Dynamic Hybrid Beamforming for RIS-Aided Near-Field Integrated Sensing and Communications

Exploiting near-field spherical wavefronts can improve the performance of integrated sensing and communication (ISAC) systems. Existing studies on near-field ISAC mainly consider either fully-digital or static hybrid beamforming, which may not be able to efficiently exploit the distance-dependent degrees of freedom (DoFs) of near-field channels. In this paper, we propose a dynamic hybrid beamforming architecture that adaptively adjusts the number of active radio frequency (RF) chains and further utilize a reconfigurable intelligent surface (RIS) to enhance the performance in ISAC systems. We formulate an energy efficiency maximization problem which aims to jointly optimize the hybrid precoding matrices at the base station, phase-shifts at the RIS, and hybrid combining matrices at the user equipment, subject to the constraint on guaranteeing the minimum beam pattern gain toward sensing targets. To tackle this intractable problem, we propose an alternating optimization algorithm by leveraging fractional programming, matrix lifting, and semidefinite relaxation techniques. Simulation results demonstrate that our proposed algorithm outperforms three baseline schemes in both optimizing the number of active RF chains and maximizing the energy efficiency.