Radio Frequency Fingerprint Extraction and Identification for Wi-Fi Signals in Noisy Channels

A noise-robust radio frequency fingerprint (RFF) extraction and identification scheme is proposed and demonstrated for Wi-Fi device recognition. Specifically, an RFF extractor pre-trained by supervised contrastive learning extracts the distinctive features from the spectrogram, which is produced by Fourier-based sychrosqueezing transform (FSST). Then a neural network-based classifier is adopted for precise device identification with the extracted RFF. Utilizing 11 commercial wireless network interface controllers (WNICs), the experimental results demonstrate the significant performance improvement for noisy channels, with classification accuracy achieving up to 95.7 ^% and 86% when the signal-to-noise ratio (SNR) is 10 and 5 dB, respectively.