Deep learning-driven defense strategies for mitigating DDoS attacks in cloud computing environments

Abstract

The kind of cyber threat prevalent and most dangerous to networked systems is the Distributed Denial of Service (DDoS), especially with expanded connection of Internet of Things (IoT) devices. This article categorizes DDoS attacks into three primary types: volumetric, protocol based and application layer of cyber attacks. It discusses the application of security threats that arise from the use of the DL models, accusing recently introduced ideas and stressing pitfalls: the issues of data and methods scarcity. There is the same need for the greater use of explainable and transparent AI to improve confidence in such security systems as is noted in the review. It also reveals that present detection performance is constrained and frequently obstructed by the poor quality of the datasets. The future work is proposed to build superior datasets and use accurate algorithm to improve the security models. This paper focuses on explainability as a way of making the AI model creation process and any consequent decisions explainable and transparent. The use of deep learning enhances the capability of cybersecurity in handling DDoS attacks and preventing or controlling them. But it has to be a part of a more large-scope platform, based on multiple types of longitudinal or cross-sectional data combined with high efficiency, explainable AI. The article ends with call to proceed with studying and advancing the AI application in response to new threats, and make the most of it to enhance protection of the contemporary networked environment.