Joint optimization of energy and delay for computation offloading in vehicular edge computing

With the advancement of information technology, Mobile Edge Computing (MEC) has been integrated into Vehicular Ad Hoc Networks (VANETs) to efficiently address the resource constraints of vehicle terminals. Within the realm of MEC, the effective utilization of wireless channels and edge server resources for task offloading to meet users’ Quality of Service (QoS) requirements has become a significant research area. This paper presents a queuing model and proposes a Predictive Vehicle-to-Vehicle Task Offloading (PV2VO) strategy based on predictable MEC state. The PV2VO strategy employs Vehicle-to-Vehicle (V2V) communication for task data upload and computation result download, thus reducing result return delay. Selective offloading is performed based on the edge server’s load status and waiting time, resulting in reduced waiting time for the edge server’s computing tasks and meeting users’ QoS requirements under high load conditions. By considering both the delay model and energy model, two preference factors are defined, and an objective function for local execution and offloading to the edge server is formulated. The genetic algorithm is then employed to determine the optimal local computing frequency, optimal transmission power, and the most suitable computation offloading approach. The proposed PV2VO strategy is simulated and compared with conventional strategies through experiments involving five different types of typical computing tasks. The comparison of the results validates the superiority of the PV2VO strategy.