Design Of Network Function Virtualization Platform And Related Deployment Algorithm

New business of telecommunication comes with the the popularity of the Internet.However,the traditional network with hardware and software tight coupled has limited expansibility probleam and high maintenance costs probleam,and it will bring great cost pressure to operators when opening up new business.In order to solve these above problems,the academic community proposed the concept of Network Function Virtualization(NFV).NFV advocates achieving network function through software and make it running on the common hardware platform with virtualization technology,so called the Virtual Network Function(VNF).So that operators can schedule and control VNF according to demand,thereby enhancing the flexibility of the network,shortening the development cycle of new business and reducing the cost.A number of open source NFV platforms which promote the development of NFV ecological environment have been emerged from industry.But there are still some problems of these platforms,such as incomplete core function and unconvenient deployment and usage experience.This thesis designed and realized a NFV platform based on MANO(Management and Orchestration),which is friendly to deployment,usage and orchestration algorithm research.This NFV platform provids six core functions: network service orchestration,infrastructure management,VNF management,descriptor catalogue and network service entity management.Network service are composed by a number of VNF and how to deploy VNF is one of the important issues of NFV platform's achieving orchestrating network service.Due to the different resource characteristics,different resources with different available sets of network service only support a specific kind of VNF,that is,resources are heterogeneous.The heterogeneity of resources makes the platform having different support capacity for different network service.As a result,the fairness problem under heterogeneous resources need considers not only the resources competition between network services but also the difference of supporting capacity of network service compared with the fairness problem under homogeneous resources.This thesis did research into VNF fairness deployment in heterogeneous resource scenario,established relevant mathematical model,proposed the Greedy for Least Influence-Factor Algorithm(GLIF)and Two Phase Algorithm based Tabu Search Algorithm(TPTS).GLIF sorts the various types of network service according to the degree of resource shortage and avoids selecting resources required by network service which has higher degree of resource shortage.TPTS uses tabu search to optimize the selection of resources,balance the degree of resource shortage and try to deploy failed requests in initial solution.The simulation results show that the GLIF and TPTS are more balanced than the traditional greedy algorithms in resource pool and deployment num of all kinds of network service,which are more suitable to meet the fair demand in resource heterogeneous scenario.