Research On Edge Server Placement In Mobile Edge Computing

With the rapid development of e-commerce,Augmented/Virtual Reality(AR/VR)and other emerging mobile applications over the years,it has promoted the growth of the computing demands of mobile users.In the traditional mobile cloud computing architecture,mobile users can offload their tasks to remote cloud servers through base stations,who can also use these cloud servers to perform tasks on their behalf.However,since the high service delay for mobile users is inevitably caused by this model,the high-quality user experience cannot be provided nowadays.Over the past few years,mobile edge computing has gradually become a popular research direction,in which the extensive attention has been paid to the placement of edge servers.The location of edge servers has caused a significant impact on optimizing server energy consumption,deployment cost,reduction of access delay and balance of workload,so the great significance is embodied in the research on edge servers.In this thesis,the heterogeneous edge server placement methods in different scenarios were studied.According to the placement position of edge servers in the network structure,the placement scenarios are divided into mobile cellular network and packet core network.Firstly,the placement scheme of edge servers was researched in mobile cellular network.The problem was defined as the multi-objective optimization problem.In the meantime,an Ant Colony Algorithm based Edge Server Placement algorithm(ACO-ESP)was proposed by considering the energy consumption and load balancing of edge servers.Through the simulation experiments,it indicates that the performance of ACO-ESP algorithm in reducing server power consumption and balancing load is improved in contrast with other optimization algorithms.Then the edge server placement scheme in the packet core network scenario was studied and the deployment cost ratio of the placement problem was considered from the perspective of service providers.A Genetic Algorithm and Simulated Annealing algorithm based Edge Server Placement algorithm(GASA-ESP)was proposed and a simulation experiment was designed to compare and analyze the optimization algorithms proposed by other studies.By using the data set of a telecom user,the algorithm proposed in this thesis was evaluated.From the results,it indicates that the proposed GASA-ESP algorithm in this thesis has great advantages in reducing response time and balancing workload of servers placed.Finally,an edge server placement management system was designed and implemented in this thesis,in which the user login registration module,user management module,device management module,and an edge server placement module based on the Edge Cloud Sim simulation platform were realized.In accordance with the final system test results,it indicates that the management system is able to complete the task of placing edge servers in different scenarios,which can also realize the efficient management of equipment such as base stations and edge servers at the same time.