摘要
针对原有在FPGA上实现高速FIR滤波器的移位加算法,进一步分析了算子调度的具体过程,讨论了在不同情况下该算法所能达到的最省资源的算子调度方案,并提出了优化的具体规则。在Xilinx spartan3系列FPGA上的实现结果表明,对于16阶固定系数FIR滤波器,相比于原有的移位加算法以及Xilinx CoregenTM生成的同等规模的分布式算法滤波器,采用优化算法后的FIR滤波器可节省资源分别达11.7%和29.7%。
The authors analyze the detailed process of calculator schedule in high speed FIR (finite impose response) digital filter with add-and-shift algorithm based on FPGA (field programmable gate array). Different calculation situations and related schedule schemes are discussed and a clear rule of optimization is proposed. At last, an example of a 16-order FIR filter is implemented on Xilinx Spartan 3 3s1000ft256 FPGA platform. The occupied resource is 11.7% less than the one generated without optimization and/or 29.7 % less than the one generated by Xinlinx CoregenTM with distribute arithmetic (DA), respectively.
出处
《北京大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2009年第2期222-226,共5页
Acta Scientiarum Naturalium Universitatis Pekinensis