动态局部可重构的FFT算法研究与优化  

Research and Optimization of Dynamic Partial Reconfigurable FFT Algorithm

作　　者：冯伟[1,2] 吴晓[1,2] 张凯龙[1,2] 

机构地区：[1]西北工业大学计算机学院,西安710129 [2]陕西省嵌入式系统技术重点实验室,西安710129

出　　处：《计算机测量与控制》2013年第3期779-781,784,共4页Computer Measurement &Control

基　　金：国家自然科学基金(61103004);陕西省重大科技创新计划课题(2010ZDKG-83)

摘　　要：对于复杂物理-信息融合的嵌入式智能系统而言,环境动态变化将驱动系统计算热点的动态改变;为了适应动态可变的实时计算需求,有效增强计算平台的灵活性及计算效能,文章重点结合典型FFT算法研究并提出了动态快速可重构的算法有效改进;首先通过改进的LZW数据压缩算法,减少系统配置时间;进而,使用基于模块化的重构方式对传统并行结构的FFT进行改进,充分利用FP-GA上的并行资源来提高FFT的动态重构速度和数据处理速度;实验结果证明,从以上三方面进行优化,能有效的提高FFT的动态重构速度,更好地满足任务处理的灵活性和实时性要求。For complex physical -- Information Fusion embedded intelligent systems, environmental dynamic changes will drive system calculation hot to be changed dynamically. In order to adapt to the dynamic real--time computing needs, and effectively enhance the flexibil- ity and computational efficiency of the computing platform, combined with the study of typical FFT algorithm, the paper effectively improves the dynamic fast reconfigurable algorithm. First we take advantage of improved LZW data compression algorithm to reduce system setup time; Furthermore, we improve the traditional FFT parallel structure based on a modular approach for dynamically reconfiguration, in which we can take full advantage of parallelism resources on the FPGA to improve data processing speed of the FFT and the reconstruction speed of dynamic module data. Experimental results show that these three aspects of the optimization can effectively improve the dynamic reconfigura- tion of the FFT speed, and better meet flexibility requirements and real time requirements.

关 键 词：快速傅里叶变换 动态可重构 模块化的重构方式 压缩算法LZW 并行处理结构 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]