Internet of Things (IoT) security dataset evolution: Challenges and future directions

• Published in: Internet of things (Amsterdam. Online) Vol. 22; p. 100780
• Main Authors: Kaur, Barjinder, Dadkhah, Sajjad, Shoeleh, Farzaneh, Neto, Euclides Carlos Pinto, Xiong, Pulei, Iqbal, Shahrear, Lamontagne, Philippe, Ray, Suprio, Ghorbani, Ali A.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.07.2023
• Subjects: Internet of Things (IoT); Intrusion detection system (IDS); IoT attacks; IoT datasets; IoT security; Machine learning; IoT security; Internet of Things (IoT); Intrusion detection system (IDS); IoT datasets; IoT attacks; Machine learning
• ISSN: 2542-6605; 2542-6605
• DOI: 10.1016/j.iot.2023.100780

The evolution of mobile technologies has introduced smarter and more connected objects into our day-to-day lives. This trend, known as the Internet of Things (IoT), has applications in smart homes, smart cities, industrial automation, health monitoring systems, and has become an essential component of the communication and networking industry. However, different communication and protocol standards, weak security defaults and the difficulty of distributing updates have exacerbated cybersecurity threats to critical applications that employ IoT. To mitigate the threats and counter these attacks, a promising approach is to develop a robust intrusion detection framework specifically aimed at securing IoT. This paper presents our efforts to catalogue and compare attacks, datasets and machine learning algorithms and architectures for intrusion detection systems for IoT devices. We classify attacks aimed at IoT devices at different layers and protocols. This work also highlights potential features that can be used by machine learning-based intrusion detection systems to detect different types of attacks. We provide a comparative study of IoT datasets used for model training and identify key properties which helps in assessing their suitability in particular scenarios. Finally, we discuss our observations and propose the research directions for building a robust IoT intrusion detection system. [Display omitted] •A comparative study of IoT datasets used for different Ml model training;•Comparison of attacks, datasets and algorithms for protecting IoT devices;•Proposing a taxonomy with respect to various IoT attacks;•Identification of potential features to detect different attack types;•Research directions for building a robust IoT intrusion detection system.