基于低管秩张量分解的互质阵列自适应波束成形算法

Abstract

The coprime array (CPA) has attracted extensive attention in the field of adaptive beamforming (ABF) because of its large array aperture and high degree of freedom. To make full use of this characteristic, several algorithms based on hole filling have been proposed to improve the performance of ABF. However, these algorithms have disadvantages in computation and noise robustness, which are difficult to adapt to complex and changeable environments. To solve this problem, this paper proposes a CPA-ABF algorithm based on low-tubal-rank tensor decomposition. First, the multi-sampling virtual signal matrix of the CPA is rearranged into a tensor form, and the missing cross-correlation information is completed using its low tubal rank. Then, signal parameters are extracted from the completed tensor data and matched with the target a priori. Finally, the ABF weight vector is obtained. The algorithm uses ADMM and Tucker decomposition to improve the efficiency of tensor completion and decomposition. The designed target matching scheme also effectively controls the algorithm error. The simulation results show the advantages of the algorithm in performance and computational complexities compared with existing methods, especially in the case of low signal-to-noise ratios and a small number of samples.