RIS Aid Security Strategy of Multi-Objective in V2X based on Deep Reinforcement Learning

With the rapid development of vehicle-connected (V2X) technology, communication security is faced with severe challenges of eavesdropping and jamming. The existing schemes are often difficult to meet the needs of dynamic environments because of the single objective, poor adaptability of scenarios, and low utilization rate of resources. Therefore, this paper proposes a security strategy based on dynamic multi-objective deep reinforcement learning (DMDRL), combined with reconfigurable intelligent surface (RIS) technology, to optimize the secure communication performance of vehicle networks. Firstly, a multi-objective reward function including safety capacity, delay, energy consumption, and coverage efficiency is constructed, and a dynamic weight mechanism is designed to adapt to different scenarios. Secondly, the multi-objective proximal policy optimization (MO-PPO) algorithm trains the RIS controller to balance the conflict between strategy stability and multi-objective. In addition, the edge cooperative fine-tuning mechanism is introduced to achieve low-cost strategy adaptation through knowledge distillation and federation aggregation. Simulation results show that compared with the traditional method, the proposed scheme can increase the safety capacity by 15%, reduce energy consumption by 12%, and significantly enhance the robustness in complex scenarios.