基于AccelDSP的自适应滤波器设计

AccelDSP Based Adaptive Filter Design

给出了一种自适应滤波器的设计和FPGA的实现方法.简要分析了最小均方误差(LMS)、归一化最小均方误差(NLMS)和延迟最小均方误差(DLMS)自适应滤波器的工作原理,指出了3种滤波器的区别和联系.详细介绍使用Xil-inx公司的AccelDSP高层次设计工具设计自适应滤波器的流程与方法,说明了可综合Matlab语言的结构特点,提高硬件使用效率的途径等设计细节,得到了设计代码和仿真结果.设计经过硬件时序仿真验证,可以在FPGA中直接实现.仿真结果说明,NLMS滤波器对于音频信号有很好的收敛特性,DLMS滤波器可以达到很高的工作频率设计.设计方法对于其它数字信号处理系统的设计和实现具有参考价值.设计过程表明AccelDSP工具可以显著地提升数字信号处理系统的设计效率.

A design approach and FPGA implementation method of adaptive filters is presented in this paper. Working theory and differences of LMS, NLMS and DLMS adaptive filters is shortly analyzed. Using Xilinx＇s AccelDSP tool, the details of design flow, synthesizable MATLAB constructs and improving hardware using efficiency method, with part of the code and simulation results, is presented as well. The design is verified by timing simulation, and can be implemented on FPGA directly. It is shown by simulation results that NLMS filters have a better convergence character, while DLMS filters have a higher working frequency. The design approach can be a good reference to other digital signal processing systems. And the design process demonstrates that AccelDSP can remarkably improve the design efficiency of DSP systems.