A Survey on Deep Learning by Hybrid Approach for DDOS Attack and Prevention

Abstract

DDoS is one of the most dangerous threats on the Internet today which prevents access to vital services. The variety of attack methods and the amount of real-time traffic that needs to be analyzed make DDoS detection difficult. On the Internet, there are a sizable variety of network security tools that may be used to both create and defend against network assaults [1]. With the aid of advanced assaulting tools, attackers can produce attack traffic that resembles regular network traffic. Several defense strategies fall short in this context of real-time DDoS assault detection. In this paper we reviewed and studied different techniques for solving DDOS (Distributed Denial of Service) attacks and in future implement we will develop a hybrid approach that combines deep learning models which provides an effective feature extraction and finds most of relevant features that sets automatically without any human interpretation. We performed a thorough analysis of the DDoS issue in this research and suggested a straightforward taxonomy to classify the assault extent and potential mitigation options. This taxonomy helps assist software developers and security experts in comprehending the typical flaws that motivate attackers to initiate DDoS attacks.