基于FPGA的LMS自适应滤波器设计

Design of LMS adaptive filter based on FPGA

提出一种基于LMS(Least Mean Square)自适应算法的滤波方法,介绍该方法在低频信号滤波上的应用及在FPGA平台上实现。传统数字滤波器FIR、IIR滤波器针对不同的系统和干扰信号,其滤波参数不固定。因此,在窄带信号的滤波处理中,传统滤波器对信号滤波降噪的效果往往受到衰减增益限制。提出的方法先以CORDIC(Coordinate Rotation Digital Computer)算法产生的正弦信号来调制采样信号,根据采样信号与基准信号误差使其权向量沿负梯度方向终止于维纳解。该方法在类直流输入及带宽窄的情况下仍能有效过滤高频噪声,读取低检测信号幅值。理论分析和实验结果表明,与传统滤波方法对比,该滤波方法在处理窄带信号的滤波上更具有优势。

This paper presents a filtering method based on adaptive algorithm LMS(Least Mean Square),which introduces the application of the method in the low frequency signal filtering, and implements it in the FPGA platform. The traditional digital filter FIR, IIR have different filter parameters for different systems and interference signals. Therefore, the effect of the traditional filter on the signal filter noise reduction is often limited by the attenuation in the narrow band signal processing. In this paper, sampling signal is modulated by sinusoidal signal which is generated by the CORDIC algorithm. According to the error of the sampling signal and the reference signal, Weight vector is terminated in the wiener solution along the negative gradient direction. The high frequency noise of the DC input can be effectively filtered in narrow bandwidth. Theoretical analysis and experimental results show that compared with the traditional filtering method, the filtering method has more advantages in the processing of narrow band signals.