Poster: Enhanced Device Identification in Cellular IoT Using IPFIX Records

Ryuta Ohishi, Norihiro Okui, Masataka Nakahara, Masakatsu Nishigaki, Tetsushi Ohki
Network and Distributed System Security (NDSS) Symposium 2025
[ Paper ] [ Web ]

Abstract

With the rapid growth of IoT devices, firmware vulnerabilities and malware infections have become critical concerns, potentially leading to unauthorized access and network breaches. To address this, accurate device identification and authentication are essential. This study proposes a method using payload-free IPFIX records to preserve privacy, combined with deep metric learning for device identification. The approach is validated in a cellular IoT environment with a dataset of 72 device models, leveraging angle-based deep metric learning algorithms.