盲源分离用于DOA估计研究

方位估计是水下目标定位、跟踪的前提。提出了一种复数域奇异值分解盲源分离方法，估计出了阵列流形，并利用阵列流形表现出的相位关系完成了目标方位估计。通过仿真实验与MUSIC高分辨方位估计方法和相关文献的方法进行了对比，结果表明该方法能很好地实现目标方位的实时估计，且性能更佳。

A study of DOA estimation by blind source separation

DOA （direction of arrival） estimation is the basis of underwater target location and tracking. Based on average time-delay correlation matrices, a new kind of complex blind source separation using the singular value decomposition is proposed. The array manifold and source signals are estimated via the method thereafter. Then the DOA estimation is performed by using the phase relationship of the array manifold. Its efficiency and stability are tested by simulation, which shows that it can accomplish real-time estimation of the target direction, and can estimate the corresponding target signal. And it is superior to the MUSIC method and the method proposed in relevant literatures.