Resource Placement For Mobile Edge Computing In Internet Of Vehicles

The proposal of Mobile Edge Computing(MEC)provides computing,storage and communication resources for the edge network,to meet the resource requirements of the expanding Internet application scenarios and increasingly diversified application services for the mobile network.As one of the important application scenarios of 5G,the Internet of Vehicles(IoVs)covers a variety of application services with strict delay and reliability requirements.As one of the critical technologies in this environment,MEC can support the demand of resources from dense access vehicle applications,and provide users with low delay,high reliable real-time services,and thus providing feasibility for intelligent driving.In MEC,resource placement will directly affect the quality of service and resource deployment cost.The researches on MEC resource placement are mainly conducted from two aspects:MEC servers(i.e.,hardware)placement and service instances(i.e.,software)placement.But the existing researches still have some defects and deficiencies under the environment of IoVs.Vehicle applications have their unique requests distribution and the demand for quality of service and resources.With fully considering the unique characteristics of vehicle applications,the placement scheme of MEC servers and service instances can provide high quality of services for vehicles.Therefore,this thesis focuses MEC server placement and service instance placement in the environment of Internet of vehicles.The major contributions of this thesis include:(1)An MEC server placement scheme with service area partition is proposed,aiming to achieve low task response time and cut down the energy waste caused by low workload status of MEC servers.In the proposed scheme,MEC servers are placed in the Road Side Units(RSUs)deployed at the intersections.The MEC server placement problem is modeled into a multi-objective optimization problem,and then is transformed to a single-object optimization problem.The proposed scheme includes two steps:candidate RSUs selection and MEC server placement location and service area determination.The algorithms are proposed to solve the problem.Simulation results show that the proposed MEC server placement scheme has good convergence and has better performance in average task response time and the ratio of server base energy consumption compared with existing schemes.(2)A service placement scheme is proposed to minimize the number of deployment instances and the times of service migration.Considering the different resource requirements of different applications,a two-stage service instance placement algorithm based on heuristic algorithm is proposed.Simulation results show that the proposed service deployment scheme has better performance on the number of services’ instances and service switching times of the system compared with other schemes.