Research On Resource Management And Task Offloading In Vehicle Cloud-fog System

Vehicular ad hoc networks are an indispensable part of intelligent transportation.How to efficiently manage the massive resources in the vehicular ad hoc networks,how to solve the problem of real-time transmission of data information in the vehicular ad hoc networks,have been a research hotspot.Cloud computing is the big data processing center of the vehicular ad hoc networks,because of its distance limit,cannot satisfy the needs of location aware,low latency and mobility support,this paper uses the combination of cloud and fog computing technology,and puts forward new architecture of vehicle cloud-fog system,design a "central cloud-road side fog-vehicle cloud" three-level system architecture which makes full use of the resources of the whole system.And systematic research on the architecture of the vehicle cloud-fog system,and systematically researches the resource management and task offloading of the vehicle cloud-fog system.In the new architecture of vehicle cloud-fog system,based on the characteristics of real-time request service of vehicle users,aiming at the problem that most of the dynamic resource optimization methods in vehicle networking do not fully consider the maximum long-term benefits of the system,this paper uses SMDP to modeling the service request process of the system,Utilize the state space,action function,reward function and State transition probability of the vehicle Cloud-Fog System,and puts forward a resource management scheme of cloud-fog computing collaboration based on SMDP,gets Optimal decision strategy based on value iterative algorithm,and compared with the resource allocation scheme of GA algorithm and greedy algorithm.At the end of this paper,the simulation results show that compared with the traditional greedy algorithm and GA algorithm allocation scheme in Blocking rate,the proposed scheme is about 37% lower than that of greedy algorithm,which is about 45.2% lower than that of GA algorithm,thus improving the service quality experience of users.Compared with long-term benefits of the system,the proposed scheme is about 10% higher than that of greedy algorithm,which is about 16.7% higher than that of GA algorithm,which makes the maximum long-term benefits of the system.On the architecture of the vehicle cloud-fog system,this paper proposes a computational task offloading scheme for cloud-fog network collaboration.Before the task is unloaded,the improved DK-means algorithm is used to cluster the task according to the delay sensitivity and calculation two indexes,the simulation results show the clustering effect is better than FCM algorithm and K-means algorithm,thus reducing the waiting delay and transmission delay of the task offloading.For the cloud-fog network collaboration task offloading scheme,taking into account multiple factors such as transmission power,task data size and computing power,the normalized value of delay and energy consumption is used as the trade-off factor of the evaluation scheme.The optimization problem is modeled to minimize the energy consumption and delay problem of the whole vehicle-linked cloud fog network.Due to the problem of NP-Hard problem,we propose an improved particle swarm optimization algorithm(DEPSO)to solve the optimal task offload strategy.Finally,the simulation results verify the convergence of the proposed algorithm,reduce the search time and improve the efficiency,and the cloud-fog collaboration scheme based on DEPSO can effectively reduce the energy consumption of the system,reduce the total time of task processing,and improve the quality of service.