基于未知信源数的相干信号DOA估计

DOA Estimation of Coherent Signals Based on Unknown Source Number

针对多重信号分类算法等大多数空间谱估计算法需要进行准确的信源数估计,且当信源数估计出现误差时性能易受影响的问题,提出了一种与子空间分解无关的最优化估计算法。在协方差矩阵的基础上通过约束条件构建新的矩阵,基于相关矩阵的最小特征值对应的特征向量得到最优化权值向量。该算法对这两个矩阵进行预处理,将虚数矩阵变为实值矩阵,然后推导得到空间谱,在谱峰搜索上利用求根的方式。该算法无需信号源个数,并且性能与MUSIC相当,具有解相干能力,能够准确地对相干信号进行估计,在小快拍的情况下仍具有良好的分辨能力。

Aiming at the estimation of source number estimation algorithm requires accurate spectrum multiple signal classification algorithm such as most space,and when the error performance is easily affected by the problem of source number estimation,a subspace decomposition and independent optimization estimation algorithm based on the covariance matrix of the constraints through the construction of new matrix is proposed.The minimum feature based on the expected direction and the correlation matrix of optimal weight vector corresponding feature vectors,the algorithm preprocess the two matrices,the imaginary number matrix into real valued matrix,and then derives the spatial spectrum using the roots of the way in the peak search.The algorithm does not need the signal source number,and its performance is comparable to that of MUSIC.It has the ability to decoherence,can estimate the coherent signal accurately,and still has good resolution in the case of small snapshots.