Design And Implementation Of Controller Deployment And Assignment Scheme For Internet-oriented SDN

With the rapid development of cloud computing,Internet of things,and mobile Internet,etc,as well as the explosive growth of network applications and network information scale and variety,users are getting more customized demanding.This brings in great challenge to existing network architecture.Therefore,future Internet and its related technologies become new research hot topics.SDN(Software Defined Networking),which emphasizes decoupling the control plane and the data plane,can not only meet the customized demand through programming,but also implement and deploy network strategies rapidly.Due to the simple network architecture and good compatibility.SDN has attracted attention from academia,as well as got support from device manufactures.Although SDN has been studied and developed in some networks,such as data center networks,enterprise networks and campus networks,etc.,the feasibility and availability of SDN used in the super-large scale Internet have not been verified.When SDN is deployed in the Internet,multiple controllers are often required.The placement of these controllers influences the performance of SDN from many aspects.Therefore,it is of great importance to find an efficient SDN controller deployment and assignment scheme for Internet.The performance optimization objects in this thesis include maximizing the network reliability,maximizing the load balance of controllers and minimizing worst-case latency of control plane.High-efficiency controller deployment and assignment scheme are proposed to optimize the placement location selection and the assignment to meet the switches' request.First,This thesis proposes the SDN network structure model that fits for the Internet multi-domain fact.Then,This thesis introduces the problem description and mathematical model on the basis of objective function.Bio-inspired Computing are used to deal with the computing complexity.Based on biotic individual behavior patterns-guide ABFO(Adaptive Bacterial Foraging Optimization)and biotic community evolution patterns-guide MSPSO-M(Multi-species Symbiotic Particle Swarm Optimizer-Mutualism)algorithm,the controller deployment and assignment scheme are designed to achieve the optimization objects.This thesis also introduces heuristic rules according to the network practical situation.According to the topologies of CERNET2,CERNET and GeANT2,this thesis simulates the proposed scheme evalutes its performance Results show that the scheme designed in this thesis realizes the optimization objects and can provide some support and reference for SDN study and development.It has practical significance to actual network planning.