Abstract

Dealing with network security has always been challenging work, especially due to the prevalence of Distributed Denial of Service attacks. A DDoS attack occurs when a hacker takes control of several computers, turning them into bots, and then sends a large number of requests simultaneously to specific servers on the internet. This causes the targeted server to become too busy to provide normal services to legitimate users, those who need to use that particular server. Various Deep-learning algorithms have been used to identify DDoS attacks in this research. To study and analyse DDoS attacks, researchers have used the CIC-DDoS 2019 dataset. In this paper, the primary goal is to make a comparison of the performance of various DL algorithms for both binary and multi-class prediction of detect DDoS attacks accurately, such as Convolutional Neural Network, Deep Neural Network, Long Short-Term Memory Network, Recurrent Neural Network, Feedforward Neural Network, Radial Basis Function Network, and Multilayer Perceptron. Along with that, the experimenter demonstrated that DDoS attacks can be better identified if they are stored in the dataset in a binary way.

Keywords: DDoS attack, Deep learning, DL algorithms, CIC-DDoS 2019 dataset, Cyber security