Sparse Bayesian learning for wideband direction-of-arrival estimation via beamformer power outputs in a strong interference environment

Sparse Bayesian learning (SBL) offers a useful tool for wideband direction-of-arrival (DOA) estimation, but its performance is limited in the presence of strong interferences. To solve this problem, this letter attempts to extend the SBL to estimate DOAs via the beamformer power outputs (BPO) because the beamformer can efficiently suppress the interferences. A Bayesian probabilistic model effective for the BPO is proposed. Based on this, a BPO-based SBL method is put forward by adopting the variational Bayesian inference to estimate the DOAs from the BPO. Simulation and experimental results confirm the good performance of the proposed method.