利用局域子空间投影提高子空间类DOA 估计算法的谱分辨力

本文提出一种新的用于子空间类DOA估计算法的阵列协方差矩阵噪声子空间投影矢量的选取方法——局域子空间投影(LSP).该投影矢量选取方法有利于压低真实信源方位附近,非信源方位对应的谱曲线高度,从而提高子空间类高分辨DOA估计算法的分辨力.LSP算法的估计偏差和信噪比分辨门限明显低于MUSIC算法,而估计方差几乎与MUSIC相同.计算机仿真结果证明了文中对LSP算法性能理论分析的正确性和LSP算法的有效性.

Improving Spectral Resolution of Subspace-Based DOA Estimation Algorithms by Localized Subspace Projection

A novel method for selecting projection vectors in the noise subspace of array covariance matrix is proposed with a view to improving the spectrum resolution of subspace-based DOA estimation algorithms,which is called LSP (Localized Subspace Projection).The new projection vector(s) obtained by LSP can depress the spectrum amplitude at the spatial location in close vicinity to sources' actual spatial positions while remaining the specular peaks on the source positions.The LSP algorithm has lower peaks location bias and SNR resolution threshold than MUSIC,while remains almost the same peaks location variance with MUSIC.Computer simulation results verify the theoretical performance analysis of the given LSP algorithm and demonstrate the effectiveness of LSP algorithm.