Research On Fog Computing Offloading Strategy For Internet Of Vehicles

As the demand for computationally intensive autonomous driving applications continues to grow,and vehicles with high mobility need to ensure driving safety,the Internet of Vehicles(IoV)communication is facing major challenges with strict end-to-end delay limits and reliability requirements.Fog computing is a network architecture that deploys distributed computing,communication,and control functions close to the end user.If a vehicle with idle computing processing capacity is used as a fog node to provide communication and computing services,such a fog computing architecture is called vehicle fog computing(VFC).Vehicle fog computing is a promising method with great potential to improve vehicle service by offloading computationally intensive tasks to fog nodes.Existing research on fog computing in IoV scenario is mainly focused on roadside unit(RSU)servers and resource scheduling.Few studies have focused on vehicle mobility,especially offloading decisions and resource allocation in the VFC scenario,but in practice the offloading decision is very important in IoV.This thesis focuses on the computation offloading problem in the VFC scenario,the purpose is to provide stable connections and reduce the tasks completion delay.The innovations and specific work are summarized as follows:(1)This thesis introduces the system efficiency utility function related to the connection duration and task completion time,as a basis for matching strategy execution and an index of system performance evaluation.Firstly,the mutual selection strategy between the vehicle and the vehicle fog node(VFN)is studied based on the matching theory.Vehicle matching expand KM algorithm(VMEKM)is applied according to the scene to complete the matching and maximize the total system efficiency.Then,based on the matching results,the communication resource allocation problem is defined and transformed.The channel allocation genetic algorithm(CAGA)is introduced to solve the problem by introducing a genetic algorithm.Simulation results show that the two algorithms can play a role in improving system efficiency and reducing system delay,respectively.Through the continuous process of the two algorithms,the allocation of public resource is completely completed,and the stability of the connections and task completion delay in the system are optimized.(2)This thesis studies the task offloading strategy for a single task when all connections in a single VFN are determined.A new task processing mechanism is designed for specific task types,and a heuristic joint offloading policy and resource allocation algorithm(JOPRA)is designed based on the mechanism to optimize the total system delay.The algorithm is combined with the designed task processing mechanism.By optimizing the proportion of offloading,the proportion of sub-task division,and the allocation of computing resource,the total system delay is reduced from the way of each task itself and within the VFN.Simulation results show that the algorithm can significantly reduce the total task delay in the system by making full use of communication and computing resource.