CAJ | 학술논문

传统位反算法在对快速傅里叶变换(FFT)的输出进行重排序时，只能以基-2形式输入数据。为此，提出一种新的基于映射迭代策略的算法，实现对任意基形式FFT输入的输出重排序，包括对映射迭代过程收敛性的证明。得出当FFT的输入点数N确定时，混合基形式下迭代次数为lbN的结论，为硬件架构的确定提供依据。
전통위반산법재대쾌속부리협변환(FFT)적수출진행중배서시，지능이기-2형식수입수거。위차，제출일충신적기우영사질대책략적산법，실현대임의기형식FFT수입적수출중배서，포괄대영사질대과정수렴성적증명。득출당FFT적수입점수N학정시，혼합기형식하질대차수위lbN적결론，위경건가구적학정제공의거。
Considering the problem that the traditional bit-reversal algorithm is only fit for radix-2 Fast Fourier Transform(FFT) data reordering, a novel approach based on mapping recursion is proposed for radix-r FFT data reordering. The mapping recursion process is proved to be convergent, of which the recursion times is lbN, while the parameter N refers to the input number of FFT, thus affording the guideline for the implementation by hardware.