Research On Orchestration Algorithms In 5G Network Slice

Network slice virtual resource orchestration technology is an emerging concept of 5G mobile network,including network slice resource mapping and resource allocation.It is a vital part of network segmentation commercialization.Traditional resource mapping and resource allocation algorithms are mostly based on single targets such as system throughput or resource utilization.The network slicing algorithm proposes higher requirements based on resource mapping and resource allocation algorithms under traditional network architecture.Based on the three-layer architecture of network slicing,this thesis studies the network slicing strategy from the perspective of users and network operators.First,considering different users' requirements for system bandwidth and delay,and analyzing the enhanced mobile broadband(eMBB)service and ultra-reliable & low-latency(uRLLC)service,the joint bandwidth resource orchestration model(e&u-BOP)is established to reduce the maximum transmission delay of the uRLLC service and improve the minimum throughput of eMBB services.The model is both a mixed integer programming model(MIP)and an NP-hard problem.For e&u-BOP,this thesis designs a high probability guaranteed heuristic greedy algorithm(HPGH)based on relaxation theory and convex optimization theory.The simulation shows that the HPGH algorithm can not only greatly reduce the complexity of the algorithm,but also realize the dynamic on-demand allocation of bandwidth resources between the eMBB service and the uRLLC service.Secondly,based on the network service quality assurance of network operators,this thesis establishes a network slice resource orchestrate model aiming at minimizing network congestion(Min-NC).Min-NC is a MIP that contains two processes of virtual network mapping and resource allocation,and is also an NP-hard problem.Therefore,this thesis decouples Min-NC into network mapping and network resource allocation process.Firstly,the network mapping result is obtained based on the subgraph isomorphism theory.Then Min-NC is reintegrated into linear programming model(LP),and finally Lagrange multiplier method is adopted to quickly solve the network resource allocation result.Simulations show that the network congestion of Min-NC nearly decreased twice compared to the traditional shortest path tree algorithm(SPT).Finally,while ensuring the network operator's network service quality,in order to further improve the network resource utilization,this thesis introduces the dynamic resource pricing strategy based on the Min-NC model,and establishes the network slice resource arrangement model(Max-PP)with the goal of maximizing the network operator's profit.For Max-PP,this thesis designs a heuristic dynamic resource pricing strategy based on the optimal network congestion control results obtained by Min-NC,and finally solves it quickly by Lagrange multiplier method.Simulations show that compared with Min-NC and SPT,Max-PP can improve network resource utilization and network operator profit,and has similar network congestion performance with Min-NC.