Mathematical and Logical Modeling of DDoS Attacks: A Hybrid Approach Using Propositional Logic and Machine Learning

Abstract

As cloud hosting continues to dominate, a major drawback is the increased vulnerability to Distributed Denial-of-Service (DDoS) attacks, which threaten the availability and reliability of internet services. This paper proposes a new hybrid DDoS detection and mitigation technique that combines mathematical logic and machine learning techniques. This frame uses propositional logic and FOL to speak of traffic patterns that are DDoS attacks, which makes it possible to reason automatically to classify attacks. Other machine learning models, including Logistic Regression, Decision Tree, Random Forest, and Deep learning architectures such as CNN and RNN, operate to achieve a higher rate of detection of anomalous traffic patterns. The approach entails real-time traffic analysis-based inference mechanisms on attack detection and logical resolution handling to minimize false positives. The results demonstrate that most of the examined models reach high accuracy on detection, having achieved perfect classification accuracy on Decision Tree and Random Forest. It was the deep learning paradigm that proceeds to achieve above 99%. The study underscored the success associated with fused logical reasoning blended towards data-driven approaches for transfer-cloud security from availability issues down to threat mitigation in their active line of operations. The prospects of the future will circle around equipping users with learning capability enough to counter DDoS experimental attack evasion maneuvers.