Deep learning-based distributed denial of service detection system in the cloud network

Abstract

Cloud computing offers an efficient solution that enables businesses and users to deliver flexible and scalable services by sharing resources. However, this shared resource pool also exposes vulnerabilities to various cyber threats, such as Distributed Denial of Service (DDoS) attacks. These DDoS attacks, due to their potential impact, can be highly destructive and disruptive. They render servers unable to serve users, leading to system crashes. Moreover, they can severely tarnish the reputation of organizations and result in significant financial losses. Consequently, DDoS attacks are among the most critical threats faced by institutions and organizations. The primary objective of this study is to identify and detect DDoS attacks within cloud computing environments. Given the challenges associated with acquiring a cloud-based dataset, the main motivation behind this research was to construct a dataset within a cloud-based system and subsequently evaluate the intrusion detection capabilities of deep learning (DL) algorithms using this dataset. Initially, an HTTP flood attack was executed after creating a network topology within the OpenStack framework. The study employed Convolutional Neural Network (CNN), Artificial Neural Network (ANN), and Long Short-Term Memory (LSTM) models for attack detection. The performance of these models was assessed using various measurement metrics, and it was found that the LSTM model delivered the most impressive results, achieving an accuracy rate of 98%.