Robust Sparse Bayesian Learning for DOA

We formulate statistically robust Sparse Bayesian Learning (SBL) for Direction of Arrival (DOA) estimation from Complex Elliptically Symmetric (CES) data using a general approach based on loss functions. Simulation results for DOA estimation are obtained for several choices of loss functions: Gauss, multivariate $t$ (MVT), Huber, and Tyler. The root mean square DOA error is discussed for Gaussian, MVT, and $\epsilon$ -contaminated data. The robust SBL estimators perform well in the presence of outliers and for heavy-tailed data and almost like classical SBL for Gaussian data.