基于箕舌线的变步长LMS自适应算法

Variable Step Adaptive Filtering LMS Algorithm Based on Tongue-Like Curve

通过建立步长因子μ与误差信号 e之间的非线性关系 ,提出一种新的基于箕舌线的变步长 LMS算法 ,并将其应用于通信降噪。该算法除了具有传统固定步长 LMS算法计算量小、稳定性较好、简单、易于实时处理等优点外 ,计算机仿真结果表明 ,其收敛速度、稳定性以及跟踪速度优于 SVSLMS算法和 NLMS算法 ,且不需进行指数运算 ,计算复杂度低于 SVSLMS算法 ,用于通信降噪取得了较好的效果。

A novel variable step adaptive filtering algorithm is presented by building a nonlinear function relationship between μ and error signal e, and it is applied to noise canceling system of communication. The algorithm is simple and easy to be implemented. Theoretical analysis and computer simulation show that the properties of the algorithm such as convergence speed, steady state error are better than those of NLMS algorithm and SVSLMS algorithm. It needs not operate the exponent, so it is easy to be implemented than SVLMS algorithm. And it obtains good results in noise canceling system of communication.