Adaptive Task Offloading Strategy in Vehicle-Assisted Mobile Edge Computing

Traditional Mobile Edge Computing (MEC) is over whelmed by time-sensitive applications in the Internet of Vehicles (IoV), leading to significant task completion delays because existing methods fail to leverage vehicle-to-infrastructure collaboration in dynamic network topologies. This paper proposes a vehicle-assisted adaptive task offloading strategy to minimize completion time through a dual-mode framework that adapts based on a vehicle's position relative to an edge server. When a vehicle is outside a server's range, the Best Service Vehicle Selection Algorithm (BSVSA) offloads tasks to the most suitable nearby vehicle while ensuring communication stability. When within server coverage, our novel Hybrid Differential Teaching Optimization Algorithm (HDTOA) determines the optimal offloading ratio and schedules tasks across edge servers to balance the computational load. Simulation results validate that our integrated approach (HDTOA+BSVSA) outperforms benchmarks like Differential Evolution (DE) and Particle Swarm Optimization (PSO), demonstrating faster convergence and lower average task execution times under heavy load. Under scenarios with a large task data size, the HDTOA reduces the average task execution time by 69.99% compared to the PSO algorithm. The strategy also provides a more balanced workload across servers, thus enhancing overall system efficiency