Abstract

The rapid expansion of the Internet of Things (IoT) ecosystem has provided a tremendous attack surface, and therefore, IoT devices are highly vulnerable to advanced cyberattacks, data breaches, and privacy invasions. Rule-based intrusion detection systems are mostly ineffective in dealing with high-dimensional and heterogeneous traffic streams that IoT environments produce. To fill in these gaps, this research examines the systematic use of Artificial Intelligence (AI) and Machine Learning (ML) methods towards IoT security augmentation, malicious activity detection, blocking of data leakage, and safeguarding of user privacy. A strict methodology, quantitative experimental approach was adopted, leveraging the Australian Centre for Cyber Security's TON_IoT20 dataset of actual network traffic, attack behaviours (i.e., DDoS, data injection, password based intrusions), and normal run log data from various IoT devices such as smart plugs, cameras, and thermostats. Data preprocessing steps involved removal of duplicates, handling of missing values by imputation, feature encoding, and scaling, followed by a 70/15/15 stratified split for training, validation, and test. Three standard ML models, Random Forest (RF), Extreme Gradient Boosting (XGBoost), and a Deep Neural Network (DNN) were used in Python under a controlled Ubuntu environment and trained on the pre-processed data. Model performance was measured by accuracy, precision, recall, F1-score, and ROC-AUC values, with further analysis by means of confusion matrices and McNemar's significance testing. The results indicate that XGBoost performed better, with 98.9% accuracy, 98.6% precision, 99.0% recall, an F1-score of 98.8%, and an ROC-AUC value of 0.996, with very low values of false positives and false negatives. Statistical testing established that the improvement of XGBoost relative to RF and DNN was significant (p < 0.05). In addition, XGBoost provided competitive training time and the quickest inference time, indicating its real-time suitability for IoT intrusion detection applications. All of these results underscore the promise of incorporating AI/ML solutions based on XGBoost in IoT security platforms to improve active threat detection, reduce false alarms, and offer improved privacy protection controls. The research provides an experimentally validated reference model towards further studies and real-world applications of AI-driven intrusion detection systems in real-time IoT environments.

References

• Abubakar, A., Adebayo, A., & Hussain, S. (2023). A comprehensive review of intrusion detection systems in IoT: Techniques and trends. Information Processing & Management, 60(2), 103198. https://doi.org/10.1016/j.ipm.2023.103198
• Abubakar, S., Nwakanma, C., & Bamisile, O. (2023). Gradient boosting ensembles for IoT intrusion detection: Tackling imbalanced datasets. ICT Express, 9(4), 569–577. https://doi.org/10.1016/j.icte.2023.01.005
• Adebayo, A., Olorunfemi, O., & Wang, H. (2023). Statistical evaluation of machine learning models for IoT intrusion detection. Security and Communication Networks, 2023, 1–15. https://doi.org/10.1155/2023/5549132
• Ahmed, S., Patel, N., & Kwon, J. (2023). Intelligent intrusion detection in IoT networks: A systematic review. IEEE Internet of Things Journal, 10(5), 4021–4035. https://doi.org/10.1109/JIOT.2023.3234567
• Al‑Garadi, M. A., Mohamed, A., Al-Ali, A. K., & Du, X. (2023). A comprehensive survey on security and privacy for IoT: Architectures, challenges, and future directions. ACM Computing Surveys, 55(4), 1–41. https://doi.org/10.1145/3540401
• Alqahtani, H., Gumaei, A., & Hassan, M. M. (2023). Hybrid CNN–LSTM model for anomaly detection in industrial IoT. IEEE Access, 11, 38461–38474. https://doi.org/10.1109/ACCESS.2023.3260058
• Barua, S., Gupta, P., & Lim, S. (2023). A comprehensive review of benchmark datasets for IoT intrusion detection. Journal of Network and Computer Applications, 218, 103677. https://doi.org/10.1016/j.jnca.2023.103677
• Bello, O. S., Akanbi, O., & Ganiyu, S. (2022). A comprehensive review of anomaly-based intrusion detection systems in IoT environments. Journal of Cybersecurity and Privacy, 2(3), 548–567. https://doi.org/10.3390/jcp2030028
• Cao, M., Li, Y., & Zhang, Q. (2023). Hybrid intrusion detection frameworks for large-scale IoT systems. Wireless Networks, 29(7), 3373–3387. https://doi.org/10.1007/s11276-023-03340-0
