Design And Implementation Of Flow Monitoring System In SDN Network

The rapid rise of mobile Internet,Internet of Things,cloud computing and big data technology have increased the growth of network traffic dramatically.When they bring positive impact on people's lives,they also pose unprecedented challenges on network management.Software Defined Network(SDN),as a new type of network in recent years,decouples the control layer and the forwarding layer from the traditional network,and open its interface,furthermore the control layer is logically centralized.The above features provide a new idea for traffic monitoring research in SDN network.This thesis takes full advantages of SDN,which is programmable,easy to expand,and has strong network control capabilities,to study network flow monitoring system in the SDN.The main contributions are as follows: 1)Research network flow monitoring and control from the perspectives of network behavior evaluation,network security and network planning,propose a flow monitoring framework in SDN network,and four functional modules are designed,which are used to collect network flow,to assess link importance based on flow,to monitor abnormal flow,and to predict network flow.2)Evaluate the importance of links in the network from the perspective of flow according to the network topology and service.3)Propose two kinds of abnormal flow detection algorithms according to the different locations of abnormalities.Using the rate of Packet_in message as a criterion for the anomaly detection in the controller,while detecting abnormal flows in the hosts through the way of extracting key five-tuple information and calculating the entropy.4)Propose the PSO-LSSVM network flow prediction model based on the basic model of LSSVM and the PSO algorithm.In this thesis,the flow-collect module,the link importance assessment module,the abnormal flow detection module and the network flow prediction module are implemented respectively.In the simulation environment built by Mininet,the function of the SDN flow monitoring system are tested.The experimental results show that the SDN network flow monitoring system can reasonably and correctly assess the importance of links in the current topology and service,and can detect abnormal flow in time and take effective control.In addition,it can accurately predict future flow based on current flow values.SDN network flow monitoring system provides a multi-angle reference information for the further management of the network.