Research On The Mapping Of Service Function Chains In NFV Scenario

With the explosive growth of mobile devices and the emergence of new network technologies,the scale of traditional networks has become larger while the income of operators has gradually decreased.Therefore,operators are eager to reduce the network complexity and to improve the efficiency of service deployment when the diversified needs of users are satisfied.The emergence of network function virtualization(NFV)technology has brought an opportunity for the solution of this problem.This technology realizes traditional network functions by running specific software on general hardwares such as servers and switches,and constructs a high-performance network with better scalability,flexibility and adaptability at lower cost.In the NFV network scenario,the Service Function Chain(SFC)describes specific network services by connecting virtualized network functions(VNF)to each other in the order required by service logic through virtual links.However,the SFC deployment also faces some challenges.How to achieve resources efficient allocation while meeting business needs is one of the major issues to be solved urgently in SFC deployment.Firstly,the thesis summarizes its research background and significance,and introduces the study status of NFV and SFC mapping at home and abroad.Secondly,the thesis introduces the NFV concept and technical framework,and analyzes the relationship between NFV and SDN.It also explains the SFC mapping principle,and summarizes the advantages and disadvantages of different mapping algorithms according to different classification regulations.Then the basic principle of genetic algorithm(GA)is described,and the advantages of GA over traditional algorithms are listed.On the basis of summarizing the existing research results of SFC mapping in NFV scenarios,this thesis focuses on the key technologies of implementing SFC mapping in both online and offline SFC deployment scenarios according to the different business models.For the online SFC deployment problem in the NFV scenario,the third chapter comprehensively considers the current network characteristics and service requirements,while relaxing the delay constraint of uses' services,and defining the delay penalty.Therefore,this chapter establishes an Integer Linear Programming(ILP)model of the SFC mapping,and proposes the service function chain mapping algorithm based on mixed genetic Algorithm(SMGA)to solve the model.The proposed algorithm achieves the joint optimization at both stages of the VNF mapping and virtual link mapping,employing the combination of genetic algorithm and precise algorithm based on the optimal numerical calculation CPLEX,respectively.Compared with the existing algorithm,simulation results show that the proposed algorithm has a certain improvement in terms of the service acceptance ratio,resource utilization ratio and VNF provider profit.For the offline SFC deployment in the NFV scenario,the fourth chapter considers the VNF types.Meanwhile,the mechanism of the mutual conversion among different types of resources in each underlying server node is adopted for reducing the impact of the underlying network resource fragmentation on the SFC mapping.Thus,this chapter establishes an ILP model aimed at minimizing resource occupation for SFC mapping,and proposes a resource types convertible SFC mapping algorithm(RTCSMA)to solve the proposed model.The proposed algorithm maximizes the number of effectively multiplexed VNFs through the resource transformation mechanism to complete the VNF mapping,and it also considers the link occupied bandwidth and link betweenness centrality to realize the virtual link mapping.Compared with the existing algorithm,simulation results show that the proposed algorithm can greatly decrease the resource consumption in the underlying physical network,reduce the number of VNF instantiations and the maximum link occupation in the underlying network.The fifth chapter summarizes the work of this thesis and presents the further research work from five aspects: network scenario,SFC model,underlying physical network topology,SFC mapping optimization objectives and optimization algorithms.