Research On Joint Optimization Of Virtualized Network Function Placement And Routing Allocation Problem In NFV

For network performance and data security,request transfer in modern networks usually requires strict execution policies.A policy is represented by a service function chain consisting of a series of network functions.Network Function Virtualization(NFV)decouples network function from specified appliances onto general shared servers in the term of software.Thus,it has been being regarded as a promising technology for Internet Service Provider to reduce high Capital Expenditures(CAPEX)and Operational Expenditures(OPEX)on the network function placement and maintenance.In NFV,the network function deployed on servers with NFV-based virtual machine is named as the virtualized network function(VNF).One critical issue for VNF placement is the resource allocation of demanded network functions,which guarantee the order of the service function chain.The problem is inherently NP-hard.Simple VNF placement algorithms do not routing allocation with respect to network requests,leading to these algorithms not applicable in large-scaled scenarios where network requests arrive to the system dynamically.Therefore,this thesis aims to research on joint optimization of VNF placement and routing allocation problem in NFV.The main research contents are as follows:(1)Research on joint optimization of VNF Placement and routing allocation for operational expenditure: These come with high OPEX for service providers,due to the complexity of the network infrastructure.To solve this problem,a strategy for OPEX optimization is proposed,which aims to minimize the startup cost,energy consumption,transmission cost and obtain VNF deployment and routing allocation optimization scheme.The VNF deployment problem as a new Mixed Integer Linear Programming(MILP)model is formulated,and three OPEX optimization algorithms are designed including Genetic Algorithm(GA).The OPEX of MILP model and optimization algorithms are compared under different resource allocation constraints.The calculation result shows that the genetic algorithm can obtain the near-optimal solutions when node resource ratio is more than 60%.(2)Research on joint optimization of VNF placement and routing allocation for throughput maximization: In the face of increasing network traffic and service requests,this poses a new optimization direction for VNF placement problem,that is,how to find an appropriate shortest routing path for each user request such that the data traffic of the request will pass through all VNFs in their orders in the service chain of the request,with the objective to maximize the network throughput,subject to various resource capacity constraints.To solve this problem,we first formulated an integer linear programming model OPL,and then designed two VNF placement algorithms to achieve a good tradeoff between maximizing throughput and minimizing routing delay.Through simulation experiments,it is found that the GAP algorithm can get the placement scheme with higher throughput,while the HEU-T algorithm can get the placement scheme in a shorter time.