基于信息熵的混合信号分离

Separation of Mixed-signals Based on Information Entropy

盲信号分离是一种从混合信号中分离出独立信号的有效方法。在源信号和信道均未知的情况下,只需要满足源信号相互独立这一条件即可分离。先通过白化去除混合信号之间的线性相关性,再通过最小化信号的信息熵,消除各信号之间的高阶相关性,从而达到分离的目的。

Blind signal separation is an effective method to extract independent signals from mixed-signals. When the source signals and the channel are unknown, source signals that statistically independent can be extracted from mixed- signals. Firstly, removing the linear correlation by whitening, and then to eliminate higher-order correlation among the mixed signals by minimize the entropy, so as to achieve the purpose of separation.