An ICA and EC based approach for blind equalization and channel parameter estimation  

作　　者：何振亚 刘琚 杨绿溪 蔚承建 

机构地区：[1]Department of Radio Engineering, Southeast University, Nanjing 210096, China [2]Department of Radio Engineering, Southeast University, Nanjing 210096, China Department of Electronic Engineering, Shandong University, Jinan 250100, China

出　　处：《Science China(Technological Sciences)》2000年第1期1-8,共8页中国科学（技术科学英文版）

摘　　要：A new on-line blind equalization approach is proposed. The approach combines over-sampling technique with independent component analysis (ICA) neural network and can give equalized output on-line employing only the received signal. Based on the fourth-order cumulants and the characteristic of the linear system, the parameters of original channel are also estimated using evolutionary computation (EC). Compared to traditional equalization methods, the proposed algorithm is of simple architecture, does not need learning sequences apart from the observation, and can achieve both blind equalization and system identification. Computer simulations show good performance.A new on-line blind equalization approach is proposed. The approach combines oversampling technique with independent component analysis (ICA) neural network and can give equalized output on-line employing only the received signal. Based on the fourth-order cumulants and the characteristic of the linear system, the parameters of original channel are also estimated using evolutionary computation (EC). Compared to traditional equalization methods, the proposed algorithm is of simple architecture, does not need learning sequences apart from the observation, and can achieve both blind equalization and system identification. Computer simulations show good performance.

关 键 词：independent component analysis HIGHER-ORDER cumulants. EVOLUTIONARY computation BLIND equal-ization . 

分 类 号：O211.67[理学—概率论与数理统计]