Prevention and Protection Against DDoS Attacks Using Machine Learning (Classification Algorithms)
Loading...
Date
2024
Journal Title
Journal ISSN
Volume Title
Publisher
Faculty of Sciences
Abstract
In the current era of extensive digitalization across various aspects of human life, we are now faced with a complex intersection of diverse systems that regulate our daily routines. Regrettably, these systems are under constant threat from rapidly evolving attacks. As a consequence, safeguarding these systems poses a significant challenge for organizations, companies and individuals alike, as they all rely on common internet services. The focus of our thesis has been on the prevention and detection of Distributed Denial of Services (DDoS) attacks through the application of machine learning techniques, specifically utilizing classification algorithms like Random Forests, Decision Trees and AdaBoost. Our evaluation centered on the effectiveness of these methods in recognizing abnormal traffic patterns associated with DDoS attacks. While Random forests, which amalgamate multiple decision trees, exhibit robustness and efficiency, Decision Trees, despite their simplicity and speed, are susceptible to overfitting. Notably, AdaBoost, which enhances model performance by assigning weights to the errors of weak classifiers, merges as the most proficient in identifying DDoS attacks in our tests. The findings indicate that AdaBoost delivers superior accuracy compared to other algorithms.