Descripción:
The rapid growth of the Internet of Things (IoT) has increased both the volume of generated data and the potential of new attacks, making intrusion detection a challenge. Hierarchical Federated Learning (HFL) has recently emerged as a promising approach to improve intrusion detection systems (IDS) by combining privacy-preserving decentralized training with scalable multi-layer aggregation. This paper presents a taxonomy of HFL for IDS in IoT networks, organized into five categories: context of application, design strategy, datasets, preprocessing techniques, and problems addressed. The taxonomy highlights current research and shows the key domains, methodologies, and strategies being employed in the field. This work offers a base for future research while identifying open challenges in developing robust and efficient HFL-based IDS for IoT systems.
Tipo de publicación: Conference Paper
Publicado en: Information Technology & Systems
Autores- Navarro-Alfaro, Jonathan
- Lara, Adrian
- Rocha, Álvaro
- Ferrás, Carlos
- Loo, Luis
Investigadores del CITIC asociados a la publicación
Dr. Adrian Lara Petitdemange
Proyecto asociado a la publicación
The rapid growth of the Internet of Things (IoT) has increased both the volume of generated data and the potential of new attacks, making intrusion detection a challenge. Hierarchical Federated Learning (HFL) has recently emerged as a promising approach to improve intrusion detection systems (IDS) by combining privacy-preserving decentralized training with scalable multi-layer aggregation. This paper presents a taxonomy of HFL for IDS in IoT networks, organized into five categories: context of application, design strategy, datasets, preprocessing techniques, and problems addressed. The taxonomy highlights current research and shows the key domains, methodologies, and strategies being employed in the field. This work offers a base for future research while identifying open challenges in developing robust and efficient HFL-based IDS for IoT systems.