Pre-correlation and post-correlation RF fingerprinting methods for GNSS spoofer identification with real-field measurement data

Abstract

Radio Frequency fingerprinting (RFF) methods are gaining popularity as physical-layer identification or authentication methods in various navigation and communication applications. Traditionally, RFF has been used in terrestrial communications to identify the genuine transmitters from spoofers and jammers. In recent literature, RFF has gained attention also in the context of satellite navigation and Low Earth Orbit (LEO) satellite communications, though this research area is still in an incipient phase. RFF studies in the context of satellite transmitters (or transceivers) are typically hindered by the challenges in acquiring high-quality raw measurement data. In this paper, we analyze via RFF methods, in both pre-correlation and post-correlation domain, the raw GNSS data collected at three locations: Tampere (Finland), Nottingham (UK), and Nuremberg (Germany). The datasets for the first two scenarios have been collected by the authors, while the third dataset is available in open access. We show that we are able to reach average classification probabilities of spoofer versus GNSS up to 99.99% (i.e., Nuremberg measurements) with pre-correlation data and up to 87.72 % (i.e., Nottingham measurements) with post-correlation data. We also discuss the challenges and limitations of RFF in the context of GNSS.