Research On Task Offloading Mechanism Of Internet Of Vehicles Based On Mobile Edge Computing

With the development of communications technologies and computer applications,Internet of Vehicles(IoV)has emerged.In the Internet of Vehicles,there are huge task processing requirements,and computation offloading technology has been introduced,which improves task processing power and efficiency.Most researches in academia nowadays are focused on designing reasonable offloading strategy to achieve the purpose of offloading,such as reducing energy consumption or lowering task execution costs.However,these strategies often ignore the internal characteristics of the task.Moreover,in research,the resource attribute of mobile edge computing(MEC)server is single,and it lacks consideration of task backlog.These unreasonable offloading strategies may lead to poor performance.Therefore,the paper studies the offloading strategy of the vehicle task represented by the Directed Acyclic Graph(DAG)structure and the offloading strategy that takes into account the backlog of task when MEC server resource attributes are diverse.First,in view of complex vehicular tasks that can be represented based on the DAG structure,the paper proposes a computation offloading strategy that optimizes energy consumption for such vehicular tasks under time constraints.According to the time dependency of the subtasks of the DAG structure,and the limitation that some subtasks can only be calculated locally,the energy consumption model for executing the task is established,and the simulated annealing and tabu search hybrid algorithm are used to solve the problem of minimizing the energy consumption of the DAG structure task computation.The simulation results show that compared to single algorithm,the hybrid algorithm has faster optimization speed and better performance.At the same time,the final result greatly reduces the energy consumption of task execution.Second,in view of the impact of task backlog on system stability,the paper proposes a multi-task computing offloading strategy for queue management and computing cost optimization.When the resource attributes of the MEC server are diverse,a calculation and queue backlog model is established to analyze the cost of vehicles,and finally the average cost is optimized under the stability of the system.In order to avoid dependence on future information as much as possible,this paper adopts the Lyapunov optimization method for transformation,and designs the offloading strategy of each time slot to achieve the overall system optimization.For solving the problem of offloading in the time slot,a chemical reaction optimization algorithm(CRO)is used.The simulation results show that compared with the general heuristic algorithm,the CRO algorithm can reduce the execution cost when the system is stable.The Lyapunov optimization method helps to maintain the stability of the system,and can also minimize the cost under different system stability requirements by adjusting the penalty factor of the method.