| With the rapid development of Internet of vehicles technology in recent years,it drives the development of intelligent driving,intelligent navigation and other technologies.However,these new intelligent applications are mainly computing-intensive services,which not only require a large amount of computation,but also have strict requirements on delay.The traditional cloud computing solutions are limited by the centralized network architecture and thus cause a high delay,which cannot effectively support those computing-intensive services.In order to solve the above problems,the industry proposes a solution for the Internet of vehicles based on fog computing,which can reduce the network delay by lowering the edge nodes with computing capability closer to the user.To analyze and verify the performance of vehicle network based on fog computing from the system point of view,this paper builds a system level simulation platform,and carries out the relevant simulation evaluation work.The main work and contributions are summarized as follows:1.From the point of view of the system,this paper designs and builds a system level simulation platform of the Internet of vehicles according to the parameters of 3 GPP protocol,such as path loss,shadow fading,fast fading,etc.On this basis,vehicle access based on load balancing and switching strategies,calculating service delays in different modes and the function of distributed fog computing module are realized2.Based on the established system-level simulation platform,the time delay of the vehicle network under the three computing modes of local computing,cloud computing and fog computing is evaluated.First,considering that there are great differences in packet sizes among different services in the actual scenario,this paper simulates the impact of different packet sizes on the delay performance under three calculation modes.When the packet size is small,the performance of local computing is better than other computing modes.As the package size increases from 5 MB to 20 MB,the delay performance of the fog computing mode is improved by about 17%compared with the other two computing modes;Secondly,because the vehicle speed will have a great impact on the vehicle distribution in the network,this paper simulates the impact of different vehicle speeds on the delay performance under three computing modes.The simulation results show that as the vehicle speed increases from 70 km/h to 140 km/h,the performance of local computing delay have barely changed,and the latency of fog computing and cloud computing modes gradually decreases.The delay performance of fog computing mode is about 10%higher than that of other computing modes.Finally,this paper simulates the delay performance of three computing modes under different CPU processing capabilities.The simulation results show that with the improvement of CPU computing capabilities,the delay performance of fog computing mode will gradually surpass the other two modes,and the delay performance of fog computing mode is about 15%higher than that of other computing modes.3.In order to effectively reduce the delay in the fog computing based vehicle network,this paper proposes a joint optimization method of computing resource allocation and data offloading decision,which aims at minimizing the delay.In view of the high complexity of joint optimization problem,this paper utilizes deep reinforcement learning to efficiently solve the original mixed integer programming problem,and the performance of the proposed algorithm is verified by simulation.Simulation results show that the proposed optimization algorithm improves the delay performance by about 8%compared with the algorithm based on the traditional offloading method.This paper studies the system level platform of the Internet of vehicles based on fog computing.The research results not only lay a simulation foundation for the architecture of the Internet of vehicles based on fog computing,but also provide performance reference for the deployment and application of fog nodes. |
