DOA Estimation Based on Second-Order Difference Co-Array for Coprime Arrays

In this paper A different approach based on the reconstruction of the data covariance matrix is presented. In the presented approach, by vectorization of the reconstructed matrix, the covariance matrix is reconstructed in two steps. Hence an extensive network of virtual sensors is created, which is introduced as a second-order difference co-array. Then, by using the well-known spatial smoothing technique full-rank covariance matrix is obtained for DOA estimation algorithms. Therefore the MN sources can be detected using a physical array with M + N sensors. The degrees of freedom (DOF) of the second-order difference co-array are significantly increased as opposed to the physical array which provides only N − 1 degrees of freedom. As a result, for a certain number of array elements, in addition to the resolution improvement, the number of detectable targets is sharply increased. Simulation results verify the effectiveness of the proposed method and MUSIC-based or ESPRIT-based DOA estimators could be designed to perform DOA estimation.