Distributed Denial of Service Attacks Detection using Deep Learning in Software Defined Network

Abstract

Software-Defined Networking (SDN) is an emerging architecture that provides high flexibility by separating forwarding functions and network logic. The programmability feature of SDN makes network management efficient. However, centralized nature of the controller poses security threats to the SDN environment. Distributed Denial of Service (DDoS) attack is the most popular cyber-attack which can cause economic loss due to network disruption. Thus, the design of a DDoS detection system that can detect attacks accurately in an SDN environment is an important research issue. Most of the work carried out in the literature on DDoS attacks detection is based on the benchmark datasets created for the Internet. However, there is a need to work on SDN datasets due to their unique features. In this work, we consider three SDN datasets for classifying DDoS attacks. We use three deep learning modules namely Multi-Layer Perceptron (MLP), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM). The performance of the models is measured using model accuracy and model loss. Results reveal that LSTM performs better than MLP and CNN for all three SDN datasets.