A Robust Sparse Bayesian Learning-Based DOA Estimation Method With Phase Calibration

Usually, the array manifolds are assumed to be known perfectly in the radar systems, but the imprecise knowledge substantially degrades the performance of estimating the direction of arrival (DOA). In this paper, the DOA estimation problem in the multiple-input multiple-output (MIMO) radar system is addressed. Assuming the antennas are well-calibrated in a uniform linear geometry, the phase errors among antennas caused by the temperature variation and other environmental conditions are unknown. By exploiting the target sparsity in the spatial domain, a new sparse model combining with phase errors is formulated. Different from the existing high-resolution and sparse-based estimation methods, we directly estimate the phase errors in the sparse reconstruction processing. By adopting the hyperparameters, a novel sparse Bayesian learning (SBL)-based method, named sparse Bayesian learning with phase errors (SBLPE), is proposed. An expectation maximum (EM)-based method is given to realize the SBLPE method efficiently. Additionally, all unknown parameters, including the noise, noise variance, the sparse matrix, etc., are theoretically derived from the prior distributions. Simulation results show that the SBLPE method outperforms the state-of-the-art methods, including the sparse-based and the subspace-based methods with acceptable complexity.