Adaptive lightweight intrusion detection system for internet of things and internet of medical things security

Abstract

The rapid development of Internet of Things (IoT) and Internet of Medical Things (IoMT) environments has increased the risks of cybersecurity, especially from DDoS, DoS, and other complex cyberattacks. Extensive research has been carried out on existing Machine Learning (ML) and Deep Learning (DL) based Intrusion Detection Systems (IDS); however, they still face several challenges such as class imbalance, high computational complexity, and poor generalization over heterogeneous datasets. This research addresses the gaps by proposing an Adaptive Weighted Feature-Optimized Lightweight Intrusion Detection System (AWFOL-IDS). The main goal is to design a scalable and efficient IDS that accurately detects attacks in both IoT and IoMT environments. AWFOL-IDS integrates Adaptive Weighted Information-Entropy Feature Selection (AWIE-FS) with a Weighted Adaptive Random Forest with Feature-Integrated Optimization (WARF-FIO) classifier with imbalance-aware sampling and weighted decision mechanisms. Experimental evaluation on CICIoT2023 and CICIoMT2024 datasets shows strong performance with 0.9989, 0.9968, and 0.9939 accuracy for 2-class, 8-class, and 34-class tasks and 0.9998, 0.9992, and 0.9856 accuracy for 2-class, 6-class, and 19-class tasks, respectively. Also, evaluation was conducted using 5-fold cross-validation, where findings showed better performance in terms of accuracy, precision, recall, and F-score. In conclusion, AWFOL-IDS is a lightweight and efficient solution for real-world intrusion detection in IoT and IoMT environments.