• Chen, Y., Luo, Q., & Wu, T. (2022). Machine learning for cyberattack detection in IoT networks: A survey and future directions. Sensors, 22(18), 6872. https://doi.org/10.3390/s22186872
• Dasgupta, S., Bera, S., & Pradhan, A. (2022). Internet of Things: Concepts, architectures, and future directions. Journal of King Saud University – Computer and Information Sciences, 34(9), 6736–6750. https://doi.org/10.1016/j.jksuci.2021.09.014
• Eze, E. C., Nwankwo, U. N., & Ahmed, M. (2023). Ensemble learning models for intrusion detection in smart IoT environments. International Journal of Information Security, 22(1), 47–61. https://doi.org/10.1007/s10207-022-00666-0
• Feng, X., Liang, J., & Zhang, T. (2024). Global trends and forecasts for IoT device proliferation. Telecommunication Systems, 85(2), 211–228. https://doi.org/10.1007/s11235-023-01127-4
• Ghosh, P., Dhar, T., & Sen, R. (2024). Deep learning approaches for intrusion detection in large‑scale IoT networks. Neural Computing and Applications, 36(4), 2159–2174. https://doi.org/10.1007/s00521-023-09078-6
• Ghosh, S., Chakraborty, A., & Bandyopadhyay, S. (2024). Privacy‑preserving AI for IoT security: A federated learning perspective. Future Generation Computer Systems, 155, 339–352. https://doi.org/10.1016/j.future.2024.01.006
• Gupta, R., Chatterjee, M., & Yadav, S. (2023). Scalable machine learning frameworks for big‑data‑driven cybersecurity. Information Sciences, 633, 235–249. https://doi.org/10.1016/j.ins.2023.05.018
• Haque, M. A., Ferdousi, S., & Alsaif, K. (2023). Evolution of AI‑driven intrusion detection: Opportunities and challenges. Security and Privacy, 6(2), e246. https://doi.org/10.1002/spy2.246
• Hossain, M. A., & Pervez, M. H. (2023). Feature selection and dimensionality reduction for IoT network anomaly detection. Wireless Communications and Mobile Computing, 2023, 1–14. https://doi.org/10.1155/2023/9975523
• Huang, Y., Chen, Z., & Zhao, W. (2023). Edge intelligence for IoT: A survey of trends and challenges. IEEE Internet of Things Journal, 10(12), 10345–10361. https://doi.org/10.1109/JIOT.2023.3247256
• Hussain, F., Wang, J., & Malik, N. (2022). An ensemble machine learning approach for effective intrusion detection in IoT networks. Sensors, 22(3), 1124. https://doi.org/10.3390/s22031124
• Jamal, M., Rawat, S., & Kaur, R. (2022). Boosting algorithms for IoT anomaly detection: A comparative analysis. Neural Computing and Applications, 34(21), 18485–18498. https://doi.org/10.1007/s00521-022-07285-x
• Khalid, M., Raza, M., & Alghazzawi, D. (2023). Network-based intrusion detection solutions for IoT gateways. Journal of Information Security and Applications, 74, 103404. https://doi.org/10.1016/j.jisa.2023.103404
• Khan, S., Rehman, A., & Alazab, M. (2024). Towards next-generation intrusion detection for IoT: A hybrid deep learning approach. Future Generation Computer Systems, 151, 30–45. https://doi.org/10.1016/j.future.2023.09.014
• Khan, T., Bashir, A., & Ahmed, R. (2023). Anomaly detection in IoT networks: A machine learning perspective. Computer Communications, 204, 1–15. https://doi.org/10.1016/j.comcom.2023.06.001
• Li, W., Tang, M., & Zhao, Q. (2024). Edge-enabled AI for real-time IoT security: Recent advances and challenges. IEEE Transactions on Industrial Informatics, 20(3), 1804–1816. https://doi.org/10.1109/TII.2023.3311895
• Lin, P., & Yu, J. (2022). Security challenges in IoT architectures: A comprehensive analysis. Computers & Electrical Engineering, 101, 108075. https://doi.org/10.1016/j.compeleceng.2022.108075
• Liu, P., Zhong, H., & Wang, R. (2023). Deep learning-based anomaly detection in large-scale networks using ensemble strategy. Computers & Security, 127, 103155. https://doi.org/10.1016/j.cose.2023.103155
• Luo, Z., Lin, J., & Hu, X. (2024). Lightweight feature selection for real-time intrusion detection in IoT gateways. Future Generation Computer Systems, 149, 250–263. https://doi.org/10.1016/j.future.2023.10.088
• Mehta, D., Patel, A., & Joshi, M. (2023). AI‑enabled automation in cybersecurity: Toward self‑healing networks. IEEE Access, 11, 103765–103779. https://doi.org/10.1109/ACCESS.2023.3302331
• Miah, M. S., Kabir, M. A., & Han, J. (2023). Improved intrusion detection in IoT networks through feature optimization. ICT Express, 9(3), 450–459. https://doi.org/10.1016/j.icte.2023.03.008
• Mishra, S., Sahu, A. K., & Panda, A. (2022). Host-centric intrusion detection strategies for resource-constrained IoT devices. IEEE Sensors Journal, 22(17), 16874–16883. https://doi.org/10.1109/JSEN.2022.3180940
