Improved sparse Bayesian learning method for direction-of-arrival estimation in non-uniforni noise

The estimation of direction-of-arrival (DOA) in the presence of non-uniform noise in array signal processing is investigated in this study. The noise covariance matrix is regarded as an arbitrary diagonal matrix in the estimation. The spatial sparsity of the incident signals in different numbers of snapshots is introduced. The signal power spectrum and noise covariance matrix are then estimated through the improved sparse Bayesian learning (SBL) method. Finally, a high-precision DOA estimation of the incident signals is achieved. The proposed method can be viewed as a further expansion of the SBL-based DOA estimator. Computer simulations show the validity of the proposed method. © 2014 Taylor & Francis.