Machine Learning-Based Detection of Wormhole Attacks in IoT Networks Using Classification Models

Abstract

The widespread adoption of Internet of Things (IoT) networks has introduced new cybersecurity challenges, particularly in the form of wormhole attacks. These attacks pose a significant threat to IoT environments by manipulating network routing without altering packet contents, making them difficult to detect using traditional intrusion detection systems (IDS). This study explores the application of machine learning (ML) techniques for detecting wormhole attacks in IoT networks. The research compares five machine learning classifiers Sparse Representation Classifier (SRC), Multi-Layer Perceptron (MLP), Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), and XGBoost based on metrics such as accuracy, precision, recall, F1-score, and computational efficiency. Data preprocessing techniques were applied to a publicly available IoT dataset to improve the performance of these models. Among the classifiers tested, XGBoost demonstrated superior performance with a detection accuracy of 99.97%, outpacing both traditional and deep learning models. The results highlight the potential of ensemble learning approaches in enhancing IoT security, especially for real-time applications in resource-constrained environments. The study underscores the importance of balancing detection accuracy with computational efficiency when selecting models for dynamic IoT networks. Future work will explore federated learning and hybrid deep learning models to further improve the detection capabilities of wormhole attacks in IoT settings.