Joint Design of Beamforming and Artificial Noise for Dual Hybrid RIS-Assisted D2D Secure Communications

Abstract Hybrid reconfigurable intelligent surfaces (HRISs) alleviate cascaded attenuation by selectively activating reflecting elements. To improve secure multi-pair device-to-device (D2D) communications, this paper considers two cooperative HRISs and artificial noise (AN). The transmit beamforming and AN vectors, HRIS reflection coefficients, and binary active/passive configurations are jointly designed to minimize the modeled total system power subject to secrecy-rate and per-HRIS amplification-power constraints. The resulting mixed-integer nonconvex problem is addressed by a four-block alternating-optimization framework employing semidefinite relaxation, successive convex approximation, majorization–minimization, and a discrete mode-selection surrogate. The numerical comparisons in the revised manuscript illustrate the power-consumption trends of the dual-HRIS architecture, the role of adaptive element allocation, and the effect of AN for the studied nominal-CSI scenarios. A mathematical multiple-eavesdropper extension, an Eve-CSI-error sensitivity study using an explicit design margin, and a computational-complexity discussion are also provided. The CSI-error sensitivity study should not be interpreted as a certified worst-case robust-optimization solution.