摘要
TherearetwomajorapproachesforBlindSignalSeparation(BSS)problem:MaximumEntropy(ME)andMinimumMutualInformation(MMI)algorithms.BasedontherecursivearchitectureandtherelationshipbetweentheMEandMMIalgorithms,anExtendedME(EME)algorithmisproposedbyusingprobabilitydensityfunction(pdf)estimationoftheoutputstodeducethecorrespondingiterativeformulasinBSS.Basedonthesimulationresults,itcanbeconcludedthattheproposedalgorithmhasbetterperformancesthanthetraditionalMEalgorithminconvolutemixtureBSSproblems.
出版日期
1999年02月12日(中国期刊网平台首次上网日期,不代表论文的发表时间)