Remote Authentication Using an Ultra-Wideband Radio Frequency Transceiver

Abstract

In this paper, we propose three new authentication methods using ultra-wideband (UWB) radio frequency (RF) signals. These mid-air-based authentication techniques are different from the conventional touch-based authentication approaches and provide a more convenient and safer user experience. In these methods, the impulse radio UWB transceiver receives the reflected RF signal corresponding to the movement pattern of the user, extracts the pattern information after noise removal, and converts it into image data suitable for use as classifier input through image signal processing. After that, it verifies whether the pattern is the authentication pattern of the user by employing a convolutional neural network and a one-class support vector machine. This report describes the three proposed authentication methods using the advantages of UWB RF signals, as well as their experimental verification. In the first authentication method, ShapeSec, authentication is performed by writing a simple shape pattern in the air. This technique is simple but very convenient. The second method, SignSec, correctly detects the handwriting of different users when signing and overcomes the security vulnerability of signing with the existing pens. Finally, BreatheSec is based on breathing patterns and takes advantage of the ability of a UWB RF transceiver to detect minute breath movements. The authentication pattern could not be imitated even after it was watched from the side, proving that the proposed method has excellent security characteristics. Experiments showed that each method has over 95% accuracy.