A Method for Mobile-Aware Multi-Vehicular Offloading in Hybrid Cloud-Edge Environment

Yu Wang, Feng Zeng, Yin Li, Yunni Xia, Linchengxi Zeng

Abstract: Recently, the Vehicle Edge Computing (VEC) paradigm offers an emerging idea to improve the problem of insufficient computational resources in motion. When the volume of tasks increases and leads to a sudden increase in computation, it is clear that pure edge resources are not sufficient for the computationally intensive applications proposed by multiple vehicles. This calls for such a hybrid environment, so it is necessary to offload computationally intensive and latency-sensitive tasks to nearby edge nodes, and establishing a hybrid architecture with centralized cloud services and multiple edge nodes and an intelligent resource management strategy can be a good solution to the above problem. However, the distributed nature and the inherently dynamic nature of the environment often present in real deployments are factors that make this problem challenging. Current existing research mainly assumes that the performance of edge servers is constant and treats offloading decisions as a static optimization problem. In this paper, we additionally consider geographically distributed edge servers with time-varying performance, while introducing a dynamic offloading strategy based on a probabilistic evolutionary game theory framework. Simulation outcomes prove that our framework beat existing methods in several metrics.