Replicating the Signature: Unsupervised Targeted Impersonation Attack on RF Fingerprinting

Presents an unsupervised targeted impersonation attack framework that estimates hardware-specific impairments of a victim RF transmitter and synthesizes over-the-air signals that mimic the victim’s RF fingerprint. The method combines unsupervised impairment estimation with signal-processing-based generation, works from different (uncompromised) transceiver hardware, and remains effective in realistic Line-of-Sight and Non-Line-of-Sight deployments. Evaluated on a BLE testbed, the attack yields large improvements over baselines (targeted success rates >80%) and includes analysis of cross-domain