Research On User Handoff Management In Sliced 5G Architecture

With the rapid development of mobile Internet technology,the number of end user has dramatically increased,and the types of services have become more diversified.Traditional communication networks are limited by physical resources and architecture and cannot meet the ever-increasing demands of users in the aspects of system capacity,data transmission rate,resource utilization,etc.However,the next-generation wireless communication system(5G)enables physical resource virtualization and network architecture flexibility by introducing SND and NFV.The system can configure,orchestrate,and reorganize these virtual resources to form different types of network slices(Network Slice,NS)according to the user service type.Different types of NSs can provide different service services,and users can choose to access NS according to the type of service.Although the sliced network architecture greatly improves the resource utilization and network flexibility,there are still some problems in user mobility management.This thesis focuses on the problem of NS re-access and virtual resource scheduling caused by handoff in sliced 5G architecture.Accordingly,the main research of this thesis is as follows:(1)Duo to the existing knowledge of 5G network is multifarious,after reading and combing a large number of domestic and foreign related literatures,the difference between 5G network and traditional network is analyzed,and the basic concepts,architecture and main application of SDN,NFV,and NS in 5G are discussed in detail to provide the basic theoretic foundation for subsequent research.(2)As the handoff process is short,temporarily forming NSs may cause a delay in the handoff response.Therefore,a throughput maximization-based user access mechanism was proposed without changing the existing network slice deployment.Meanwhile,a practical implementation scheme of the mechanism is put forward.Andusing greedy algorithm,particle swarm optimization and genetic algorithm to solve the proposed mechanism respectively and analyze comparatively,and the results obtained by using the particle swarm algorithm are superior to the other two methods.In addition,the proposed mechanism is compared with power-based access,load-based access and random access,the simulation results show that the handoff success ratio and system throughput of the access mechanism are significantly better than the other three schemes.(3)As the diversity of service types of handoff users,the existing slice types in the system may not be able to meet the user service requirements.If the resources are temporarily scheduled according to service requirements to re-form NSs,the continuity of the user communication may be affected by the delay in the response to the user's handoff request.Therefore,a slice resource scheduling scheme based on mobile prediction is proposed.The core idea of the scheme is to reserve resources for each base station in advance according to the mobile prediction result to reduce the delay caused by NSs' reconstruction.Firstly,Markov and online learning process are combined to predict user mobility.Then,a network slice resource scheduling model is established based on the prediction results and a utilization maximization-based resource scheduling scheme is proposed.Finally,this scheduling scheme is compared with load-based scheduling and random scheduling,and the result shows that the scheduling scheme can guarantee high resource utilization.We summarize the work of this thesis and look forward to the further research direction.