Machine Learning-Based Intrusion Detection for Smart City Internet of Things Networks

Abstract

The increasing deployment of Internet of Things (IoT) devices in smart city infrastructures has significantly expanded the network attack surface, making effective intrusion detection a critical security requirement. Traditional intrusion detection systems struggle to cope with the volume, heterogeneity, and dynamic behaviour of IoT network traffic, often resulting in high false alarm rates and missed attacks. This study investigates the effectiveness of supervised machine learning techniques for detecting cyberattacks in IoT-based smart city networks using the TON_IoT dataset. A progressive modelling approach is adopted, beginning with Logistic Regression as the baseline model, followed by Random Forest as an ensemble method, and culminating in an optimised XGBoost model. Preprocessing and feature engineering address dimensionality, feature representation, and the challenges associated with imbalanced IoT traffic distributions. Experimental results demonstrate that ensemble and boosting-based models significantly outperform linear approaches. Among the evaluated models, XGBoost achieves the highest detection performance, substantially reducing missed attacks while maintaining robust classification accuracy. The findings demonstrate that advanced ensemble learning combined with robust feature engineering provides a reliable and scalable solution for securing smart city IoT networks.