This repository implements the code proposed by [1]. The implementation is not complete and only covers the device classification, as this is part of a learning process.
The results shown in the Jupyter Notebook are the outcome of a training using only 100 frames from 10 devices, hence the scruffiness and lack of precision. We did not proceed with a larger training due to a lack of hardware and time.
[1] G. Shen, J. Zhang, A. Marshall, and J. Cavallaro. “Towards Scalable and Channel-Robust Radio Frequency Fingerprint Identification for LoRa,” IEEE Trans. Inf. Forensics Security, 2022.