Wavelet Fingerprinting of Radio-Frequency Identification (RFID) Tags

Abstract

Unintentional modulations of the electromagnetic signal of radio-frequency (RF) emitters are used to identify individual sources of signals as unique from emitters of the same type in a procedure known as RF fingerprinting. It allows for the identification and tracking of physical threats, prevention of unauthorized access, and detecting cloning of sensitive devices. Machine learning techniques assist RF fingerprinting by providing automatic recognition of these unique aspects of individual RF emitters. RF identification (RFID) tags are a common RF emitter used to track supplies and are also present in credit cards and passports to allow for automatic recognition or monetary transfers. Despite advances in RFID cryptography, RFID tags can still be easily cloned and tracked. Here, we implement RF fingerprinting to authenticate individual RFID tags at the physical layer. Features are extracted using the dynamic wavelet fingerprint, and supervised pattern classification techniques are used to identify unique RFID tags with up to 99% accuracy.