The Power of Precision: High-Resolution Backscatter Frequency Drift in RFID Identification

Physical-layer identification uses manufacturing variations to create unique identifiers for each device. A decade ago, this concept was applied to RFID tags using backscatter frequency drift (BFD), a specific kind of ‘fingerprint’ determined by the difference between the actual backscatter signal received and the expected backscatter link frequency (BLF). However, BFD has been undervalued due to its low performance in tag identification, achieving less than 30% accuracy. In this study, we reevaluate BFD, focusing on the issue of frequency resolution as the cause of its poor performance. The problem doesn't lie in the BFD's uniqueness, but in the inferior way we measure the frequency of a backscatter signal, which is limited by the current air interface protocol. This situation is akin to trying to identify human fingerprints using low-quality imaging. We propose a practical solution to improve the frequency resolution from kilohertz to sub-hertz, without requiring hardware or protocol changes. Our findings show that this high-resolution BFD approach significantly enhances the distinguishability to 99.4% and the identification accuracy to 94% when tested on a dataset of 7,135 RFID tags across nine models.