User Authentication Method Using Active Acoustic Sensing

Abstract

Personal authentication of the wearable device is important because it has become personalized. Personal identification number (PIN) is generally used to identify the user. However, the user has to remember the PIN and it is easily attacked by shoulder hacking. Therefore, in this study, we propose an authentication method using active acoustic sensing to the user's body. Concretely, the speaker transmits the ultrasonic signal to the user's body and the microphone receives the reflected signal from the body. Each user has a different body composition, and it causes a different frequency response. The system authenticates the user utilizing the difference. Moreover, to improve authentication performance, we add four hand poses to authenticate. Since the proposed method requires one hand for authentication, it is suitable for the wearable computing environment. We implemented the prototype and evaluated the proposed method with nine participants. The evaluation results confirmed that the accuracy of user identification was 91.2% and the equal error rate of user authentication was 2.94% using four hand poses.