Transient-Based Radio Frequency Fingerprinting using Singular Superlet Transform and Deep Learning

Radio Frequency Fingerprinting (RFF) is a technique for identifying wireless devices based on their intrinsic physical characteristics. In the last 10 years, a variety of techniques have been applied to this problem, including the use of Deep Learning (DL) algorithms in combination with Time-Frequency (TF) representations of the signals emitted by the wireless devices. The implementation of RFF using the transient part of the signal has some advantages, but due to the Heisenberg–Gabor uncertainty principle, transients are difficult to localise simultaneously in time and frequency when using classical estimators such as the short-time Fourier transform or the continuous wavelet transform, either optimising the time or frequency resolution or finding a suboptimal compromise. This letter proposes the novel application of DL in combination with the Singular Superlet Transform (SSLT), which has not yet been introduced in the RFF literature. SSLT is a recently introduced TF superresolution representation that consists of increasingly bandwidth-constrainedwavelets and is particularly suitable for transient signals. The original RFF signal is converted into images that are input to a multihead convolutional neural network (CNN). The letter evaluates the SSLT-based approach on two different public RFF datasets for different values of Signal to-Noise Ratio (SNR). The results show that the proposed approach is able to outperform other combinations of DL and TF transforms used in the literature for RFF.

FRANC Dimc;  BALDINI Gianmarco; 

2025-11-24

JOHN WILEY AND SONS INC.

JRC140072

2476-1508 (online),   

https://onlinelibrary.wiley.com/doi/abs/10.1002/itl2.70085,    https://publications.jrc.ec.europa.eu/repository/handle/JRC140072,   

10.1002/itl2.70085 (online),