Power Efficiency Physical Layer Security for Multiple Users in IRS-Assisted Uplink Channels: Learning to Phase Shift

• Published in: GLOBECOM 2023 - 2023 IEEE Global Communications Conference pp. 5037 - 5042
• Main Authors: Cheng, Xiangrui, Liu, Yiliang, Su, Zhou, Luo, Xuewen, Xu, Qichao, Peng, Haixia, Benslimane, Abderrahim
• Format: Conference Proceeding
• Language: English
• Published: IEEE 04.12.2023
• Subjects: Complexity theory; Optimization; Physical layer security; Power demand; Radio frequency; Simulation; Uplink
• ISSN: 2576-6813
• DOI: 10.1109/GLOBECOM54140.2023.10437473

This paper investigates the power efficiency of physical layer security (PLS) in intelligent reflecting surface (IRS)-assisted multi-user uplink channels. Existing research works usually focus on enhancing secrecy performance, and neglect measures to improve power efficiency. In this paper, the optimization problem is formulated to minimize the sum radio frequency (RF) power of multiple users in the uplink channel subject to secrecy outage probability constraint. This problem is solved by an alternating optimization (AO) algorithm that includes three optimization sub-problems, i.e., phase shift matrix, receiving matrix, and RF power optimization. Furthermore, to reduce the complexity of the proposed AO algorithm, a deep learning (DL)-based approach is proposed to optimize the sophisticated phase shift matrix optimization process. Simulation results demonstrate that the proposed scheme can significantly reduce the average RF power, and the DL-based scheme achieves similar performance as AO algorithm while reducing the time complexity significantly.