A high-resolution DOA estimation method based on the Newton-like method

Abstract

Focusing on the problem that when the direction of arrival (DOA) interval of target signals is small and the performance of traditional DOA estimation methods, such as the multiple signal classification (MUSIC) method and reweighted smoothed $l_0$ norm (RSL0) method, is seriously degraded, this paper proposes a high-resolution DOA estimation method based on the Newton-like method to solve it. Firstly, the covariance matrix of the received signal is processed by column vectorization, and the signal sparse model is established according to it. Then, the sparse signals are weighted by the MUSIC method to promote sparse recovery. Secondly, a set of exponential functions is used to approximate the $l_0$ norm, the Lagrangian function is established by combining the constraint function, and the positive definiteness of the Hessian matrix of the Lagrangian function is analyzed. Due to the Hessian matrix of the Lagrangian function cannot always be guaranteed to be positive definite, a Newton-like method is proposed according to the positive definite part of the Hessian matrix to achieve sparse recovery and high-resolution DOA estimation. The simulation results confirm the strength of the proposed method in DOA resolution.