• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2006, Vol. 28 ›› Issue (4): 60-62.

• 论文 • 上一篇    下一篇

一种解相关变步长归一化LMS自适应算法

龙剑友[1,2] 王梓展[1] 夏舜晖[1] 段正华[1]   

  • 出版日期:2006-04-01 发布日期:2010-05-20

  • Online:2006-04-01 Published:2010-05-20

摘要:

本文对变步长(VSS)自适应滤波算法进行了分析,针对输入信号高度相关时算法收敛速度下降导致性能下降的问题,提出了一种解相关变步长归一化LMS自适应算法,引入相关性原理,使得算法保持良好的收敛性能。计算机仿真结果与理论分析相一致。

关键词: 归一化 变步长 解相关 LMS

Abstract:

In this paper, we discuss and analyze the variable step-size adaptive algorithm(VSSLMS), in which the descent of its convergence speed leads to the reduction of performance when input signals are heavily correlated. As a result, we propose an uneorrelated variable step-size normalized LMS adaptive algorithm. Due to the principle of correlation, the algorithm can achieve good performance. The simulations agree to the theoretical analysis.

Key words: normalization, variable step-size, uncorrelated, LMS