Research On Several Optimization Problems In Software-Defined Networking

With the rise of new technologies such as artificial intelligence,big data,cloud computing and the Internet of Things,people have put forward new requirements for the functions and performance of the computer networks behind these technologies.Due to the closed and rigid characteristics of traditional network architecture,it is more and more difficult to meet people's demand for services.Based on these technical backgrounds,Software-Defined Networking(SDN)came into being.Due to its advantages of strong controllability,scalability,and programmability,the application scope of SDN is becoming wider and wider.However,some optimization problems in traditional networks also exist in SDN.Although some network optimization methods in traditional networks are still applicable in SDN,some unique features of SDN also bring some new problems.It is necessary to study some optimization problems in SDN based on these new features.In this dissertation,the optimization problems in SDN are studied from two popular problems: energy optimization and virtual network embedding optimization.The main work and innovations of this dissertation are summarized in the following two aspects:(1)In terms of energy optimization,we fully pay attention to the limited bandwidth resources of data center network and the limitation of SDN switch flow table capacity.Because of the limitation of network resources,the method of flow aggregation inevitably leads to the failure of flow transmission.The existing research on energy optimization ignores the problem of flow reliability.In this dissertation,we fully consider the Qo S(Quality of Service)requirements of flow and propose a minimum energy consumption algorithm based on flow priority scheduling strategy.Firstly,we established an integer linear programming model of the Software-Defined Data Center Network(SD-DCN)energy optimization problem based on Qo S requirement constraints.Secondly,we propose a minimum energy consumption heuristic algorithm based on three different flow scheduling strategies to achieve a balance between energy optimization and flow reliability.Experimental results show that the proposed algorithm can effectively save energy under low traffic load and medium traffic load.Under high traffic load,a high Qo S satisfaction ratio can be obtained.(2)In terms of virtual network embedding optimization,we fully pay attention to the unique feature of controller in SDN,and existing researches divide the virtual SDN network embedding process into three independent stages: controller placement,node embedding,and link embedding.Due to the interdependence of the use of resources at various stages,this multi-stage approach will affect the performance of virtual SDN network embedding.In order to solve this problem,a novel virtual SDN network embedding algorithm and its improved algorithm based on coordinated mechanism are proposed by establishing a coordinated mechanism that binds the controller and virtual node location and considers the subsequent link embedding at the node embedding stage.Specifically,in the node embedding stage,we use the coordinated mechanism to embed the controller and virtual node.We only need to embed the virtual node,and combine the two steps of controller placement and virtual node into one step,and concentrate the underlying physical nodes mapped by virtual nodes as much as possible to reduce the cost of subsequent link embedding.Experimental results show that the proposed algorithm can effectively improve the embedding performance of virtual SDN network.