ML-Based Intrusion Detection System (IDS) Optimization for Cybersecurity in Internet of Things Networks
Open Access DepositedInternet of Things (IoT) enables diverse devices across telecommunications,healthcare, Information Technology (IT), and manufacturing to interconnect for enhanced automation and operational efficiency. However, rapid IoT proliferation has substantially expanded attack surfaces, created sophisticated intrusion vectors and increased cyber- attack vulnerability due to inherent device limitations including constrained processing power, limited storage capacity, and energy resources. These constraints necessitate weaker encryption algorithms and shorter security keys, compromising network security. While edge layer networking addresses processing limitations through proximity-based data centers, it introduces significant financial and operational challenges that many organizations cannot sustain. This research addresses the critical need for an optimized, robust, and scalable intrusion detection system capable of high-speed, high-accuracy attack detection in energy-constrained enterprise IoT environments. BoT-IoT and ToN-IoT datasets were leveraged to develop and evaluate eight machine learning models through a comprehensive comparative analysis. This investigation encompasses four ensemble models (XGBoost, Random Forest, CatBoost, and LightGBM) and four deep learning models (DNN, CNN, RNN, and ANN). Through extensive performance evaluation, it was demonstrated that XGBoost emerges as the optimal solution for energy-constrained enterprise IoT networks. The The proposed model achieves superior performance with over 98% accuracy, minimal classification time, and the lowest false positives, effectively detecting and classifying high-frequency network intrusions while maintaining compatibility with resource-limited IoT infrastructure requirements.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.