Broadband two-dimensional off-grid DOA compressive beamforming based on block-sparse Bayesian learning

The planar microphone array's two-dimensional DOA compression beamforming technology offers wide recognition space and clear imaging, holding significant promise for various applications. However, early versions of compressed beamforming encountered issues with grid mismatch, where the true source DOA did not fall on the discrete grid. In this paper, a broadband off-grid DOA compressive beamforming based on block sparse Bayesian learning (BSBL) is proposed. We adopt a block sparse signal reconstruction model to address off-grid DOA compensation, further extending it to a broadband multi-snapshot model for enhanced noise robustness and universality. Leveraging the BSBL algorithm based on fast marginal likelihood maximization (FMLM), we efficiently solve the above model, achieving rapid broadband multi-snapshot off-grid compressive beamforming. Simulation and experimental findings demonstrate the method's superior spatial resolution, effectively mitigating basis mismatch, while multiple snapshots efficiently reduce noise interference in localization. •A broadband 2D off-grid compressive beamforming algorithm using fast block sparse Bayesian learning is proposed.•The computation time of the multi-snapshot block sparse Bayesian algorithm is less affected by the number of snapshots.•This method achieves more accurate source intensity, sparser imaging and lower time cost than the interior point method.