A combined LMS with RGA algorithm of the co-channel separation system

Abstract

In this paper, we present the method which is combined least mean square (LMS) algorithm with real-parameter genetic algorithm (RGA) for optimizing the coefficients of adaptive filter in the amplitude-locked loop (ALL) separation system. The proposed algorithm is adopted to control the value of the step size in order to improve the slow rate of convergence. Therefore, the mean-square error (MSE) could be minimized under the channel signal-to-noise ratio (SNRc). Another purpose is to successfully separate the co-channel signals by eliminating signal distortion and noise interferences. Finally, we compared the simulation results of proposed algorithm to the traditional LMS algorithm. We obtained the performance of LMS+RGA is better than adaptive LMS algorithm.