Title :
A solution to blind separation of convolutive communication mixtures in frequency domain
Author :
Duan, Taotao ; Zhang, Xudong
Author_Institution :
Dept. of Electron. Eng., Tsinghua Univ., Beijing, China
Abstract :
In this paper, an efficient algorithm to deal with convolutive blind source separation of communication signals in frequency domain is presented. The classical JADE algorithm is adopted to perform linear complex-valued ICA in each frequency bin. To solve the permutation ambiguity problem, a new permutation technique based on high-order statistics is proposed. Then, a post processing procedure called CMA is added to compensate for signal distortion. Simulations are included to demonstrate the effectiveness of the proposed algorithms.
Keywords :
blind source separation; convolution; higher order statistics; independent component analysis; CMA; blind separation; classical JADE algorithm; communication signals; convolutive communication mixtures; frequency domain; high order statistics; linear complex valued ICA; permutation ambiguity problem; signal distortion; Blind source separation; Finite impulse response filter; Frequency domain analysis; Sensors; Signal processing algorithms; Speech; Time domain analysis; blind source separation; communication; convolutive mixture; frequency domain;
Conference_Titel :
Consumer Electronics, Communications and Networks (CECNet), 2012 2nd International Conference on
Conference_Location :
Yichang
Print_ISBN :
978-1-4577-1414-6
DOI :
10.1109/CECNet.2012.6202000
