Optimizing Vehicular Edge Computing: Graph-Based Double-DQN Approaches for Intelligent Task Offloading

In vehicular edge computing, optimizing task offloading is crucial to balance computational needs, reduce delays, and cut costs amid the dynamic challenges of the vehicular environment, including vehicle mobility, network topology, and computing resource variability. This study introduces a task offloading scheme enabling vehicles to dynamically choose local execution or offloading to nearby vehicles, edge servers, or the cloud. The primary goal is to optimize task offloading by simultaneously minimizing cost and delay. Achieving this involves integrating graph convolutional networks and deep reinforcement learning, enhancing decision-making efficiency and network representation for agents. The fusion of Graph Convolutional Networks with Double-DQN strengthens overall network representation and decision-making. The optimization challenge is formally structured as a Markov Decision Process. Simulation results highlight the proposed scheme's superiority, showcasing its effectiveness in achieving cost-efficiency by maximizing resource utilization, minimizing costs, and optimizing task offloading while reducing task rejection.