一种改进的快速自适应滤波算法

An improved fast adaptive filtering algorithm

为了解决传统自适应滤波最小均方(Least Mean Square,LMS)算法中收敛速度与稳态误差之间的矛盾,提出了一种改进算法。该算法在已有变步长LMS算法基础上,引入遗忘因子来影响步长的更新。仿真表明,改进后的算法比原算法不但具有更快的收敛速度,而且具有更小且稳定的稳态误差。

In order to settle the conflict between convergence speed and steady-state mean-error of traditional LMS（Least Mean Square） equalization filtering algorithm,an improvement to the existing variable step LMS algorithm is proposed by introducing the forgetting factor,with which the update of step length can be controlled.Simulation results show that the improved algorithm is faster convergence and smaller mean-error compared with the original one.