Stacked FAN-Based Radio Frequency Fingerprint Identification

With the rapid advancement of Internet of Things (IoT) technology and the large-scale deployment of wireless terminal devices, device authentication has become a critical measure for ensuring IoT security. Radio Frequency Fingerprinting (RFF) technology, based on emitter identification, extracts RFF features strongly correlated with hardware imperfections to achieve precise authentication of wireless devices, thereby enhancing the security of the IoT. This paper introduces an RFF recognition network, IQ-SFAN, which integrates a CNN-based IQ data feature extraction layer with a three-layer stacked FAN (SFAN). This network directly extracts RFF features from raw, unprocessed IQ signals to identify different devices. Experimental results on the public Wisig dataset demonstrate that the IQ-SFAN achieves an identification accuracy of 98.3% at a Signal-to-Noise Ratio (SNR) of 30 dB and 94.5% at 10 dB, indicating robust noise resistance. Furthermore, experiments on the AIS dataset show identification accuracies of 92.9% and 92.3% at SNR levels of 20 dB and 10 dB, respectively, validating the generalization capability of IQ-SFAN.