摘要
Recently,anewsoft-insoft-outdetectionalgorithmbasedontheMarkovChainMonteCarlo(MCMC)simulationtechniqueforMultiple-InputMultiple-Output(MIMO)systemsisproposed,whichisshowntoperformsignificantlybetterthantheirspheredecodingcounterpartswithrelativelylowcomplexity.However,theMCMCsimulatorislikelytogettrappedinafixedstatewhenthechannelSNRishigh,thuslotsofrepetitivesamplesareobservedandtheaccuracyofAPosterioriProbability(APP)estimationdeteriorates.Tosolvethisproblem,animprovedversionofMCMCsimulator,namedforced-dispersedMCMCalgorithmisproposed.Basedontheaposteriorivarianceofeachbit,theGibbssamplerismonitored.Oncethetrappedstateisdetected,thesampleisdispersedintentionallyaccordingtotheaposteriorivariance.Extensivesimulationshowsthat,comparedwiththeexistingsolution,theproposedalgorithmenablesthemarkovchaintotravelmorestates,whichensuresanear-optimalperformance.
出版日期
2008年03月13日(中国期刊网平台首次上网日期,不代表论文的发表时间)