欠定条件下的空间同频信号盲分离

Underdetermined Blind Separation of Overlapped Signals of the Same Frequency

当分离多个频域的混叠信号,尤其是遇到观测信号少于源信号时,传统的时频域信号处理方法出现困难。文章介绍了一种可以实现同频信号成功分离的方法——基于两阶段法的欠定盲信号分离法。该方法不需要源信号以及混合信道等先验知识,仅根据观测信号就能分离出混叠在一起的多个同频信号。仿真结果表明,该方法分离效果良好,实现了对空间同频信号的准确分离。

It is difficult for the traditional time-frequency domain analysis method of signals to separate the same frequency signals,especially when the observation signals are less than original signals.In this paper a method is introduced which can separate the same frequency signals—the underdetermined blind signals separation based on the two-stage method.This method does not need to know the information of source signals and the mixing signals.It can realize the separation of overlapped signals of the same frequency only using the observation signals.Simulation results show that the algorithm can effectively separate the overlapped signals of the same frequency from observation signals.