Mixing Matrix Estimation in Underdetermined Blind Source Separation Based on Objective Function and Artificial Bee Colony Algorithm

Abstract

To solve the problems of existing methods for mixing matrix estimation in underdetermined blind source separation such as the defect that separation performance is vulnerable to the initial value and low estimation accuracy, probability model is used to describe the distribution of the observed signals in this paper, and the objective function based on the maximum likelihood is obtained, which turns the problem of mixing matrix estimation into the problem of parameters estimation. The objective function is optimized by the artificial bee colony algorithm and mixing matrix estimation is obtained. Compared with some existing methods, the proposed method has more precise estimation accuracy.