关键词: 二维DOA估计, L型互质阵, 虚拟扩展, 解耦原子范数最小化, 酉变换, 角度匹配

Abstract: To reduce the computational complexity of the gridless matrix reconstruction method for the L-shaped coprime array and improve the success probability of angle matching under low signal-to-noise ratios, a low complexity gridless two-dimensional direction-of-arrival (DOA) estimation method for the L-shaped coprime array is proposed. This method utilizes the conjugate augmentation method to realize the array virtual extension of the x-axis and z-axis coprime arrays by solving the cross-correlation functions between array elements; According to the idea of matrix-form atomic norm, the array interpolation is realized by the decoupled atomic norm minimization method; The estimated values of the angles between each axis are obtained by the root multiple signal classification (MUSIC) method; According to the spatial consistency between the signal subspace and the array manifold matrix, the angle matching is realized by solving the cost function. To further reduce the computational complexity, the matrix-form atomic norm is combined with the unitary transformation, and the unitary array interpolation is realized by the real-valued decoupled atomic norm minimization method. The results show that, on the one hand, the proposed method improves the accuracy of DOA estimation, reduces the computational complexity, and enhances the probability of successful angle matching; On the other hand, the computational complexity is further reduced by sacrificing part of the DOA estimation accuracy and array degrees of freedom. The feasibility and advantages of the proposed method were verified through simulation experiments.

Key words: two-dimensional DOA estimation, L-shaped coprime array, virtual extension, decoupled atomic norm minimization, unitary transformation, angle matching