Research On Virtual Network Embedding Algorithm In Software Defined Networking

In recent years,the rapid development of various emerging technologies such as mobile Internet,cloud computing,big data and Internet of Things have led to the surge of network traffic and the clear trend of diversification of service requirements of users.The traditional ossification network architecture and complex network management technology face difficulties and challenges.As a promising architecture that decouples the control plane from the data plane,the software-defined networking(SDN)is expected to overcome the shortcomings of traditional network architecture and provide diversified service support for users.Network virtualization technology builds multiple heterogeneous virtual networks on a shared substrate network in a flexible and on-demand manner,which can realize the adaption among application scenarios,network functions and infrastructure platforms,so as to better support the diversified service requirements of users.SDN and network virtualization will become an important enabling technology of the next generation network.In this thesis,the SDN network architecture and network virtualization technology are combined and the problem of virtual network embedding in SDN is studied.The main contents of this thesis are as follows:Firstly,the key technologies of SDN and network virtualization are briefly introduced.The current research works on virtual network embedding algorithms are summarized and analyzed.As the existing virtual network embedding algorithms fail to comprehensively consider the reliability of nodes and links in the substrate network and the resource consumption of nodes on the substrate path,thus may cause problems such as limited virtual network reliability and load imbalance of the substrate network.In order to solve the problem,a virtual network embedding algorithm based on multi-objective optimization is proposed.Considering the constraints of node and link mapping,and substrate available resource,the virtual network embedding problem is formulated as a multi-objective optimization problem which achieves the network load minimization and the embedded reliability maximization.Since the optimization problem is a multi-objective integer programming problem,which is difficult to be solved directly,the ideal point method is applied to transform it into a single-objective optimization problem containing the feasible solution to the two subproblems,i.e.,network load minimization subproblem and embedding reliability maximization subproblem.Then the discrete particle swarm optimization algorithm and Dijkstra algorithm are applied to solve the problem,and the corresponding optimal virtual network embedding strategy can be obtained.For the network scenario where multiple virtual network requests being dynamic arrival,and considering the difference of performances such as the expenditure and power consumption of nodes and links in the substrate network,a virtual network embedding algorithm based on expenditure and power consumption joint optimization for SDN is proposed.Based on the analysis and evaluation of the expenditure and power consumption required for embedding virtual nodes and virtual links,the embedding cost function is defined as the weighted sum of the embedding expenditure and power consumption of virtual network.Under the constraints of resource requirements,the SDN virtual network embedding problem is formulated as a cost function minimization problem.The optimization problem is an integer linear programming problem,which is difficult to be solved directly.A time window-based batch embedding strategy is proposed to dynamically process online virtual network requests.The virtual network request in certain time window is transformed into virtual node embedding subproblem and virtual link embedding subproblem,and the corresponding heuristic algorithms are applied respectively to solve the two subproblems,the embedding strategy of virtual network request in substrate network can then be obtained.