Fast two-dimensional sparse Bayesian learning with application to pulse Doppler multiple-input-multiple-output radars

Sparse Bayesian learning (SBL) for multiple-input-multiple-output (MIMO) radars can achieve significant performance in target localisation by exploring sparsity in the target space. In this study, the authors develop a fast two-dimensional (2D) SBL algorithm using Laplace priors (2D SBL-LPs) by deriving two auxiliary algorithms. The proposed algorithm is implemented for pulse Doppler MIMO radars in order to estimate the target parameters in three dimensions of angle, Doppler, and range. Simulation results show that the 2D SBL-LP compared with famous Bayesian compressed sensing algorithm not only drastically reduces the computational complexity, but has better performance in targets parameters estimation. Furthermore, in estimation of range-angle-Doppler parameters, the 2D SBL-LP outperforms the matched filter, 2D smoothed L0, 2D iterative adaptive approach, fast orthogonal matching pursuit (FOMP), and spectral projected gradient for l(1)-norm minimisation algorithms.