一种新的变步长LMS自适应滤波算法

A novel variable step size adaptive filtering LMS Algorithm based on sample function

通过建立步长因子μ与误差信号e之间的非线性关系,提出一种新的基于抽样函数的变步长LMS算法,并进行了计算机仿真.结果表明,该算法除了具有传统LMS算法计算量小、稳定性较好、简单易于实时处理等优点外,其收敛速度、稳定性以及跟踪速度均优于传统固定步长LMS算法抽样函数、SVSLMS算法.

This paper presents a novel variable step adaptive filtering algorithm by establishing a nonlinear function relationship between μ and error signal e. The algorithm is as simple and easy to implement as the classical LMS algorithm does. Both the theoretical analysis and computer simulation prove that the properties of the algorithm such as convergence speed and steady state error are better than those of classic LMS algorithm and SVLMS algorithm.