Authors: Maha M. Althobaiti *
Affiliations:
Department of Computer Science, College of Computing and Information Technology, Taif University, Taif, Saudi Arabia
Machine learning (ML) plays a key role in intrusion detection systems (IDS) and Internet of Things (IoT) security by improving the ability of cyber-physical systems (CPSs) to resist attacks from malicious users. CPSs combine physical components with networking and communication technologies to ensure safe and efficient operations. However, attackers often try to disrupt or disable the computing resources of these systems. This paper presents a new ML-based IDS framework designed for CPSs. To develop this framework, an open-source dataset containing different types of cyberattacks and related detection features was used. The dataset was labeled and preprocessed to make it clean, balanced, and suitable for training ML models. Preprocessing steps included handling missing values, normalizing features, and balancing the class distribution. Two ML algorithms—Random Forest (RF) and Stochastic Gradient Descent (SGD)—were applied to build and train classification models for intrusion detection. The experimental results showed that the RF model achieved a high accuracy of 99.5%, outperforming the SGD model, which reached 93.6% accuracy. In addition to accuracy, model performance was also measured using precision, recall, and F1 score. The results demonstrate that the proposed IDS is effective in detecting cyberattacks and improving IoT security. It offers a scalable and reliable solution for protecting CPS environments. This research contributes to the development of more secure CPSs by enhancing the trustworthiness, robustness, and flexibility of IoT systems.
Intrusion detection, Cyber-physical systems, Machine learning, IoT security, Classification models
https://doi.org/10.21833/ijaas.2025.06.009
Althobaiti, M. M. (2025). Intelligent intrusion detection for IoT and cyber-physical systems using machine learning. International Journal of Advanced and Applied Sciences, 12(6), 92–105. https://doi.org/10.21833/ijaas.2025.06.009