A modified ESPRIT algorithm based on a new SVD method for coherent signals

In this paper, an improved high-resolution ESPRIT algorithm for direction of arrival (DOA) estimation of coherent signals is proposed, which is based on a new SVD method, the MSVD-ESPRIT algorithm for short. The main idea is to construct a matrix with the maximum eigenvector according to certain rules, then fix this matrix and get two signal subspaces by singular value decomposition; at the last we use the rotation invariant to get DOA. Simulation results indicated that compared with other improved algorithm, this improved ESPRIT algorithm´s resolution and robustness is obviously better, especially in low SNR case.