Enhancing Cybersecurity in IoMT Environments: Machine Learning-Powered Anomaly-Based Intrusion Detection Systems
Résumé
Intrusion Detection Systems (IDS) play a crucial role in protecting intelligent systems, particularly in critical domains such as cybersecurity and data analysis. In this work, we focus on anomaly-based IDS powered by machine learning, which must effectively detect a wide range of attacks to ensure robust security. This approach is especially essential in IoMT (Internet of Medical Things) environments, where protecting sensitive data and preventing risks related to cyberattacks are paramount. To address these challenges, we evaluated five supervised machine learning models using the CICIoMT2024 dataset, which contains various types of attacks classified into six categories: DDoS, DoS, MQTT, benign, reconnaissance, and spoofing. Our results demonstrate that our ensemble learning-based IDS, leveraging models such as Random Forest, CatBoost, LightGBM, and XGBoost, achieved a top accuracy of 99.39%, highlighting its strong anomaly detection capabilities. These results underscore the solution's potential in ensuring the integrity and confidentiality of medical information in IoMT environments.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |