基于正交匹配追踪的远近场混合源定位算法

Mixed Far-Field and Near-Field Source Localization Algorithm Based on Orthogonal Matching Pursuit

针对远近场混合源定位问题,提出了一种基于正交匹配追踪的参数估计算法。首先,提取接收数据协方差矩阵的反对角线元素作为新的数据向量,通过空域角度网格划分对向量进行稀疏表示;然后,利用正交匹配追踪(OMP)算法计算远场和近场目标的角度参数;接着,对空域进行距离网格划分,并构建与距离参数相关的稀疏重构问题;最后,利用OMP算法实现远近场目标的区分以及距离参数的求取。与以往的参数估计算法相比,该算法无需计算高阶统计量和进行矩阵分解,具有较低的计算复杂度;同时,由于OMP算法受信噪比和快拍数影响较小,故该算法在低信噪比和小快拍数下具有较高精度。

Aimed at mixed far-field and near-field source localization problem, a parameter estimation algorithm based on orthogonal matching pursuit(OMP) algorithm is proposed. Firstly, the anti-diagonal elements of received data′s covariance matrix are extracted as a new vector. By dividing the spatial into a large number of angular grids, the vector′s sparse representation is carried out. Secondly, the angular parameters of mixed far-field and near-field sources are calculated with the OMP algorithm. Thirdly, the spatial is divided into several range grids, and a sparse reconstruction problem related with range parameter only is constructed. Finally, with the OMP algorithm, the targets in far field or near field are distinguished, and the range parameters are calculated. Compared with traditional parameter estimation algorithm, the algorithm does not need high-order statistics and any matrix decomposition. Thus, it has lower computational complexity. Meanwhile, since the OMP algorithm is less affected by the signal noise ratio(SNR) and the snapshots, the algorithm maintain a high accuracy in low SNR and small snapshots scenarios.