Research On Two-dimensional Direction Of Arrival Estimation Algorithms For L-shaped Array

Spatial spectrum estimation, also known as Direction of Arrival (DOA), which is a basic problem ofarray signal processing, is widely used in civil and military realms, such as sonars, communications. AL-shaped array, which has a simple structure and is easy to be implemented, is often used in the fieldsof estimation of two-dimensional direction of arrival. We investigate the two-dimensional direction ofarrival estimation algorithms for L-shaped array, which has theoretical significance and practical values.The main work in this paper is summarized as follows:(1) A2D-DOA estimation algorithm based on Estimation Signal Parameters via RotationalInvariance Techniques (ESPRIT) with L-shaped array is proposed. Without spectral peak searching,this algorithm works well. The algorithm has much better2D-DOA estimation performance thanconventional ESPRIT algorithm, obtains automatically paired elevation angle and azimuth angleestimation, and can identify more DOAs than conventional ESPRIT algorithm.(2) A2D-DOA estimation algorithm based on Propagator Method (PM) with L-shaped array ispropose. This algorithm does not require peak searching and eigenvalue decomposition of receivedsignal covariance matrix and obtains automatically paired elevation angle and azimuth angleestimation. It has great2D-DOA estimation performance and has a low computational cost.(3) A2D-DOA estimation algorithm based on ROOT-MUSIC algorithm with L-shaped array isinvestigated. This algorithm gets the parameter estimations containing the elevation angle and theazimuth angle via determination of roots of a polynomial. A cost function is constructed to get thepaired elevation angle and azimuth angle estimation. This algorithm does not require peak searchingand has a better2D-DOA estimation performance than ESPRIT and PM.(4) A2D-DOA estimation algorithm based on trilinear decomposition with L-shaped array isinvestigated. This algorithm links the2D-DOA estimation problem to the trilinear model. Trilinearalternating least squares is employed for obtaining the estimated direction matrices.2D-DOA isestimated according to the LS principle. The algorithm needs a few iterations to achieve convergenceand has much better2D-DOA estimation performance than ESPRIT.