计算DFT谱的改进递归算法  

Improved recursive algorithm of DFT implementation

作　　者：王宏伟[1] 郑文秀[1] 

机构地区：[1]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《系统工程与电子技术》2013年第11期2263-2268,共6页Systems Engineering and Electronics

基　　金：国家自然科学基金(61271276)资助课题

摘　　要：计算离散傅里叶变换(discrete Fourier transform,DFT)谱的传统算法是快速傅里叶变换(fast Fourier transform,FFT)算法。FFT算法适合全景频谱分析,不适合并行计算。在只需要计算少数感兴趣频点频谱时,基于递归计算的Goertzel算法比FFT算法更加快速、灵活。在介绍Goertzel算法的基础上,又提出了两种计算DFT谱的改进递归算法。其中,改进递归算法2比Goertzel算法更加合理,更加适合同时进行数据采集和数据处理。递归算法的递归单元所需要的加法器、乘法器等硬件资源少且相对固定,有利于利用多个递归单元构造多路并行工作的滤波器组来分析频谱,使信号处理的速度得到提高。The implementation of discrete Fourier transform （DFT） is the traditional fast Fourier transform （FFT） algorithm which is suited to the global spectrum analysis, not suited to the parallel calculation. When several interested spectrum need to be calculated, the Goertzel algorithm based on the recursive calculation is faster and more flexible than the FFT algorithm. After introducing the Goertzel algorithm, two novel recursive algorithms to calculate DFT spectrum are proposed. The improved recursive algorithm 2 is more appropriate for the data acquisition and processing at the same time. The needed gdder and multiplier of the recursive elements are relatively fixed so that it is favorable for constructing the parallel filter band with several recursive elements to make the signal processing faster.

关 键 词：快速傅里叶变换算法 GOERTZEL算法 递归算法 并行计算 

分 类 号：TN971.6[电子电信—信号与信息处理]