A MULTI-TIER ML-BASED FRAMEWORK FOR DETECTING DISTRIBUTED DOS ATTACKS IN IOT NETWORKS

Abstract

Through the thorough deployment and assessment of machine learning-based detection systems, this study investigates the growing threat of Distributed Denial of Service (DDoS) assaults directed at Internet of Things (IoT) ecosystems. As IoT deployments expand rapidly across industries, these resource- constrained devices present unique security challenges and attractive targets for cybercriminals. This paper proposes a novel detection approach specifically tailored for IoT environments, implements and compares the efficacy of various machine learning classifiers for DDoS detection, systematically identifies individual attack vectors in existing IoT models, and provides complete source code implementation for reproducible research.  According to our experimental findings, ensemble-based methods outperform single classifiers in terms of detection accuracy (97.8%) while retaining a manageable computing overhead. The lightweight detection framework we propose integrates edge computing components to enable real-time threat mitigation with minimal impact on IoT device performance. With the help of this research's comprehensive Python implementation and thorough code documentation, practitioners can deploy and modify the solution to suit their unique IoT setups.