Potential Game for Computation Offloading in Edge-Vehicle Collaboration Computing
Xi Liu, Jun Liu, Weidong Li · IEEE Transactions on Consumer Electronics · 2025
We address the problem of computing offloading in the edge-vehicle collaboration computing system, where edge servers (ESs) and smart vehicles (SVs) provide computing and sensing services to users. A sensing resource-sharing model is proposed, where the sensing devices can serve multiple tasks simultaneously. We consider a computing environment with multiple rational users, where each user has three options: local computing, ES computing, and SV computing. The computing offloading problem is formulated as an ordinal potential game to realize distributed offloading decisions by individual users. We show that the proposed computing offloading game achieves a Nash equilibrium in which no user has incentives to change their decisions. A novel distributed computation offloading algorithm is proposed, which enables users to participate in the offloading game in a distributed manner. We then analyze the convergence of the proposed algorithm and show that it admits the finite improvement property. Furthermore, the approximation and the price of anarchy of the proposed allocation algorithm are analyzed. The results show that the proposed algorithm can converge to a Nash equilibrium in a fast manner. Additionally, the average percent gap between solutions obtained by the proposed algorithm and optimal solutions is 3.21%.