Research On High Precision Direction Of Arrival Estimation Based On Fast Sparse Bayesian Learning

Direction of arrival(DOA)estimation is an important part of array signal processing,which is widely used in many engineering fields,such as radar,sonar,medicine,etc.In underwater applications,high-precision DOA estimation can provide finer details,providing richer information for subsequent target recognition and classification.The sparse Bayesian learning(SBL)algorithm can provide high precision DOA estimation,but SBL algorithm has high computational complexity and grid mismatch problem.Based on efficient SBL algorithm and off-grid algorithm,this paper aims to simultaneously improve the DOA estimation accuracy and computational complexity of DOA estimation algorithm.The research contents are as follows:This paper systematically summarizes the research on DOA estimation methods at home and abroad.The advantages and disadvantages of traditional DOA estimation,classical sparse reconstruction and SBL algorithm are analyzed.In order to solve the problem of heavy computation and grid mismatch of SBL algorithm,the fast SBL algorithm and off-grid algorithm are mainly studied,which provides a basis for the follow-up research work of this paper.The theoretical model of DOA estimation is introduced,traditional DOA estimation,compressive sensing and SBL algorithms are studied in detail.The advantages and disadvantages of various algorithms are analyzed more clearly through simple simulation experiments.In order to reduce the computation amount of SBL algorithm,the inverse-free sparse Bayesian learning algorithm is studied.By introducing the basic properties of smooth function,it is extended to the form of complex valued matrix,and the variational method is used to solve the iterative formula of each hidden variable parameter.A fast multisnapshot inverse-free sparse Bayesian learning algorithm is proposed.The performance of the proposed algorithm is verified by comparing it with the advanced algorithm and processing the experimental data.In order to make up for the grid mismatch problem of the grid algorithm,the off-grid sparse reconstruction DOA estimation model was introduced.At the same time,to accelerate the algorithm speed,the space alternate variational estimation method was used to solve the offgrid model,and an off-grid space alternating multisnapshot sparse Bayesian learning algorithm was proposed.And after that,fast mean-field sparse Bayesian learning algorithm was analyzed in detail in order to make up for the fact that the two proposed fast algorithms could not minimize the correct variable target problem and could not adapt well to large-scale problems.The above two algorithms are unified into a mean-field.A complex-valued multisnapshot offgrid fast mean-field sparse Bayesian learning algorithm is proposed.By comparing it with the advanced algorithm and processing the experimental data,the algorithm is verified to improve the accuracy and running time.