In this paper, we focus on estimating Direction of Arrival (DOA) and removing heavy clutter embedded with measurementnoise. A correlated Gaussian process is chosen to model destructive effects of clutter. Also, a white Gaussian process isselected to describe measurement noise caused by sensor array. After adding these distortions to the off-grid model, we utilizeSparse Bayesian Learning and principal component analysis (as a preprocessing stage) in order to remove these distortionsas well as estimating of true DOAs. Finally, at the end we will show how ignorance of clutter from model or combine it withmeasurement noise degrade DOA estimation. This will be demonstrated by various numerical simulations

Keywords