DRL-Based Task Offloading and Resource Allocation Strategy for Secure V2X Networking

抄録

Vehicle-to-Everything (V2X) networking is poised to further integrate vehicular communication with Mobile Edge Computing (MEC) technologies, enhancing autonomous vehicle driving and intelligent applications. However, the rapid growth in computational demands and vehicle mobility often results in inefficient task processing, particularly in resource-constrained environments. Additionally, the broadcasting nature of wireless channels poses significant risks to user privacy and network confidentiality. This paper addresses the challenges of task offloading and resource allocation in secure V2X networking, to improve the computational efficiency of tasks. To achieve this, we propose a Deep Reinforcement Learning (DRL)-based algorithm to obtain the optimal task offloading and resource allocation strategy by employing the Actor-Critic (AC) framework. In this algorithm, a Convolutional Neural Network (CNN) is employed as the actor network to output offloading decisions, efficiently extracting features from high-dimensional state spaces. The critic network evaluates the offloading decisions using a subcarrier allocation strategy based on a greedy algorithm and a power allocation strategy derived from an improved water-filling (IWF) algorithm. The proposed power allocation strategy offers a closed-form solution for subcarrier transmit power. Simulation results demonstrate that the proposed approach significantly enhances the computational efficiency of task processing while ensuring secure communication.