Intelligent Reflecting Surface-Assisted Power Efficiency Physical Layer Security in Uplink Multiple-User Channel

This article investigates the radio frequency (RF) power efficiency issue of intelligent reflecting surface (IRS)-assisted physical layer security (PLS) in multiple-user uplink channels. Traditional methodologies relying on secrecy rate metrics prove impractical when faced with unknown eavesdropper channel state information (CSI). To tackle this issue, the concept of secrecy outage probability is introduced within IRS-assisted PLS. This paper formulates the total RF power minimization problem of multiple users in the uplink channel, considering scenarios where the eavesdropper's CSI is unknown. An alternating optimization scheme is proposed to address the problem via alternately optimizing the RF power of individual users, receiving beamforming vectors, and phase shift matrix, minimizing the total RF power while ensuring compliance with secrecy outage probability constraints. Innovative methodologies, such as manifold optimization, dynamic weighted quadratic transformation, and deep learning, are proposed to approximate optimal solutions and reduce complexity in phase shift optimization. Simulation results indicate that the proposed scheme markedly decreases RF power, with the DL-based approach demonstrating satisfactory performance while substantially reducing time complexity.