Lightweight Preprocessing and Feature Extraction for LoRa RF Fingerprint Identification

Bothereau, E, Gerzaguet, R, Gautier, M, Zhang, J ORCID: 0000-0002-3502-2926, Chillet, A, Marshall, A ORCID: 0000-0002-8058-5242 and Berder, O (2026) Lightweight Preprocessing and Feature Extraction for LoRa RF Fingerprint Identification In: ICC 2026 - IEEE International Conference on Communications, 2026-5-24 - 2026-5-28.

Text ICC 2026 LoRa RFFI.pdf - Author Accepted Manuscript Available under License Creative Commons Attribution. Download (1MB) | Preview

Abstract

Deep learning-based radio frequency fingerprint identification (RFFI) achieves strong performance but suffers from high computational demands, limiting its suitability for resource-constrained and real-time scenarios. In this work, we presented a lightweight preprocessing method and machine learning-based framework for LoRa RFFI. In particular, our preprocessing method exploits the instantaneous phase difference of the LoRa signal, enabling fine spectral characterization and revealing residual hardware-dependent patterns. A three-stage framework involves feature extraction training, enrollment and authentication. The feature extraction training is machine learning-based and achieved by extremely randomized trees and linear discriminant analysis. The enrollment is done by distribution matching and the authentication is performed by calculating their probability density. We carried out extensive evaluation using a public LoRa RFFI dataset and compared with the state-of-the-art (SOTA) deep learning-based approach. Our proposed method achieved a comparable classification accuracy and channel robustness with the SOTA approach. In addition, our method required significantly less computation and memory, resulting in an explicit and efficient model across all evaluated scenarios.

Item Type:	Conference Item (Unspecified)
Uncontrolled Keywords:	4605 Data Management and Data Science, 46 Information and Computing Sciences, 4611 Machine Learning, Networking and Information Technology R&D (NITRD), Machine Learning and Artificial Intelligence
Divisions:	Faculty of Science & Engineering Faculty of Science & Engineering > School of Computer Science & Informatics Faculty of Science & Engineering > School of Computer Science & Informatics > Trustworthy Computing
Depositing User:	Symplectic Admin
Date Deposited:	09 Feb 2026 10:08
Last Modified:	19 Sep 2026 02:30
DOI:	10.1109/ICC59461.2026.11587750
Related Websites:	Publisher
URI:	https://livrepository.liverpool.ac.uk/id/eprint/3196888
Disclaimer:	The University of Liverpool is not responsible for content contained on other websites from links within repository metadata. Please contact us if you notice anything that appears incorrect or inappropriate.