Research On Multi-Constrained QoS Routing Optimization Based On Traffic Classification In Software-Defined Networks

Software Defined Networking（SDN）is a new network architecture that loosely couples the control plane and the forwarding plane by means of Numerical Control separation,utilizing network programmability to flexibly control network traffic and route scheduling.In order to guarantee the network quality of service（QoS）for different types of service data transmission,this paper proposes a multi-constrained QoS routing optimization method based on traffic classification under SDN architecture.The main contents and contributions are as follows:（1）To address the problem of differentiated QoS guarantees for different service types,the service types are classified into session class,streaming class,interaction class,and background class according to the QoS requirements,and an improved classification model（Improved multi-granularity cascade Forest,Igc Forest）based on multi-granularity cascade forest is proposed.Firstly,the Z-score method is used to normalize the dataset so that the feature samples with different range values have the same metric.Secondly,the noise and edge samples in the dataset are removed by the A-SMOTE algorithm.Finally,the gc Forest classification model is improved by introducing a binning-based confidence sieving mechanism in the cascade forest structure,so that each level of the cascade forest only passes the feature samples with low confidence to the next level.The random sampling method is introduced in the multi-granularity scan to reduce the dimensionality of the transformed features.Simulation experiments show that the Igc Forest classification model can accomplish the classification tasks of different business types and has higher classification accuracy and performance compared with the random forest,SVM,gc Forest,and gc Forest_cs classification models.（2）To address the problems of multi-constraint QoS routing and differentiated QoS guarantees for data transmission for different service types,we propose a multi-constraint QoS dual path routing optimization algorithm in SDN based on traffic classification（Multi constraint QoS Dual path routing optimization algorithm based on Traffic classification in SDN,SDN-MQDT）.Firstly,the logarithm of the packet loss rate and the overhead caused by the successful delivery of the data stream by the link are used as the cost function,and the multi-constraint problem is simplified by aggregating multiple constraint performance metrics into the cost function through a Lagrangian relaxation algorithm.Secondly,a directed acyclic graph based on the destination node is generated to reduce the routing computation time,and the data is forwarded through the QoS metric in the data plane to achieve the differentiation guarantee of QoS.Finally,the reverse link deletion algorithm is used to obtain two non-intersecting optimal paths to guarantee data transmission in case of link or node failure.Simulation experiments show that SDN-MQDT can reduce the routing computation time and meet the QoS flow satisfaction of different service types.（3）Combining the traffic classification model and multi-constraint QoS routing algorithm for secondary development of the Ryu controller,we design and implement a visual network management system with three functions:authorizing unknown data transmission,viewing the flow table of authorized data transmission,and manually adding or deleting the flow table.The test results show that the system can monitor and manage the transmission paths of authorized unknown data and improve the efficiency of route management.