An Efficient Dictionary Learning-Based 2-D DOA Estimation Without Pair Matching for Co-Prime Parallel Arrays

In this paper, we propose an efficient 2-D direction-of-arrival (DOA) estimation method for co-prime parallel arrays. By using the second-order statistics of the received signals with different time lags, a conjugate augmented spatial-temporal virtual array is constructed to extend the array aperture and enhance the degrees-of-freedom (DOFs). Vectorizing the cross covariance matrix of the virtual array, the 2-D DOA estimation can be cast as a 1-D sparsity-based recovery problem. A dictionary learning technique is then proposed to compensate the off-grid bias caused by discretization. Simulation results show the effectiveness of the proposed method as compared with the state-of-the-art methods.