Towards deep learning multi-classification of DDoS attacks in software-defined networks

Abstract

The landscape of Software-Defined Networking (SDN) is dynamically evolving, introducing an excess of security challenges, notably due to the Distributed Denial of Service (DDoS). This paper delivers profoundly into the exhibition and systematic classification of Deep Learning (DL) tools for DDoS attacks multi-classification. Several models have presented a binary categorization of normal and abnormal traffic, without discrimination among diverse identified attack types. This paper proposes an experiment of a set of models in simple and hybrid forms to identify five attack types in addition to normal traffic. Used solutions have been deployed in supervised mode using the CIC-DDoS2019 dataset. The resultant accuracy recorded is around 97% for LSTM simple model and DNN-GRU hybrid solution that have the highest values.