Classification of RFID Tags with Wavelet Fingerprinting

Abstract

Passive radio frequency identification (RFID) tags lack the resources for standard cryptography, but are straightforward to clone. Identifying RF signatures that are unique to an emitter’s signal is known as physical-layer identification, a technique that allows for distinction between cloned devices. In this work, we study the effect real-world environmental variations have on the physical-layer fingerprints of passive RFID tags. Signals are collected for a variety of reader frequencies, tag orientations, and ambient conditions, and pattern classification techniques are applied to automatically identify these unique RF signatures. We show that identically programmed RFID tags can be distinguished using features generated from DWFP representations of the raw RF signals.