Abstract
The recent trend for extremely massive antenna arrays and high frequencies facilitates localization and sensing, offering increased angular and range resolution. In this letter, we focus on the emerging technology of Dynamic Metasurface Antennas (DMAs) and present a novel framework for the design of their analog beamforming weights, targeting high accuracy near-field localization at the THz frequency band. We derive the Cram\'{e}r-Rao Bound (CRB) for the estimation of the positions of multiple users with a DMA-based receiver, which is then utilized as the optimization objective for the receiver's discrete tunable states of its metamaterials. Leveraging the DMA structure, we reformulate the localization objective into a constrained Rayleigh quotient maximization problem, which is efficiently solved via two schemes: one based on projection and a greedy one. Our simulation results verify the validity of our near-field localization analysis, showcasing the effectiveness of the proposed near-field localization designs over the optimum exhaustive search solution and state-of-the-art schemes.