Blind Signal Separation Algorithm In Convolutive Mixture Model

This paper is about the research of blind signal separation in convolutive mixture model. At first, we introduce the basic model and the mathematical principle of blind signal separation. The latest research evolution and the application in engineering are also introduced. And then, we bring forward the time-domain model and the criteria of recognition of MIMO convolutive mixture system, and introduce briefly the general time-domain algorithm of convolutive mixing signal.The relationship of blind deconvolution and blind signal separation is discussed in this paper. We deduct the iteration equation and performance parameter of blind deconvolutionalgorithm based on maximum likelihood estimation criteria and extend the algorithm of blind signal separation in instantaneous mixture into the field of blind deconvolution.We introduce the frequency-domain model of blind signal separation in convolutive mixture and put forward the principle analysis and mathematical deduction of a frequency-domain algorithm of blind signal deconvolution based on high order cumulant. At last, we simulate this algorithm in MIMO liner convolutive mixture system using MATLAB software and analyse the results.