A comprehensive plane-wise review of DDoS attacks in SDN: Leveraging detection and mitigation through machine learning and deep learning

Abstract

The traditional architecture of networks in Software Defined Networking (SDN) is divided into three distinct planes to incorporate intelligence into networks. However, this structure has also introduced security threats and challenges across these planes, including the widely recognized Distributed Denial of Service (DDoS) attack. Therefore, it is essential to predict such attacks and their variants at different planes in SDN to maintain seamless network operations. Apart from network based and flow analysis based solutions to detect the attacks; machine learning and deep learning based prediction and mitigation approaches are also explored by the researchers and applied at different planes of software defined networking. Consequently, a detailed analysis of DDoS attacks and a review that explores DDoS attacks in SDN along with their learning based prediction/mitigation strategies are required to be studied and presented in detail. This paper primarily aims to investigate and analyze DDoS attacks on each plane of SDN and to study as well as compare machine learning, advanced federated learning and deep learning approaches to predict these attacks. The real world case studies are also explored to compare the analysis. In addition, low-rate DDoS attacks and novel research directions are discussed that can further be utilized by SDN experts and researchers to confront the effects by DDoS attacks on SDN.