一种高性能、低运算量自适应滤波器算法

由于以往小计算量最小均方算法在减小算法计算量的同时，增大了算法中的梯度估计误差。所以这些算法虽然计算量小但性能却比传统的LMS算法降低了许多。针对该不足，本文用线性限幅函数代替原算法中的符号函数，利用线性限幅函数良好的过渡特性和放大性能提出了一种改进型算法。利用系统辨识实验将改进算法与原算法进行比较表明，改进后的算法不但计算量小而且稳态性能大大提高。该改进算法可被用于高性能。低功耗自适应滤波器的设计与实现。

Research of LMS Adaptive filter Algorithm with Low Computational Load

When computation of LMS(Least Mean Square) algorithm is reduced, error of the gradient estimate is increased. Due to it the performance of these algorithms are worse than traditional LMS algorithm. In order to promote the performance, author presents an improved algorithm which uses the linear limit function to replace the sign function. Experimental results of system identification show that the improved algorithm has better performance. The algorithm can be used for reducing the computational load in software filter and reducing the power consumption in hardware filter.