脉冲噪声环境下基于矩阵差分的远近场混合源定位

为了提高脉冲噪声环境下基于二阶协方差矩阵差分（COV-MD）的远近场混合源定位算法的估计性能，本文提出了基于分数低阶协方差矩阵差分（FLOC-MD）和基于压缩变换协方差矩阵差分（CTC-MD）的远近场混合源定位算法。所提出的算法首先利用一维MUSIC谱峰搜索获得远场源信号的方位角估计，然后利用矩阵差分法实现远近场信号源的分离得到扩展的近场源分数低阶协方差矩阵（或压缩变换协方差矩阵），最后在利用类旋转不变方法（ESPRIT-Like）估计得到的近场源方位角的基础上，再次利用一维MUSIC谱峰搜索获得近场源距离的估计。计算机仿真结果表明:CTC-MD算法和FLOC-MD算法在强脉冲和低信噪比情况下的估计性能都要明显优于COV-MD算法和其他基于二阶统计量的远近场混合源定位算法，同时CTC-MD算法的性能要好于FLOC-MD算法并且不依赖于脉冲噪声的先验信息。 

Mixed Far and Near Field Sources Localization Based on Matrix Differencing Under Impulsive Noise Environment

? ?In order to improve the performance of mixed far and near field sources location algorithm based on matrix differencing of the second-order covariance matrix （COV-MD） under impulsive noise environment， two revised algorithm based on matrix differencing of the fractional low-order covariance matrix （FLOC-MD）and compressed transform covariance matrix （CTC-MD） are proposed in this paper. Firstly， the proposed algorithms use one-dimensional MUSIC spectral peak search to obtain the direction of arrival （DOA） of far field source signal. Then the matrix difference algorithm is used to eliminate the far field components and reconstruct the extend near field components. Finally， based on the DOA of near-field source which is estimated by the ESPRIT-like method， the range of near field source is estimated by another one-dimensional MUSIC spectral peak search. Computer simulation results show that the performance of CTC-MD algorithm which is not depend on the prior information of the impulsive noise is better than FLOC-MD algorithm，and the performance of CTC-MD and FLOC-MD algorithm are both better than that of COV-MD algorithm and other mixed far and near field sources location algorithms based on second-order statistics in the case of strong impulsiveness and lower signal-to-noise ratio.