An Initial Phase-Added and RIS-Assisted Physical Layer Security Scheme Based on Deep Reinforcement Learning

The usage of physical layer security (PLS) in the Internet of Vehicles (IoV) has become widespread. With the majority of vehicles requiring meticulous energy management, the enhancement of energy efficiency in the realm of PLS strategies has garnered significant attention from researchers. The conventional approach concentrates on allocating energy between confidential information and artificial noise (AN), leading to additional energy consumption for AN. In contrast, this study introduces a novel initial phase-added and reconfigurable intelligent surface (RIS)-assisted PLS (IPARISS) approach that optimizes the transmission phase to achieve security in a more energy-efficient manner. A phase shift is introduced at the legitimate transmitter to increase the difference between the signal received at the legitimate receiver and the eavesdropper. Furthermore, RIS is employed to enhance interference at the eavesdropper. To increase the bit error rate (BER) at the eavesdropper with low energy cost, the phases, RIS position, and power allocation strategy are jointly optimized using a deep reinforcement learning (DRL) framework, specifically the constraint-optimized proximal policy optimization (COPPO)-IPARISS algorithm. Simulation results indicate that the proposed phase optimization scheme improves security performance while reducing the power consumption of the transmitter under different modulation techniques. The impact of the RIS height on energy efficiency under quadrature phase shift keying (QPSK) modulation is analyzed. Finally, the robustness of proposed approach is verified through simulating the effect of different speeds of vehicles.