采用信号相邻采样点联合估计的欠定盲源分离算法

摘要

Two-step strategy is the basic method of under-determined blind source separation currently,and the estimation of mixing matrix is a prerequisite for recovering the source signals. We consider the situation that the mixing matrix has been estimated,and matrix subspace method is applied to estimate source signals. In order to improve anti-noise capability of the underdetermined blind source separation algorithm,the relevant characteristics of adjacent sampling points of signals are used to determine the ownership of the signals at any time. Finally,simulations show the performance and validity of the proposed algorithm.%两步策略是当前欠定盲信号分离的基本方法,混叠矩阵的估计是恢复源信号的先决条件.考虑在混合矩阵已经估计的前提下,通过矩阵子空间的方法估计源信号,并且利用信号相邻采样点的相关特性,联合判断任意时刻信号的归属,提高欠定盲源分离算法的抗噪能力.通过仿真实验显示了该算法的性能及实用性.