摘要
针对传统LMS算法运算量大收敛性能差的缺点,提出了一种减小运算量并且提高收敛性能的LMS自适应滤波算法。首先从理论上介绍并分析了两种自适应滤波算法--量化误差算法(运算量小)和变步长算法(收敛速度快),接着将这两种算法的优点有机结合,提出了一种改进算法。通过在MATLAB下的辨识仿真研究和对误差曲线的分析,证明了结合后的改进算法在运算量和收敛速度方面都优于传统LMS算法。算法对于IP电话中回声消除的自适应滤波问题提供了一个较好的算法。
Based on the traditional LMS algorithm with large amount of computing and poor convergence performance, an improved LMS algorithm is proposed. It has a decreased computing amount and an improved convergence property. First, two algorithms are introduced and analysed: quantization error algorithm and variable step algorithm, and then the advantages of these two algorithms are combined. Through MATLAB simulation, it is proved that the new algorithm is superior to the traditional LMS algorithm both in computing amount and convergence speed, which provides a better algorithm for adaptive filtering in echo cancellation in the IP telephone system.
出处
《计算机仿真》
CSCD
北大核心
2009年第4期324-327,364,共5页
Computer Simulation
关键词
回声消除
自适应滤波算法
量化误差算法
可变步长算法
Echo cancellation
Adaptive filter algorithm
Quantization error algorithm
Variable step algorithm