基于信源数目估计的超定盲信号分离

针对盲信号分离中信道噪声大、信号分离效果差等问题，在传统主分量分析和特征值分解方法的基础上，提出了一种基于信源数目估计的超定盲信号分离方法。首先，采用主分量分析和最大似然估计方法分别对混合矩阵和噪声协方差进行估计，用于对信道噪声的估计与去除；然后，采用交叉验证法对源信号维数进行估计，实现盲信号分离。为了验证提出算法的分离效果，对轻拖尾与轻拖尾混合信号以及重拖尾与轻拖尾信号混合情况进行仿真实验验证，结果表明该算法具有良好的分离效果。

Overdetermined blind signal separation based on source number estimation

For the problems of large channel noise and poor signal separation effect in blind signal separation,an overdetermined blind signal separation method based on source number estimation is proposed on the basis of traditional principal component analysis(PCA) and eigenvalue decomposition(EVD) methods.Firstly,PCA and maximum likelihood(ML) estimation methods are used to estimate the mixing matrix and noise covariance for estimation and removal of channel noise,respectively.Then the cross-validation method is used to estimate the source signal dimension and achieve blind signal separation.The simulation experiments for the mixed signal of light trailing and light trailing and the mixed signal of heavy trailing and light trailing are conducted to verify the separation effect of the proposed algorithm.The results show that the algorithm has good separation effect.