HoRFFI: High-Openness RF Fingerprint Identification with a Similarity-Enhanced Variational Information Bottleneck

2026-08-05

AI Summary

This work addresses the challenge of identifying newly authorized devices and rejecting unknown ones in highly open scenarios with only a few known device classes available during training. To this end, the authors propose the HoRFFI framework, which integrates similarity-enhanced variational information bottleneck (SVIB), feature-space data augmentation, and clustering-guided similarity learning to construct a transferable and generalizable embedding space, thereby substantially reducing reliance on diverse training categories. Evaluated on public LoRa and Wi-Fi datasets, the method achieves notable improvements, increasing new-class identification accuracy by 0.112 and 0.288, respectively, and boosting AUC scores by 0.029 and 0.060. These results demonstrate its effectiveness in enabling scalable radio frequency fingerprinting and robust open-set rejection.

Abstract

Radio frequency fingerprint identification (RFFI) is a promising technique for wireless device authentication. However, practical RFFI systems must enroll newly authorized devices while rejecting previously unseen ones, even when the feature extractor is trained on only a few labeled base-device classes, giving rise to a high-openness RFFI problem. Existing open-set recognition methods typically rely on feature spaces learned from a large and diverse set of known-device classes, limiting their applicability in practical scenarios. To address this challenge, we propose HoRFFI, a high-openness RFFI framework that supports scalable device identification and unknown-device rejection using only a small number of labeled training devices. HoRFFI employs a similarity-enhanced variational information bottleneck (SVIB)-based supervision mechanism, which reduces the encoder's dependence on training-class diversity and learns a more transferable embedding space. This supervision mechanism uses feature-space augmentation and clustering to derive inter-sample similarity information, which provides supplementary supervision for regularizing the embedding space. Experiments on public LoRa and Wi-Fi datasets show that HoRFFI achieves absolute improvements of \(0.112\) and \(0.288\) in novel-class accuracy, respectively, and corresponding absolute AUC improvements of \(0.029\) and \(0.060\) over the best-performing baselines.

Problem

• Radio Frequency Fingerprint Identification
• High-Openness
• Open-Set Recognition
• Device Authentication
• Unknown Device Rejection

Innovation

• high-openness RFFI
• similarity-enhanced variational information bottleneck
• few-shot open-set recognition
• RF fingerprint identification
• transferable embedding space