Orthogonal Polarization Array DOA Estimation Based on Matrix-distributed Reconstruction

‍ ‍The accuracy of covariance matrix estimation greatly affects the performance of subspace algorithms. The ideal covariance matrix estimation is obtained through infinite signal operations； however， in practice， a limited number of samples are mostly used for covariance matrix estimation. Scholars have shown that the covariance matrix of an array is located in a subspace composed of all possible directional vectors of the received signal. The complete reconstruction matrix composed of subspaces can be used to reconstruct the covariance matrix. This method requires integrating all possible received signals of the array and obtaining a reconstruction matrix composed of its principal components. A sampling covariance matrix reconstruction method based on a uniform circular array composed of orthogonal dipoles is proposed to reduce the sampling covariance matrix error and computational complexity in low snapshot count and low signal-to-noise ratio down sampling. The overall array was divided into subarray 1 and subarray 2， where only a spatial phase difference is present within the subarray and no polarization-sensitive characteristics. The spatial phase difference between subarrays was the same； however， the polarization sensitive characteristics were different. Based on the structural characteristics of a uniform circular array， a special reconstruction matrix is provided. The special reconstruction matrix does not depend on signal characteristics and is only related to the circular array structure. Through theoretical analysis and simulation testing， it can be concluded that the special reconstruction matrix is superior to the reconstruction matrix integrated in the full angle domain and polarization domain in terms of dimension and complexity. This method applies the reconstruction algorithm to the field of polarization-sensitive arrays while reducing computational complexity. Subsequent simulation tests at low signal-to-noise ratio and low number of snapshots showed that the reconstructed covariance matrix can effectively improve the orthogonality between the signal subspace and the ideal noise subspace. Using the reconstructed covariance matrix for polarization space spectrum estimation improves the source resolution in harsh environments.