Off-Grid Space-Frequency Environment Sensing for Next Generation Cognitive Radars

This paper proposes a novel 2-D spectrum sensing method to improve environmental awareness capabilities compared with traditional grid-based techniques. Instead of using fixed angles of arrival (AOA), as dictated by a grid of points, the off-grid approaches allow for flexibility in estimating angle displacements at a reduced computational complexity increase. The recovery process is formulated as a regularized maximum likelihood (RML) estimation, leveraging block-sparsity. The optimization is tackled with a maximum block improvement (MBI) method to estimate noise power, the 2-D profile, and angular displacements. Moreover, two refinement strategies are used to improve the angle estimation accuracy. The method is validated through numerical simulations modeling realistic environments.