Two-Dimensional DOA Estimation with Partial Damaged Sensors in Rectangular Array Based on Tensor Reconstruction

In the field of direction of arrival (DOA) estimation, there exists no problem of angle merging and it has low sidelobe characteristics with the uniform rectangular array. But when sensors in the uniform rectangular array are partially damaged, the two-dimensional DOA estimation performance of the original algorithms will seriously decline, and even have a performance failure in some cases. To solve the problem, we study the inexact augmented Lagrange multiplier method (IALM). Through utilizing features of low-rank matrix, obtained by the tensor unfolding in different modes, in the tensor model of the received signal, we propose the IALM-ESPRIT algorithm. Compared with the direct ESPRIT algorithm, the performance of two-dimensional DOA estimation is significantly improved when partial sensors damaged.