RIS-aided integrated sensing and communication: Beamforming design and antenna selection

This work considers an integrated sensing and communication system, where a reconfigurable intelligent surface (RIS) is utilized to manage interference and radar signals. The sensors are attached to the RIS to sense multiple targets. A joint design of the base station transmit beamforming and RIS phase shift matrix is proposed to minimize total interference and maximize the worst received signal power at the RIS sensors. Due to highly coupled transmit beamforming and RIS phase matrices, the optimization problem is decoupled into two subproblems and solved iteratively by semidefinite programming and a manifold-based Riemannian steepest descent algorithm. We further design energy-aware beamforming to eliminate the interference induced by radar probing signals. Antenna selection with the ${\ell }_0$ norm is introduced to exclude redundant antennas while maintaining the sufficient multiple beams for multiple users and targets with minimized required antennas. Due to the nonconvexity of the ${\ell }_0$-norm, we relax the number of active transmit antennas as a weighted ${\ell }_1$-norm and employ a concave approximation for the constraint on the radar beampattern. Numerical results illustrate that the proposed algorithms can effectively reduce interference and strengthen the received signal power for radar sensing, achieving mutual benefit for communication and sensing performance.