Minimum Mutual Information Criteria for Convolutive Blind Source Separation

In this paper ,the criteria for convolutive blind source separation(CBSS) is analyzed. We prove that the output stochastic processpsilas spatial independence is the necessary and sufficient condition for achieving separation. Using the definition of spatial independence, we converted this condition into minimum mutual information of output variables. Then we optimize this criteria with the method of steepest descent and get the recovered sources. Computer simulations show that the conclusion of this paper is correct.