基于NLMS线性相位FIR滤波器的自适应设计

Designing of Linear Phase FIR Filters Using Modified NLMS Algorithm

在以往具有线性相位的滤波器自适应方法中，最常用的是最小均方（ＬＭＳ）方法．但用此法设计时必须小心地选择迭代步长，同时还可能遇到梯度噪声放大问题．文中提出了一种用于设计具有线性相位的有限冲击响应（ＦＩＲ）滤波器自适应方法．它用经过改进的归一化最小均方（ＮＬＭＳ）算法来保证所得结果具有线性相位，且克服了梯度噪声放大问题，同时也不必选择步长．进一步的改进使它的计算更高效。

The linear phase filters are often wanted in many applications. There are several kinds of methods proposed for the designing of FIR filters, including the adaptive designing methods. Such methods based on least mean square (LMS) algorithm encounter the gradient noise amplifying problem and have to select the step size carefully. In this paper, the general normalized least mean square (NLMS) adaptive algorithm was modified to ensure the linear phase of the FIR filter obtained in this way without the gradient noise amplifying. Another modification on this algorithm was adopted to provide more efficient computation needed in the real time implementation.