Controller Deployment Algorithm In Software Defined Networking

The vertical integration architecture and distributed packet forwarding mechanism of traditional network lead to complex network control and reconfiguration,as well as highly limited packet transmission performance.In order to overcome the limitations of traditional network architecture,software-defined networking(SDN)is proposed as a new network paradigm,which separates control and forwarding through open interface and supports network programmability,so as to achieve flexible management and configuration of network and rapid deployment of new services.To enhance the scalability and reliability of SDN networks,multi-controller deployment mechanism can be used in the control plane.Thus,how to optimize and design a reasonable controller deployment strategy to achieve network performance improvement has become an urgent problem to be solved.This thesis focuses on SDN controller deployment problem.The main contents are as follows:In this thesis,we firstly give an overview of SDN from the aspects of concept,characteristics,architecture and key technologies.Then,according to the classification of optimization objectives,we summarize the research work on SDN controller deployment algorithm.Stressing the capacitated controller deployment problem in SDN networks,we propose a network cost optimization-based SDN controller deployment algorithm.By defining network cost as the weighted sum of network response time and controller deployment cost and jointly considering the optimization constraints on controller deployment,the capacity and type matching of the controllers,etc.,the capacitated controller deployment problem is formulated as a network cost minimization problem.The formulated problem is an integer non-linear programming problem,which is difficult to solve directly.Thus,we propose a two-stage heuristic algorithm which first tackles the controller deployment subproblem under the unlimited capacity constraint,and then solves controller-type matching subproblem.Specifically,in the first stage,we propose a minimum eccentricity-based controller deployment strategy to determine the number and location of controllers.In the second stage,we propose a controller-type matching strategy based on greedy algorithm to determine the type of deployed controllers.In the SDN network scenario with byzantine attacks,a multi-objective optimization-based controller reliable deployment algorithm is proposed.We formulate the interaction delay between switches and controllers,and the synchronization delay among controllers,and then formulate the multi-objective optimization problem based on the minimization of interaction delay and synchronization delay under constraints on the number,capacity of controllers and delay.In order to solve the optimization problem,we apply the multi-objective evolutionary algorithm based on decomposition to design the optimal controller reliable deployment strategy.Firstly,the multi-objective optimization problem is decomposed into multiple scalar optimization subproblems.Then,the initial reference points,population,neighboring individuals and the maximum number of iterations are initialized.Finally,new individuals are generated through operations such as crossover and correction,and neighboring individuals are updated according to the aggregate function value.In order to determine the initial reference points,we propose two heuristic algorithms for the interactive delay minimization subproblem and the synchronization delay minimization subproblem to obtain the feasible solution for controller deployment.In addition,a controller reliable deployment strategy based on greedy algorithm is proposed for small scale network scenarios.