Abstract: Aiming at the problem that the estimation accuracy of the existing estimation methods of the direction of arrival (DOA) was low under the condition of low signal-to-noise ratio, small fast beat and multiple sources, we proposed a DOA estimation method based on parallel coordinate descent algorithm. Firstly, the airspace was uniformly divided into equal angles, and the super-complete redundant dictionary was constructed. Secondly, the sparse signal was reconstructed by using the idea of parallel coordinate descent algorithm, and the sparse coefficient matrix of the signal in spatial space was obtained. Finally, the l₂-norm of the sparse matrix row vector was mapped to the spatial grid to obtain an accurate DOA estimate. The simulation experiment results show that the proposed method is superior to subspace algorithm, greedy algorithm and convex optimization algorithm under the conditions of low signal-to-noise ratio, small fast beat and multiple sources, and has lower root mean square error (RMSE), higher DOA estimation accuracy and higher operational efficiency.

Key words: direction of arrival estimation, compressed sensing, parallel coordinate descent algorithm, vector hydrophone array signal