Grid Evolution: Joint Dictionary Learning and Sparse Bayesian Recovery for Multiple Off-Grid Targets Localization

Abstract

In this letter, we propose an efficient grid evolution multiple targets localization framework for off-grid targets. First, we propose a more accurate localization model, enabling grid evolution by considering all the grids as random variables to be inferred. Then, the localization problem is formulated as a joint sparsifying dictionary learning and sparse signal recovery problem. Finally, the joint optimization problem is solved under the general framework of sparse Bayesian learning (SBL). Different to previous SBL based localization algorithms, we adopt the hierarchical Laplace distribution for sparse prior, rather than the Sudent’s t distribution. We compare the proposed framework with state-of-the-art off-grid targets localization algorithms as well as Cramer–Rao lower bound. Numerical simulations highlight the improved performance of the proposed framework in terms of localization error, noise robustness, and required number of measurements.