An Ensemble-based Machine Learning Approach for Botnet-Based DDoS Attack Detection

In the realm of Software-Defined Networking (SDN), where network dynamics evolve rapidly, safeguarding against Distributed Denial of Service (DDoS) attacks remains paramount. This research introduces a novel machine learningdriven approach designed to fortify network security by effectively identifying botnet-based DDoS attacks. This approach uses an ensemble-based Random Forest classifier in conjunction with feature selection, data preparation, and other methodologies. To identify the most significant features supporting attack detection, the feature selection makes use of cutting-edge methodologies including principal component analysis, mutual information, and correlation analysis. Performance evaluation is conducted using a comprehensive set of metrics, which includes recall, precision, accuracy, and F1 measure. Notably, this model excels in addressing class imbalance, elevating its effectiveness by deploying the Synthetic Minority Over-sampling Technique (SMOTE) to harmonize dataset disparities. Empirical findings underscore the model's proficiency, delivering high-fidelity detection of botnet-driven DDoS attacks in an SDN environment as evidenced by elevated accuracy and balanced accuracy scores.