变步长解相关Volterra LMS算法

Volterra滤波器的非线性使输入向量自相关矩阵包含了输入信号的高阶统计量，导致矩阵特征值扩展很大，因此LMS算法收敛速度一般很慢。从降低输入信号的相关性出发，提出了一种变步长解相关Volterra NLMS算法。解相关能显著加快LMS算法的收敛速度，变步长能够改善算法的稳态性能，两者的有机结合，能明显改善算法性能。仿真结果表明，在不同输入信号相关性情况下，该算法有更好的收敛速度和稳态性能。

Variable step size decorrelation Volterra LMS algorithm

For Volterra filter is nonlinear,the impact of nonlinear operations on the input signal results in that the date correlation matrix consists of higher order statistics and eigen values spread,so its convergence speed is very slow.This paper proposes a kind of algorithm of second order Volterra LMS algorithm using varying step size and decorrelation.Decorrelation can accelerate the convergence speed,and varying step size can improve steady state.The combination of both sides is able to improve the original algorithm＇s convergence performance obviously.The simulation results have shown that the convergence speed of the improved algorithm is faster than that of VLMS algorithm,and the steady state of the improved algorithm is better than that of VLMS algorithm.