Trust‐Aware RF Fingerprinting for Open‐Set Radio Unit Authentication in Open Radio Access Network
Radio Science · 2026-11 · 61 (11)
DOI: 10.1029/2026rs008790 ↗Kurz gefasst
Fachartikel in der Zeitschrift Radio Science, online erschienen am 18.11.2026. Verfasst von G. Kavitha und 2 weiteren Autor:innen. Die Arbeit nennt 44 Literaturangaben.
Abstract (Originalsprache)
Open Radio Access Network (O‐RAN) has emerged as a key architecture for next‐generation wireless communication. However, its disaggregated design increases vulnerability to False Base Station (FBS) attacks that allow an unauthorized Open Radio Unit (O‐RU) to impersonate a legitimate network entity. Existing Radio Frequency (RF) fingerprinting approaches usually treat authentication as a closed‐set classification task with a cross‐entropy optimization that limits their ability to reject unknown devices. Moreover, classifier confidence alone may produce overconfident predictions for unknown radios, while wireless noise and replay‐based signal manipulation can further degrade authentication reliability. This study proposes a trust‐aware open‐set RF fingerprint authentication framework for secure O‐RU verification that combines prototype‐based embedding learning and zero‐trust authentication. The local RF hardware distortions and long‐range temporal dependency are jointly captured by a lightweight hybrid 1D Convolutional Neural Network‐Transformer encoder. A Trust‐Adaptive Prototype Loss (TAP‐Loss) is proposed to learn compact intra‐class, inter‐class separability, and confidence‐aware prototype learning simultaneously. The framework is evaluated using the POWDER RF Fingerprinting Data set, In each fold, three transmitters are used as known O‐RUs, while the remaining transmitter is reserved as the unseen O‐RU, under AWGN and DSP‐based replay attack conditions. Experimental results show an unknown device detection rate of 98.07 ± 0.75%, a false acceptance rate of 1.82 ± 0.31, and a false rejection rate of 1.67 ± 0.28, demonstrating that the proposed method is robust and reliable for FBS detection and provides a secure O‐RAN authentication under realistic adversarial wireless conditions.
Zitieren
APA
Kavitha, G., Abarnaswara, R., & Sugitha, G. (2026). Trust‐Aware RF Fingerprinting for Open‐Set Radio Unit Authentication in Open Radio Access Network. Radio Science, 61(11). https://doi.org/10.1029/2026rs008790
Vancouver
Kavitha G, Abarnaswara R, Sugitha G. Trust‐Aware RF Fingerprinting for Open‐Set Radio Unit Authentication in Open Radio Access Network. Radio Science. 2026;61(11). doi:10.1029/2026rs008790
BibTeX
@article{Kavitha2026,
author = {Kavitha, G. and Abarnaswara, R. and Sugitha, G.},
title = {Trust‐Aware RF Fingerprinting for Open‐Set Radio Unit Authentication in Open Radio Access Network},
journal = {Radio Science},
year = {2026},
volume = {61},
number = {11},
doi = {10.1029/2026rs008790}
}
EUVIMED