Scholarlink

<p>The enormous growth in the number of Internet of Things (IoT) environments has resulted in greater exposure to cyberattacks (Denial of Service (DoS), Distributed Denial of Service (DDoS), and malicious software attacks among others), necessitating effective and intelligent intrusion detection systems (IDS). This work proposes DTXGRF, a lightweight ensemble-based IDS that combines Decision Tree, Random Forest, and XGBoost using a soft voting strategy to improve approach detection performance in IoT networks. The proposed model is examined on two recent benchmark datasets, CIC-IoT2023 and IoTID20, providing comparison against state-of-the art machine learning classifiers. Experimental results show that the ensemble model achieves superior and balanced performance, with an accuracy of 95.03% on CIC-IoT2023 and near-perfect performance on IoTID20, along with high precision, recall, F1-score, and ROC-AUC values. The performance improvements are due to the ability of the combined model to learn complementary decision boundaries which increase robustness and reduce variance. Moreover, the cross-validation results indicate that the proposed approach leads to a stable and generalizable model. These findings demonstrate that the DTXGRF model is an effective and scalable solution for real-time intrusion detection in resource-constrained IoT environments. Moreover, the proposed ensemble model is well suited for real-time intrusion detection and can be efficiently deployed in parallel or high-performance computing (HPC) environments due to its scalable and modular architecture.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intrusion detection system-based ensemble machine learning to improve IoT network against cyber attacks

  • 期刊论文

  • 发表日期：

    2026年4月22日

查看全文
  • Shayma Wail Nourildean,
  • Wafa Mefteh,
  • Ali Mohsen Frihida

摘要

The enormous growth in the number of Internet of Things (IoT) environments has resulted in greater exposure to cyberattacks (Denial of Service (DoS), Distributed Denial of Service (DDoS), and malicious software attacks among others), necessitating effective and intelligent intrusion detection systems (IDS). This work proposes DTXGRF, a lightweight ensemble-based IDS that combines Decision Tree, Random Forest, and XGBoost using a soft voting strategy to improve approach detection performance in IoT networks. The proposed model is examined on two recent benchmark datasets, CIC-IoT2023 and IoTID20, providing comparison against state-of-the art machine learning classifiers. Experimental results show that the ensemble model achieves superior and balanced performance, with an accuracy of 95.03% on CIC-IoT2023 and near-perfect performance on IoTID20, along with high precision, recall, F1-score, and ROC-AUC values. The performance improvements are due to the ability of the combined model to learn complementary decision boundaries which increase robustness and reduce variance. Moreover, the cross-validation results indicate that the proposed approach leads to a stable and generalizable model. These findings demonstrate that the DTXGRF model is an effective and scalable solution for real-time intrusion detection in resource-constrained IoT environments. Moreover, the proposed ensemble model is well suited for real-time intrusion detection and can be efficiently deployed in parallel or high-performance computing (HPC) environments due to its scalable and modular architecture.