Toward Secure and Scalable Vehicular Edge Computing With Zero-Energy RIS Using DRL

Abstract

The escalating demands of advanced automotive technologies exert considerable pressure on modern vehicles’ computational capabilities. This scenario underscores the critical need for vehicular edge computing (VEC) networks, which leverage 5G/6G communications to facilitate computational offloading. However, providing seamless access to these services while simultaneously adhering to stringent latency and security requirements presents a formidable challenge. The advent of reconfigurable intelligent surfaces (RIS) heralds a new era of possibilities, which includes enhancing connectivity, boosting data transmission rates, consequently reducing delay, and improving the physical layer security of communication channels. This paper delves into the utilization of zero-energy RIS (ze-RIS) in the context of vehicular computation offloading. Our primary goal is to ensure secure access while optimizing operational efficiency in compliance with various task-related and environmental requirements. The ze-RIS-assisted secure task efficient offloading (DRSTO) scheme is a novel deep reinforcement learning (DRL) framework that cleverly switches communication connections to optimize task offloading efficiency and security thereby resolving this issue. At its core, our assessment strategy revolves around the DRSTO model’s secrecy and efficiency factor which serve as both a performance measure and a reward function. Time efficiency and rate of confidentiality are used to evaluate this aspect which provides a thorough evaluation of the scheme’s success. Extensive testing and comparison have shown that the DRSTO scheme’s efficiency factor can be significantly increased, from 6.05 to 18.10. In addition, the rate of job success has increased dramatically, from 2.12% to 4.63%. When compared to other models that were evaluated, the DRSTO scheme consistently better on several metrics including reward, time frames per step (TFPS) ratio and DRL characteristics.

Keywords

• Vehicular communication
• security
• intelligent surfaces
• edge computing
• DRL