基于分离矩阵的步长自适应EASI算法

首次提出将分离矩阵用于对在线盲分离算法中步长自适应控制,并在认真分析相关固定步长和变步长EASI算法的基础上,将其加以运用,提出了一种新的步长自适应EASI算法。该算法步长是基于分离矩阵的,其学习速率由信号的分离程度自适应地选取,因而能很好地解决收敛速度和稳态误差之间的矛盾。计算机仿真结果与理论分析相一致,证实了该算法明显优于其他固定步长或变步长的EASI算法。

Adaptive step-size EASI algorithm based on separating matrix

This paper proposes to use separating matrix to control the step-size of on-line blind source separation algorithms for the first time.After detailed analyzing relevant fixed step-size and variable step-size equivariant adaptive source separation via independent algorithms,the paper presents a new adaptive step-size EASI algorithm.Because the variability of the new algorithm's step-size is based on separating matrix,its learning ratio is chosen adaptively according to separating degree,therefore it can improve convergence speed and reduce the misadjustment error in the steady state simultaneously.Computer simulations confirm the theoretical analysis and show the algorithm performance is superior to other EASI algorithms.