Improved wireless security for GMSK-based devices using RF fingerprinting

Gaussian minimum shift keying (GMSK) is employed in current global system for mobile (GSM) networks and remains a viable option for future wireless systems. Physical layer (PHY) security is demonstrated using RF fingerprints from GSM-GMSK signals to thwart malicious activities that are commonly orchestrated through unauthorised use of digital identities. Enhanced security is provided through specific emitter identification by classifying RF fingerprints from near-transient and midamble regions of GSMGMSK bursts using multiple discriminant analysis with maximum likelihood estimation (MDA/ML). While midamble classification performance using instantaneous phase responses is notable, results using GSM-GMSK neartransient fingerprint features are nearly 12% better at SNR = 20 dB and accuracies of 88 94% are achieved at SNR = 6 dB. While transient feature performance is consistent with previous 802.11a results, midamble performance is poorer and various factors are identified as potential contributors to the increased classification challenge posed by GMSK signals.