Research On Traffic Analysis And Identification On SDN Environments

With the rapid development of the Internet and the diversification of business hosted by the Internet. A variety of network attacks make use of the increasing network traffic. DDoS attack is a typical form of these network attacks. The attacker schedule these collected hosts send falsified data simultaneously to bring the victim down which makes the traffic analysis and identification more and more significant.Software Defined Networks is a novel architecture where the control planes and data planes are decoupled. Among the first and widespread mechanisms to support SDN architecture is the OpenFlow protocol which can be effectively applied with a rich diversity in load balancing, traffic management and routing. In this paper, we extend these functionalities for abnormal traffic detection in SDN architecture. In this paper we firstly introduce the feature, detection, defense and trace of DDoS attack, the character of SDN and OpenFlow. Then we summarize the existed flow-based DDoS detection method. We collect five proper features for DDoS detection and KNN-based classification method and SVM-based classification method was proposed for classifying the collected features. The feasibility of the method is validated through experiment based on SDN controller NOX and SDN switch NetFPGA.