A Self-supervised Framework for Generalized RF Fingerprinting and Zero-shot Anomaly Detection in Dynamic IoT Networks
Keywords:
Domain-informed augmentation, Few-shot learning, RF fingerprinting, Semi-supervised learning, Specific emitter identificationAbstract
Radio frequency (RF) fingerprinting leverages hardware-induced signal imperfections to uniquely identify devices, but conventional supervised deep learning methods degrade sharply under realistic conditions, including few-shot labeled data, open-set environments, channel distortions, temporal drift, modulation diversity, and dynamic operational settings, challenges prevalent in large-scale IoT, UAV swarms, and tactical networks. This work introduces RF-SSL, a semi-supervised framework designed for robust few-shot device identification and zero-shot anomaly detection amid extreme data scarcity and diverse RF impairments. RF-SSL integrates domain-informed augmentation (DIA) simulating channel effects, hardware imperfections, operational variations, and invariance-preserving transforms. It combines multi-task and contrastive self-supervised pretraining on unlabeled RF signals with shallow-to-medium CNN fine-tuning, with hyperparameters optimized for contrastive temperature (τ ≈ 0.1) and learning rate (1×10⁻³). Evaluation spans few-shot accuracy (1–20 shots), zero-shot anomaly AUC, and robustness to SNR, channel models, temporal drift, unseen modulations, convergence speed, and ablation studies. RF-SSL uniquely unifies physics-informed, multi-category augmentation with multi-task contrastive pretraining, providing systematic, multi-dimensional benchmarking under simultaneous data scarcity, drift, and modulation shifts. Results show ~3× faster convergence, 15–40% improved robustness under impairments and drift, and better preservation of performance on unseen modulations (few-shot accuracy up to 0.95, anomaly AUC ~0.88). Ablations confirm DIA and contrastive pretraining as principal contributors, while shallow CNNs optimize speed, accuracy, and robustness. Overall, domain-informed semi-supervised learning significantly enhances data efficiency, temporal stability, and generalization in RF fingerprinting, though absolute performance remains constrained in extreme impairment scenarios.
References
K. Ashton, “That ‘Internet of Things’ thing,” RFID Journal, vol. 22, no. 7, pp. 97–114, 2009.
D. R. Reising, M. A. Temple, and M. J. Mendenhall, “Improving intra-cellular security using air monitoring with RF fingerprints,” in Proc. IEEE WCNC, pp. 1–6, 2010.
O. Ureten and N. Serinken, “Wireless security through RF fingerprinting,” Canadian Journal of Electrical and Computer Engineering, vol. 32, no. 1, pp. 27–33, 2007.
S. Riyaz, K. Sankhe, S. Ioannidis, and K. Chowdhury, “Deep learning convolutional neural networks for radio identification,” IEEE Communications Magazine, vol. 56, no. 9, pp. 146–152, 2018.
K. Sankhe et al., “No radio left behind: Radio fingerprinting through deep learning of physical-layer hardware impairments,” IEEE Transactions on Cognitive Communications and Networking, vol. 6, no. 1, pp. 165–178, 2020.
Y. Xie, Z. Li, and M. Li, “Precise power delay profiling with commodity WiFi,” in Proc. ACM MobiCom, pp. 53–64, 2015.
T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” in Proc. ICML, pp. 1597–1607, 2020.
K. Davaslioglu, S. Boztaş, M. C. Ertem, Y. E. Sagduyu, and E. Ayanoglu, “Self-supervised RF signal representation learning for NextG signal classification with deep learning,” IEEE Wireless Communications Letters, vol. 12, no. 1, pp. 65–69, 2023.
J. Y. Lee, D. H. Park, and S. J. Oh, “Contrastive learning for unsupervised radar extrinsic self-calibration,” in Proc. IEEE ICRA, pp. 5863–5869, 2023.
T. J. O’Shea, T. Roy, and T. C. Clancy, “Over-the-air deep learning based radio signal classification,” IEEE Journal of Selected Topics in Signal Processing, vol. 12, no. 1, pp. 168–179, 2018.
M. Xie, J. Hu, and S. Guo, “Segment-based anomaly detection with approximated sample covariance matrix in wireless sensor networks,” IEEE Transactions on Parallel and Distributed Systems, vol. 26, no. 2, pp. 574–583, 2015.
H. Li, K. Ota, and M. Dong, “Learning IoT in edge: Deep learning for the Internet of Things with edge computing,” IEEE Network, vol. 32, no. 1, pp. 96–101, 2018.
K. Sankhe et al., “ORACLE: Optimized radio classification through convolutional neural networks,” in Proc. IEEE INFOCOM, pp. 370–378, 2019.
W. J. Scheirer, A. de Rezende Rocha, A. Sapkota, and T. E. Boult, “Toward open set recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 7, pp. 1757–1772, 2013.
T. J. O’Shea and J. Hoydis, “An introduction to deep learning for the physical layer,” IEEE Transactions on Cognitive Communications and Networking, vol. 3, no. 4, pp. 563–575, 2017.
B. Schölkopf, J. C. Platt, J. Shawe-Taylor, A. J. Smola, and R. C. Williamson, “Estimating the support of a high-dimensional distribution,” Neural Computation, vol. 13, no. 7, pp. 1443–1471, 2001.
P. Robyns, E. Marin, W. Lamotte, P. Quax, D. Singelée, and B. Preneel, “Physical-layer fingerprinting of LoRa devices using supervised and zero-shot learning,” in Proc. ACM WiSec, pp. 58–63, 2017.
X. Fu et al., “Semi-supervised specific emitter identification via dual consistency regularization,” IEEE Internet of Things Journal, vol. 10, no. 21, pp. 19257–19269, 2023.
C. Liu et al., “Overcoming data limitations: A few-shot specific emitter identification method using self-supervised learning and adversarial augmentation,” IEEE Transactions on Information Forensics and Security, vol. 19, pp. 500–513, 2024.
W. Wang, C. Luo, J. An, L. Gan, H. Liao, and C. Yuen, “Semisupervised RF fingerprinting with consistency-based regularization,” IEEE Internet of Things Journal, vol. 11, no. 5, pp. 8624–8636, 2024.
Y. Wang et al., “Few-shot specific emitter identification via deep metric ensemble learning,” IEEE Internet of Things Journal, vol. 9, no. 24, pp. 24980–24994, 2022.
C. Zhu et al., “Few-shot specific emitter identification via integrated complex variational mode decomposition and spatial attention transfer,” arXiv preprint, 2025.
L. Xie, L. Peng, J. Zhang, and A. Hu, “Radio frequency fingerprint identification for Internet of Things: A survey,” Security and Safety, vol. 3, Art. no. 2023022, 2023.
M. Sun et al., “Few-shot specific emitter identification: A knowledge, data, and model-driven fusion framework,” IEEE Transactions on Information Forensics and Security, vol. 20, pp. 3247–3259, 2025.
J. Zhang, Q. Liu, S. An, and Y. He, “Cross-domain few-shot specific emitter identification via contrastive self-supervised learning,” IEEE Communications Letters, vol. 29, no. 7, pp. 1564–1568, 2025.
S. Yang et al., “Few-shot specific emitter identification via asymmetric dual-path masked autoencoder,” Digital Signal Processing, vol. 163, Art. no. 105201, 2025.
