Analyzing the Evolution of DDoS Attack Detection: Traditional vs. Modern Machine Learning Models

Distributed Denial of Service (DDoS) attacks continue to pose a significant threat to network and system security. This research explores the transition from traditional to modern machine learning models in DDoS attack detection. Their effectiveness in identifying DDoS attacks, considering factors like accuracy, scalability, and real-time detection capabilities was explored. Traditional models are discussed in their historical context, along with their limitations in handling sophisticated DDoS strategies. In contrast, this paper investigates the capabilities of modern machine learning models, specifically designed to address the contemporary threat landscape. LSTM networks demonstrate their proficiency in sequential data analysis, CatBoost’s gradient boosting approach showcases its efficiency in large-scale datasets, and GANs provide insights into their anomaly detection potential. Additionally, the J-48 algorithm, a decision tree classifier, is evaluated for its adaptability to DDoS attack detection. This paper presents a comparative analysis that reveals the strengths and weaknesses of both traditional and modern machine learning models, aiding the security practitioners and researchers in selecting the most suitable approach based on their specific requirements and network conditions.