Removing the RF Fingerprint: A Least Squares Approach to Compensate for a Device's Hardware Impairments

Radio Frequency (RF) Fingerprinting is proposed as a key method of authentication and physical layer security in 5G and beyond networks. The process could classify a device based on the impairments in the received signal that are caused by hardware imperfections in the transmitter. This paper proposes a method that allows a device to add a filter to the transmit chain that removes the RF fingerprint from the transmit signal. We assess the effectiveness of this technique by computing several divergence-based metrics on the RF fingerprints to determine if the fingerprint has been successfully “removed.” Our results indicate that this technique can remove a statistical fingerprint calculated on the received waveform's time domain responses.