Improved Bayesian compressive sensing-based direction of arrival estimation

The spurious peaks of Bayesian compressive sensing with Gaussian prior model are considered when used in Directions Of Arrival(DOA)estimation.In order to suppress spurious peaks,we improve a Bayesian prior model and propose a kind of DO A estimation method where we estimate the noise background of the beam output,mark directions of targets using binary indicator variables and bring in a signal variance based noise power estimation method,and then combine the model with variational Bayesian inference to improve the performance of DOA estimation.Numerical simulation and experimental results show that,compared with the state-of-the-art nethods,the improved method yields fewer spurious peaks,higher peak-background ratio and superior detection performance.