Research On Methodology Of Virtual Network Function Intelligent Orchestration In NFV

Virtual Network Function(VNF)is a Network Function implemented on a general-purpose server using Network Function Virtualization(NFV)technology.Network Service Request(NSR)becomes a request to run a set of VNF deployed on the physical network.However,VNF deployment has been proved to be NP-Hard problem.Aiming at the problem of deploying the dynamic and complex NSR in 5G network to the physical network and adaptively performing VNF migration,there are still great challenges of difficulty in solving and inefficient resource utilization,so the research of VNF orchestration method is of great significance.Aiming at the VNF orchestration problem in 5G dynamic and complex network scenarios,a cost optimization oriented VNF deployment method was proposed,and on this basis,an energy-aware VNF migration method was studied.Finally,based on the research results,a VNF intelligent orchestration system was designed and implemented.The detailed research content is as follows:(1)In order to overcome the challenges of solving difficulties and inefficient resource utilization caused by different arrival rates and complexities of NSR,a method combining Graph Convolution Network,Sequence to Sequence and Reinforcement Learning,referred to as GSR,was proposed.The Asynchronous Advantage Actor-Critic algorithm is used to train the S2S model.The experimental results show that GSR has the best resource utilization efficiency and NSR acceptance rate compared with the other six algorithms,and the solution speed is faster.(2)In order to solve the problem of energy inefficiency caused by dynamic NSR,an energy-aware VNF migration method based on improved KM algorithm was proposed.Aiming at the matching problem of node migration,this paper proposed an improved node matching algorithm based on KM.As for how to perform migration,it designed a greedy node migration strategy and an energy-aware link migration algorithm.Finally,the experimental evaluation shows that the proposed method significantly reduces the number of active physical nodes and energy consumption compared with the state-of-art method.(3)Based on the proposed VNF deployment and migration method,an intelligent VNF orchestration system is designed and implemented.Based on the requirement analysis of the system,the overall design is clarified.Then the detailed flow chart of the system is given.Finally,the system is developed and implemented.