A Novel Hybrid Flow-Based Handler with DDoS Attacks in Software-Defined Networking

Abstract

In this paper, we firstly introduce a new combined approach to enhance the performance of classification in network traffic. The proposed combination mainly focuses on taking advantages of two classification algorithms, Support Vector Machine (SVM), Self Organizing Map (SOM). We utilize both advantages that SVM takes a little time to produce outputs with a high accuracy, SOM makes a reliable prediction based on their neurons. Next, we propose a hybrid flow-based work mechanism, which applies the proposed combination SVMs-SOM, for handling with Distributed Denial-of-Service (DDoS) attacks, network component protection from resource exhaustion in Software-Defined Networking. In the introduced scheme, multiple Linear SVMs first classify flow entries existing in flow-tables from OpenFlow switches. In the case that a flow's position is located between two margin lines or a vague region in the Linear SVM representation, it will be then forwarded to a SOM to make a final decision. Afterwards, an attack classifier, a policy enforcement module will be applied to attack flows with the purpose of attack diminution, the SDN controller protection. Besides, we also provide readers with a new view of DDoS attacks in the Software-Defined Networking. Thorough practical experiments conducted in the Software-Defined Networking environment, it is proved that the proposed classification combination outperforms original algorithms,, the novel hybrid mechanism can be an effective, innovative approach to face with DDoS attacks, protect the OpenFlow switches, the SDN controller from being overloaded.