Variational Bayesian DOA Estimation Based on Hidden Markov Models

Aiming at the direction-of-arrival (DOA) estimation problem in pulse-disturbed and wireless transmission environments, this paper presents a variational Bayesian DOA estimation approach utilizing a Hidden Markov Model (HMM). First, a method for joint sparsity of signals and noise is used to build a hierarchical Bayesian structure and, by means of mixed-noise modeling, to simulate practical scenarios. Next, a Forward-Backward algorithm for Hidden Markov Models is used to model the changes in noise state over time and thus capture the temporal correlation of impulse disturbances. Finally, it computes the posterior probability via variational inference and iteratively adjusts the arrival angle for higher accuracy. Simulation results show that, in the presence of mixed-noise conditions, this scheme has achieved relatively accurate direction-of-arrival (DOA) estimation with lower computational costs compared to other Bayesian learning methods.