Detection of Distributed Denial of Service Attacks Based on Deep Learning Approaches: A Survey, Taxonomy, and Challenges

DDoS attacks are among the most dangerous dangers to the digital world, according to recent theoretical and empirical research. Over time, DDoS attack mitigation strategies have developed to guarantee security. In the past, several traditional techniques, including heuristics and signatures, were employed to detect DDoS attacks encoded with different characteristics. The advanced obfuscation strategies used by new generations of DDoS attackers were too formidable for detection tools designed for traditional DDoS attacks. Since DL-based systems beat traditional DDoS attack detection techniques in discovering novel DDoS attack variations, Deep Learning (DL) is being employed more and more in DDoS attacks. Additionally, DL-based methods offer quick DDoS attack prediction together with superior detection rates and DDoS attack analysis. Thus, this work is interested in examining recently suggested DL-based DDoS attack detection systems and their development. It provides a comprehensive examination of the most current advances in DL-based detection methods. This survey's main objective is to give readers a thorough grasp of the applications of DL for detection. The outcome of this review discusses various DL methods, their strengths and weaknesses, datasets, challenges of recent research work, and future enhancements of present works.