• Nandhini, S., Rajalakshmi, K., & Bose, S. (2024). A paradigm shift in cyber defense: Machine learning approaches and future directions. Expert Systems with Applications, 233, 120934. https://doi.org/10.1016/j.eswa.2023.120934
• Okafor, B., Arshad, J., & Madanian, S. (2023). AI‑enabled intrusion detection for IoT: Challenges and research opportunities. ACM Computing Surveys, 56(7), 1–30. https://doi.org/10.1145/3590907
• Okafor, L., Madueke, S., & Ekpo, C. (2023). Scalability challenges of intrusion detection systems in IoT environments: A review. Journal of Network and Computer Applications, 221, 103624. https://doi.org/10.1016/j.jnca.2023.103624
• Pahlavan, M., Sadeghi, A., & Ghiasi, M. (2023). Security challenges in IoT ecosystems and emerging AI solutions: A review. Computers & Security, 126, 103087. https://doi.org/10.1016/j.cose.2022.103087
• Rajput, A., Dhamdhere, K., & Pande, S. (2024). Privacy-preserving intrusion detection using federated learning in IoT. Future Internet, 16(2), 54. https://doi.org/10.3390/fi16020054
• Rashid, F., Chen, L., & Bhuiyan, M. (2022). Unsupervised learning for anomaly detection in evolving cybersecurity landscapes. Journal of Information Security and Applications, 69, 103237. https://doi.org/10.1016/j.jisa.2022.103237
• Roy, R., Mondal, S., & Banerjee, I. (2023). A layered perspective on IoT system architecture and security. Information Systems Frontiers, 25(3), 871–889. https://doi.org/10.1007/s10796-022-10344-7
• Saha, D., Choudhury, T., & Debnath, B. (2023). Ensemble-based anomaly detection for IoT traffic. Sensors, 23(4), 2047. https://doi.org/10.3390/s23042047
• Sahu, P., Tripathi, R., & Saxena, N. (2023). Evaluating machine learning algorithms on benchmark IoT intrusion datasets: A comparative analysis. Journal of Information Security and Applications, 74, 103468. https://doi.org/10.1016/j.jisa.2023.103468
• Sankaran, S., Paul, R., & Devi, S. (2023). Hybrid IDS frameworks for heterogeneous IoT systems. Computers & Security, 129, 103396. https://doi.org/10.1016/j.cose.2023.103396
• Shah, R., Agarwal, R., & Park, J. (2024). Deep learning-based intrusion detection for IoT with attention mechanisms. Computers & Security, 134, 103505. https://doi.org/10.1016/j.cose.2023.103505
• Singh, H., & Rajesh, M. (2023). Enhancing anomaly detection in IoT using ensemble machine learning approaches. Journal of Network and Computer Applications, 210, 103617. https://doi.org/10.1016/j.jnca.2022.103617
• Sujatha, R., & Thomas, J. (2023). Low‑rate DDoS attack detection in IoT using autoencoder‑GRU hybrid models. Computer Networks, 226, 109687. https://doi.org/10.1016/j.comnet.2023.109687
• Tian, H., Li, Z., & Xu, Y. (2023). Comparative study on machine learning-based IoT intrusion detection models. Expert Systems with Applications, 215, 119254. https://doi.org/10.1016/j.eswa.2022.119254
• Tian, S., Yu, Z., & Cheng, F. (2024). Explainable AI in cybersecurity: A review of methods and challenges. ACM Computing Surveys, 56(8), 1–34. https://doi.org/10.1145/3608930.
• Vekaria, J., Patel, K., & Shukla, R. (2023). Impact of data preprocessing on IoT intrusion detection performance: A case study. Wireless Networks, 29(6), 2187–2203. https://doi.org/10.1007/s11276-023-03323-8.
• Velmurugan, S., Reddy, A. K., & Kumar, B. V. (2024). AI-driven intrusion detection frameworks for IoT: A comprehensive survey. Engineering Applications of Artificial Intelligence, 132, 107924. https://doi.org/10.1016/j.engappai.2023.107924
• Wang, Y., Chen, G., & Qiu, L. (2024). Adaptive online learning strategies for evolving IoT security threats. ACM Transactions on Internet Technology, 24(1), 15. https://doi.org/10.1145/3620094.
• Wu, K., Han, Y., & Li, F. (2024). Emerging security threats in large-scale IoT ecosystems: A case study. Future Internet, 16(1), 12. https://doi.org/10.3390/fi16010012.
• Xu, L., Liu, X., & Zhang, Y. (2023). Lightweight gradient boosting for IoT intrusion detection. Knowledge-Based Systems, 266, 110478. https://doi.org/10.1016/j.knosys.2023.110478
• Zhang, K., Wang, L., & Sun, Y. (2024). Balancing accuracy and efficiency in IoT security with edge-based machine learning. IEEE Transactions on Network and Service Management, 21(2), 1609–1621. https://doi.org/10.1109/TNSM.2023.3332107.
• Zhao, H., Yu, P., & Wang, K. (2024). Transformer-based network intrusion detection for industrial IoT. IEEE Transactions on Industrial Informatics, 20(1), 541–552. https://doi.org/10.1109/TII.2023.3329110.