrafficconditions.ThisweaklimitisafunctionalofadiffusionprocessdeterminedbytheMarkovchainmodulatingtheinputandoutputrates.Thefirstpassagetimeofthereflectedprocessisexamined.ItisshownthatthemeanfirstpassagetimecanbeobtainedviaasolutionofaDirichletproblem.ThenthetransitiondensityofthereflectedprocessisderivedbysolvingtheKolmogorovforwardequationwithaNeumannboundarycondition.Furthermore,whenthefastchangingpartofthegeneratoroftheMarkovchainisaconstantmatrix,therepresentationoftheprobabilitydistributionofthereflectedprocessisderived.Upperandlowerboundsoftheprobabilitydistributionarealsoobtainedbymeansofasymptoticexpansionsofstandardnormaldistribution.
出版日期
2004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)