AN OVERVIEW OF USING OF FRACTAL ANALYSIS FOR DETECTING DDOS NETWORK ATTACKS

Abstract

Distributed Denial-of-Service (DDoS) attacks are among the most severe cybersecurity threats, continuously evolv- ing and causing extensive financial losses worldwide. Unlike traditional Denial-of-Service (DoS) attacks, DDoS attacks leverage multiple compromised systems, creating a coordinated effort to overwhelm network resources and disrupt ser- vice availability. The growing complexity of these attacks, often indistinguishable from legitimate traffic, presents sig- nificant challenges for detection and mitigation. This article examines various machine learning techniques for DDoS detection, including Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Trees, and Naïve Bayes, all of which demonstrate high accuracy in classifying attack patterns on synthetic datasets. While effective in controlled environments, these methods often struggle with the nuances of real-world network traffic, where hybrid and novel attack types obscure detection efforts. To address these challenges, we explore the application of fractal analysis, a promising approach for identifying self-similarity in network traffic.Fractal analysis, which captures the self-similarity in network traffic, has shown potential for identifying abnormal patterns indicative of DDoS activity. Although it has limitations, fractal analysis can improve the detection process when combined with statistical features and machine learning algorithms.Fractal characteristics, such as those quantified by the Hurst Exponent, reveal long-term dependencies and auto- correlation structures in traffic, making them suitable for detecting irregularities commonly associated with DDoS attacks.Our analysis demonstrates that fractal-based methods, when combined with statistical and machine learning approaches, can enhance detection accuracy and improve adaptability to real-world scenarios. Although no single method offers a universal solution, this study underscores the importance of using diverse techniques to effectively monitor and protect against DDoS threats. Further research should focus on integrating multifaceted detection models to better address the evolving landscape of cybersecurity threats posed by DDoS attacks.