Luqman, Muhammad and Zeeshan, Muhammad and Riaz, Qaiser and Hussain, Mehdi and Tahir, Hasan and Mazhar, Noman and Khan, Muhammad Saffeer (2025) Intelligent parameter-based in-network IDS for IoT using UNSW-NB15 and BoT-IoT datasets. Journal of the Franklin Institute, 362 (1). p. 107440. ISSN 0016-0032, DOI https://doi.org/10.1016/j.jfranklin.2024.107440.
Full text not available from this repository.Abstract
Keeping in mind the ubiquity of Internet of Things devices and the heterogeneity of protocols, intruders can easily meddle in an IoT network and jeopardize its security. This work proposes an intelligent intrusion detection system for IoT networks by applying machine learning algorithms, namely Random Forest and Support Vector Machine. We achieved high efficiency and reduced overfitting by balancing 11 classes in the BoT-IoT dataset. The dataset was prepared using preprocessing techniques that include handling missing values, one-hot encoding, and scaling the features for both binary and multi-class classification. An important feature selection and extraction process was used for RF and SVM model training with two hyperparameter tuning techniques. The proposed model achieves efficient training times for SVM and RF with three cross-validation techniques using 10 folds. In machine learning, an accuracy of 99.60% for binary and 98.31% for multi-class classification was achieved. For deep learning algorithms, the abnormal traffic in the UNSW BoT-IoT dataset consists of user datagram protocol and transmission control protocol packets. To address this imbalance, we merged two important network datasets, UNSW BoT-IoT and NB-15, using feature engineering based on common features, and used the deep learning model LSTM for training. After applying preprocessing techniques and important feature extraction, we performed multiple iterations with 10 and 16 epochs for model validation and achieved an accuracy of 99.89% and 99.97% for multi-class classification.
| Item Type: | Article |
|---|---|
| Funders: | UNSPECIFIED |
| Uncontrolled Keywords: | In-network IDS; Internet of Things (IoT) security; BoT-IoT; NB-15; Real-time intrusion detection; Cyber Threat Intelligence (CTI) |
| Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science |
| Divisions: | Faculty of Computer Science & Information Technology |
| Depositing User: | Ms. Juhaida Abd Rahim |
| Date Deposited: | 04 Mar 2025 02:57 |
| Last Modified: | 04 Mar 2025 02:57 |
| URI: | http://eprints.um.edu.my/id/eprint/47800 |
Actions (login required)
 |
View Item |
