Hybrid V2V/V2I Task Offloading in Vehicular Edge Computing: A Double-Layer Stackelberg Game Approach
Shujuan Wang, Hao Peng, Huafeng Li · IEEE Internet of Things Journal · 2025
Task offloading is a promising way to support computation-intensive applications in a resource limited network such as the Internet of Vehicles (IoVs). However, great challenges exist in applying task offloading efficiently in practical IoV scenarios. For one thing, vehicles and RoadSide Units (RSUs) are reluctant to participate in the cooperation due to the lack of incentive. For another, multiple vehicles generate various computation tasks at the same time, which further intensifies the resource competition among vehicles. Although existing work have explored the potentiality of Stackelberg game theory in stimulating the cooperation between vehicles/servers, etc., from various aspects, the classical single-layer modelling is over simplified and fails to make the best use of all available resources. To solve these issues, this paper focuses on simultaneously incentivizing vehicles and RSUs as computation assistants, to provide their resources in completing the task together. A novel double-layer Stackelberg game-based approach is proposed to perfectly characterize the collaboration and competition among vehicles/RSUs in the hybrid V2V/V2I offloading scenario. The interactions between multiple user vehicles and a RSU are formulated as the first-layer Stackelberg game, and that between a user vehicle and multiple service vehicles are defined as the second-layer Stackelberg game. The existence of Nash Equilibrium in each Stackelberg game is proved and two algorithms are designed to approximate the achievable optimal offloading strategy under practical vehicular environment. Simulation results prove that the proposed approach achieves remarkable performance advantages in terms of completion delay, utilities of vehicles, RSUs and the overall system.