非平稳卷积混合信号的盲分离

Convolutive Blind Source Separation of Non-stationary Source

传感器采集到的信号是由多目标源、环境噪声等经多途径卷积混合的形式。为有效地去除环境因素如干扰、传输延迟等的影响,提出一种新的盲信号分离方法。利用非平稳信号的多重去相关和最小二乘准则来估计混合矩阵A或解混矩阵W以及信号和噪声功率。实验结果表明,该算法具有良好的分离效果。

Signals obtained by sensor arrays are convolutive mixtures for multiple-sources and environment noises through multiple-paths. A novel approach of the blind source separation is proposed to remove the interferences of time-delay and environment noises. It takes up the multiple decorrelation approach assuming non-stationary signals and uses a least square optimization to estimate A or W as well as signal and noise powers. Experiment results illustrate the algorithm can get high separation performance for both non-stationary and convolved mixing signals.