A combined BP‐MF algorithm for wideband off‐grid DOA estimation with DP prior

Abstract

Abstract Direction‐of‐arrival (DOA) estimation based on sparse Bayesian learning (SBL) framework has attracted extensive attention. The accuracy of on‐grid DOA estimation is restricted by the prescribed grid, while off‐grid approaches resolves the problem of grid mismatch partly. This letter deals with off‐grid DOA estimation problem of wideband signals. A factor graph is designed to describe the problem with Dirichlet process (DP) prior, which clusters the sparse structure of wideband signals. The authors employ the combined belief propagation‐mean field (BP‐MF) rule on the factor graph and lead to a message passing algorithm. Simulation results verify the superiority of the proposed combined BP‐MF algorithm for off‐grid DOA estimation compared to other state‐of‐art methods.