Research On User Access Policy In Heterogeneous Vehicular Networks Based On MEC

Multi-access Edge Computing(MEC)is a promising technology goes with 5G.By putting resources next to the edge of the network and users,MEC greatly reduces latency and relieve the computing pressure on terminal devices.Nowadays,it has been widely used in various fields.With the deployment of MEC server applications,more and more attention has been paid to the problems and challenges it faces in the field of Internet of Vehicles,such as resource allocation,user association and mobile management issues.Unmanned Aerial Vehicle(UAV)greatly expands the coverage of edge computing for its flexibility and low cost,which has attracted lots of attention in the application of MEC.However,in previous studies,little consideration has been given to the application of UAV in the vehicular network,which can realize flexible data transmission.In the same time,network node heterogeneity has become an evitable trend because of current complex vehicular network.Therefore,how to use MEC technology to transmit video contents,improve the quality of experience(QoE)of vehicular users(VUEs)and realize the dynamic association between VUE and network nodes in time-varying network is still a hotspot and hard issue in the field of vehicular network.Focusing on the scenario of UAV-assisted vehicular network,this thesis studies the long-term association policy between VUE and resource nodes with limited computing,caching,communication and energy resources.The main research contents are concluded as follows.(1)Research on single user association policy in heterogeneous UAV-assisted vehicular network based on MEC.We proposed a MEC-based user association scheme by putting caching and computing resources on the edge nodes of the network such as road side unit(RSU),UAV,and macro base station(MBS).Based on the time-varying nature of the link and the heterogeneity of resource nodes in communication,computing and caching,the availability of resource nodes and the QoE users attain through links are described with two indicators:video transmission capacity and reliability.With the goal of maximizing user QoE,this thesis establishes a VUE association model during multiple time slots under some resource constraints.Considering that the VUE association problem during multiple time slots is an NP-hard problem,we transform it into a shortest path problem and find a sub-optimal solution.Simulation results show that our model and algorithm can effectively improve user QoE compared to other correlation mechanisms.(2)Research on multiple user association policy in heterogeneous UAV-assisted vehicular network based on MEC.Considering user and environment states and the time-varying nature of resources in vehicular networks,MDP is used to model the transient environment,which includes changes in the number of connected nodes caused by user requests for access,the change of the link distance caused by vehicles’ movements and the change of the channel condition caused by small-scale fading.Secondly,the utility of the link between the VUE and the resource node is descried based on the heterogeneity of nodes.Finally,we use the dynamic programming algorithm to optimize the user association policy by considering the next state transition.At the same time,the effectiveness of our model and proposed algorithm can be proved by numerical simulations and corresponding comparison algorithms.