Research On Traffic Management Optimization In Software-defined Networks

In recent years,the increasing network scale of large cross-domain enterprises with multiple branches and data centers,and the application of novel technologies such as cloud computing,big data,Internet of Things,and artificial intelligence resulted in a huge amount of network traffic with special requirements within enterprise branches,between enterprise branches,and between enterprise cross-domain data centers,which is a great challenge to effectively improve network resource utilization.The traffic management method based on Software-Defined Networking can improve the utilization efficiency of network resources through the methods such as flow relocation and flow splitting under limited network resources.However,existing traffic management methods still have problems such as one-sided and static link weight calculation,difficultly in selecting the optimal path that meets the Qo S requirements of network services,and inability to avoid packet disordering while reducing the flow completion time.The dissertation makes full use of the advantages of the SoftwareDefined Network architecture and optimizes the aforementioned problems in SDN network traffic management according to various application scenarios and actual requirements of enterprise networks.The main research contents in this dissertation are as follows:(1)Focusing on the one-sided and static problem of link weight calculation in the current routing protocols,a dynamic link weight calculation method that considers the multi-attribute status of network links and the requirements of network services is proposed.In this regard,the existing SDN architecture is expended and the advantages of the SDN architecture are fully taken in the proposed method that comprehensively considers multiple attributes of the link,and realizes the dynamic calculation of the network link weight.To be specific,the method uses the contrast characteristics of the information entropy evaluation data set to solve the dynamic computing problem of link weight.Furthermore,the one-sided problem of link weight is solved by considering multiple attributes of the link and network service requirements.Experimental results show that the link weight calculated by this method has better dynamics and comprehensiveness when selecting the path.(2)Focusing on the problem of optimal path selection for the Qo S requirements of network services,a stepwise filtering Qo S optimal path selection scheme(SWQo S)is proposed under the SDN architecture.Specifically,the proposed scheme takes the network service requirement standard as the criterion of path selection and decides the optimal path that meets the Qo S requirements of the network service flow in step-by-step filtering selection of candidate paths between the source node and the destination node.When there is no path that fully meets the network service requirements in the candidate path set,the analytic hierarchy process combined with the fuzzy synthetic evaluation method is used to look for an optimal path with the path state closest to the network service flow transmission requirements.Experimental results show that compared with other routing methods,the proposed scheme can select the optimal path and all available paths that meet the requirements of network services,and the selected paths are optimal in terms of bandwidth,delay,jitter,and packet loss rate.(3)Focusing on the problem of multi-path data transmission inter cross-domain data-center of enterprises,which may easily cause packet disorder and unbalanced flow allocation,an SDN-based multi-path flow splitting scheme is proposed.This scheme is unequally to split flow aim at the shortest Flow Completion Time(FCT)according to the residual bandwidth capacity and delay of the path using multiple link-disjoint paths inter data-center.Experimental results show that this scheme can not only effectively avoid the problem of packet disorder,but also reduce flow completion time.