Robust adaptive beamforming in uncalibrated arrays: Sparsity-based array calibration for covariance matrix reconstruction

• In uncalibrated arrays, we propose a sparsity-based joint array calibration and the interference plus noise covariance matrix (INCM) reconstruction framework for robust adaptive beamforming (RAB). • The implicit strong interference is used with the sparse Bayesian learning technique to automatically estimate gain-phase errors and interference direction. • The estimated gain-phase errors are used to correct the reconstructed INCM, which greatly improves the performance of INCM reconstruction-based RAB. Robust Adaptive Beamforming (RAB) based on interference plus noise covariance matrix (INCM) reconstruction suffers significant performance degradation in uncalibrated arrays, where each sensor has unknown gain-phase errors. In this paper, we propose an INCM reconstruction-based RAB framework, where the reconstructed INCM is corrected by the sparsity-based array calibration. First, we present the sparse representation of uncalibrated array output and integrate the gain-phase errors into the model. Then, we perform spatial sparse reconstruction using sparse Bayesian learning (SBL), and the gain-phase errors and interference DOA can be obtained. The accuracy of estimated results can be guaranteed because of the implicit high power of interference, and the estimated results are directly used to reconstruct INCM. Finally, the INCM reconstruction-based RAB is proposed. Note that the potential high power of interference is applicable in practice. By using the proposal on simulation experiments, we observed significant improvements over the compared methods in uncalibrated arrays.