基于盲信号分离的同频信号的串行分离技术  

Serial separation of signals with same frequency based on blind signal separation

作　　者：周治宇[1] 陈豪[1] 

机构地区：[1]中国空间技术研究院西安分院,陕西西安710100

出　　处：《信息与电子工程》2009年第4期308-313,共6页information and electronic engineering

基　　金：国家重点实验室基金资助项目(9140C5305020705;9140C5306030707)

摘　　要：传统的时域和频域处理等方法很难分离多个频域的混叠信号,文中提出一种利用盲信号分离技术串行分离同频信号的新方法——FastICA盲分离算法。该方法充分利用盲信号分离技术不需要知道信号先验信息的特性,能正确地、一个一个地分离出在频域中混合在一起的信号,且能分离功率相差100万倍的同频信号。在存在比信号功率大得多的高斯白噪声情况下,FastICA盲分离算法具有良好的分离性能。It is difficult to deal with the separation of signals with same frequency using traditional time-domain or frequency-domain signal processing method. A new blind signal separation method for solving the problem was proposed, which didn＇t need any information of the source signals and could correctly separate the signals with same frequency one by one. The simulations demonstrated that the Fast Independent Component Analysis（FastICA） introduced in the paper could separate the signals with same frequency, and 1 000 000 times difference of power. And the FastICA not only had a good performance under Gaussian white noise condition, but also had a fast convergence. The method introduced in the paper will has a good application prospect.

关 键 词：同频信号分离 盲信号分离 独立分量分析 快速ICA算法 

分 类 号：TN911.7[电子电信—通信与信息系统]