Detection of DDoS Attacks in SDN Using Machine Learning Approaches: A Review

Over the past several years, the Distributed Denial-of-Service attack has emerged as a big threat to Software-Defined Networks because of its frequent attack on SDNs. The DDoS attack, partially or fully, stops the genuine users of network services from accessing the network resources. A large amount of network resources is consumed to ensure the security of the SDN controllers. This factor leads to the poor performance of the SDN controllers in managing the entire network. The research and situation become tougher and stimulating as there are numerous categories of DDoS attacks against SDNs. In order to protect SDN networks from DDoS attack, one must understand the basic characteristics of SDN; network traffic behavior when DDoS attack occurs against SDN; and behaviors of DDoS attack. Different studies on DDoS attacks against SDN had shown many specific clear characteristics and features that can be used as indicators and in turn help us to identify DDoS attacks. With the help of machine learning techniques, one can detect DDoS attacks on SDNs. In this paper, various machine learning techniques which can be helpful in detection of DDoS attacks on SDNs are discussed. Also, a proposed solution is suggested at the end where the best machine learning (ML) with highest accuracy, precision, and recall will be used to create an inference engine for early detection of DDoS attack in SDN.