盲分离中LMS和RLS两种算法的比较

Comparison of the LMS and RLS Algorithms for Blind Source Separation

盲源分离试图从给定的一组混合观察数据中恢复未知的独立信源。本文简要阐述LMS和RLS两种自适应算法,用Matlab对一组混合通信信号进行分离实验,考察算法的特性和效果,并进行比较分析。结果表明:LMS算法与RLS算法相比,RLS算法的收敛性能更好一些,而在RLS算法中,自然梯度RLS法又是最优的。

Blind source separation attempts to recover unknown independent sources from a given set of observed mixtures. The two adaptive algorithms -LMS and RLS are introduced in this paper. The separation simulation of a set of mixed communication signals is constructed using Matlab. The characteristics and effects of the two algorithms are also observed. Through the comparison of them, the result shows that the RLS algorithm has better convergence than the LMS algorithm and the natural gradient al- gorithm is the best in the RLS algorithm.