Improving Reconnaissance Detection in IoT Environments Using Machine Learning with Heterogeneous Data Integration

Reconnaissance attacks, which involve gathering information on network structure and vulnerability information, pose significant risks to Internet of Things (IoT) environments. IoT devices, with their limited resources and diversity, are especially vulnerable, potentially leading to severe threats like Denial of Service (DoS) attacks. Current detection methods often rely on data from single sources, limiting their generalizability and effectiveness in varied IoT scenarios. These approaches often fail to capture the diversity of real-world IoT environments, reducing their suitability for real-time threat detection. This paper presents a novel approach to enhancing reconnaissance detection in IoT by integrating Machine Learning (ML) techniques with heterogeneous data sources. We combine datasets from different IoT environments, specifically CICIoT2023 (home devices) and CICIoMT2024 (medical devices) to create a comprehensive training set that reflects IoT network diversity. Our ML-based model, trained on this integrated dataset, demonstrates a detection accuracy of 96%, highlighting its potential as a scalable, effective, and real-time intrusion detection solution across diverse IoT environments. This research addresses the limitations of existing methods and underscores the importance of using diverse data to strengthen IoT network security against reconnaissance attacks and other cyber threats.