基于零均值特性的箕舌线变步长 LMS算法

Tonguelike Curve VSLMS Algorithm Based on the Characteristic of Zero Mean

摘要:

针对基于箕舌线变步长 LMS算法步长更新公式易受噪声干扰的问题,根据高斯白噪声的零均值特性,提出基于零均值特性的箕舌线变步长 LMS算法,使算法的抗噪声干扰能力增强。若 2算法选取相同参数 α、β,则基于零均值特性的箕舌线变步长 LMS算法相对于箕舌线变步长 LMS算法具有小的稳态误差。在保证算法收敛的条件下,基于零均值特性的箕舌线变步长 LMS算法相对箕舌线变步长 LMS算法具有较快的收敛速度。

Abstract:

The tonguelike curve VSLMS algorithm’s step-formula can be disturbed easily by noise jamming. Ac. cording to the characteristics that the mean of white Gaussian noise was zero, tonguelike curve VSLMS algorithm based on zero mean was put forward in order to improve tonguelike curve VSLMS algorithm’s ability of anti-noise. If two algorithms chose the same parameters α、β , tonguelike curve VSLMS algorithm based on zero mean would have less steady-state error than tonguelike curve VSLMS algorithm. Under the condition that the two algorithms were convergent, the convergence rate of tonguelike curve VSLMS algorithm based on zero mean was bigger than the convergence rate of tonguelike curve VSLMS algorithm.