Coalition Formation Game Based Information-Energy Collaboration in Vehicle Edge Computing Networks

Abstract

In vehicle edge computing network (VECN), a key issue for the roadside unit (RSU) is how to handle computationally intensive tasks timely and use energy efficiently. When an RSU cannot finish the task on time, computation offloading will be an effective approach. However, when offloading tasks most RSUs neglect the energy supply and do not consider computation requirement and energy supply jointly. In this paper, we consider a VECN where there are multiple RSUs. When an RSU has a computational task with latency constraint, it will offload the task to the nearby RSUs and/or cloud servers, and purchase electricity from the neighbors and/or power grid if it is also short of power. To conserve energy and reduce cost, not only the collaboration between the VECN and the cloud/grid, but also the collaboration of RSUs within the VECN is investigated. Specifically, we first formulate the information-energy collaboration problem, and study how the strategies of computation offloading and energy sharing interact with each other, where the RSU's individual revenue is considered. Next, since the problem is a mixed nonlinear programming problem that is hard to be solved, we reformulate it into a double-level coalition formation game, based on which an algorithm with low complexity is proposed. Finally, we prove theoretically that through finite iterations, the game-based algorithm can converge to a stable Nash equilibrium which is the solution to the original problem. Simulation results show that our scheme is superior to the existing ones in reducing edge network expenses while improving individual revenue.