Estimation of Time Varying 2D DOAs Based on the Variational Bayesian Inference

This paper considers the estimation of multiple 2D directions-of-arrival (DOAs) of sources, wherein both the number of sources and the DOA of a specific source are time-varying. Our work leverages dynamic models of 2D-DOAs in the spatial domain to predict the 2D-DOAs, and the likelihood function for a square array is formulated. Such modeling allows us to exploit the Bayesian estimation framework to recursively compute the posterior of multiple 2D-DOAs. In order to reduce computational complexity and ensure the conjugacy of prior and posterior PDFs, variational Bayesian inference (VBI) is employed further. Simulation results show that our approach outperforms the conventional 2D-MUSIC method, particularly under low signal-to-noise ratio (SNR) conditions, thus verifying the effectiveness of the proposed algorithm.