基于源数估计的盲源分离

Blind Source Separation Based on Signal Number Estimation

在信号源少于传感器观测到的混合信号时,未知信号源数目的估计一直是已有盲分离算法中一个未解决的问题。就盲分离在阵列信号处理中的应用,提出了一种基于四阶累积量的源数估计方法。由于四阶累积量对高斯噪声的抑制作用,从而可提高估计的分辨性能。给出了详细的计算方法,并用蒙特卡洛试验证实了该方法优于通常的源数估计算法。将其用于盲源分离,通过实例证明了该方法的正确性和有效性,从而解决了盲分离中信号源个数的估计问题,为盲源分离技术的应用进一步奠定了基础。

The determination of the number of unknown source signals when there are fewer sources than the mixtures in blind source separation （BSS） has always been an unsolved problem. Based on the application of BSS in array signal processing, a method of estimation signals number was proposed. Due to fourth-order vumulants, the method can improve the estimation performance. The detailed solution to the problem of number estimation was explained, and Monter-Carlo trials demonstrated that the performance of this method was better than the usual algorithms. The validity of the method in BSS is proved through experiments, Thus the method can solve the problem of estimation of signal number in BSS, it paves the way to wider application of BSS methods.