Abstract:

In order to gain high accuracy of two dimensional direction of arrival (2D-DOA) estimation based on co-prime structure, double parallel co-prime array is designed and an off-grid process sparse Bayesian learning algorithm is proposed through non-uniform virtual array. In this way, degrees-of-freedom is extended and off-grid influence on 2D-DOA estimation accuracy is reduced at the same time. Firstly, the double parallel co-prime array is built by vertical extension on the parallel co-prime array. Sccondly, a non-uniform virtual array is developed and the sparse Bayesian learning algorithm is utilized to reconstruct signals. In addition, the energy relationship of the true DOA’s two nearest grids is utilized to conduct a low computational complexity off-grid process through Taylor expansion. Finally, the elimination rule and the selection rule are proposed to fuse estimation results of the two direction parallel co-prime array. Theoretical analysis and simulation results show the validity of the proposed array and DOA estimation method is proved through.