有限长卷积反演问题研究  

A Study on Finite length Deconvolution

作　　者：张胜付[1] 吴黎光[1] 夏志云 张克[1] 吕杰[1] 

机构地区：[1]南京理工大学电子工程与光电技术学院

出　　处：《南京理工大学学报》1998年第1期47-50,共4页Journal of Nanjing University of Science and Technology

摘　　要：该文讨论解有限长序列卷积反演的３种方法。一种为时域迭代算法，它根据待求序列长度决定迭代次数，且不存在迭代收敛问题。另一种方法适合于卷积长度、ＦＦＴ点数大约是原卷积序列长度的两倍这种情形，它比经典ＦＦＴ方法节约３３％的计算量。第３种方法与第２种类似，也可节约１５％计算量。与其它方法相比，该文的方法简单易行，运算量少。This paper discusses three methods for problem of finite length deconvolution. The first method is called iterative algorithm for time domain, its iterative times are decided only by the length of unknown seriers. An important result is that the iterative algorithm has no problem of convergency. The secend algorithm is suitable for the length of convolution, points of FFT are about 2 times of the length of known series, it can save about 33% computational cost compared to FFT. The third method is similar to the second, it can save 15% computational cost compared to FFT. So the methods given in this paper are very simple and easy to program.

关 键 词：逆变换 迭代算法 Z变换 卷积反演 FFT 

分 类 号：TN911.1[电子电信—通信与信息系统] O174.22[电子电信—信息与通信工程]