A Comprehensive Review of Deep Learning Techniques for the Detection of (Distributed) Denial of Service Attacks

Abstract

(Distributed) Denial of Service (DoS/DDoS) attacks are performed to bring down a target by flooding it with
useless traffic. Because the DoS/DDoS attackers often change their styles and attack patterns, the nature and
characteristics of these attacks need to be examined cautiously. Developing mechanisms to detect this menace
is a challenging task. Recently, deep learning has played a major role in the growth of intrusion detection solutions. In recent years, significant attempts have been made to construct deep learning models for countering
DoS/DDoS threats. In this review, we provide a taxonomy of DoS/DDoS attacks and deep learning-based DoS/
DDoS detection approaches. Then, the article focuses on the recent (from 2016 onwards) defensive methods
against DoS/DDoS attacks that exploit the advantages of deep learning techniques and discusses the key features of each of them. As datasets are imperative for deep learning techniques, we also review the traditional and contemporary datasets that contain traces of DoS/DDoS attacks. The findings from the review articles are as well summarized and they urge that more effort be made to strengthen the existing state-of-the-art approaches to coping with the dynamic behavior of the attackers. The imbalances in the surveyed articles are also highlighted. Finally, we outline a few key research directions that will need additional focus in the near future to ensure good security against DoS/DDoS attacks using deep learning approaches.