부산대학교 도서관

This paper opens up a new pathway of the utilityof breathing pattern for user verification. We demonstrate thatit is possible to capture people’s breathing pattern leveragingcommodity WiFi devices. While prior solutions for biometricsbaseduser recognition usually require dedicated devices (e.g.,video cameras or IR sensors), this paper introduces the firstgeneral, low-cost breathing-based user verification system usingcommodity WiFi devices. The proposed system is based onthe fact that the breathing pattern always keeps consistentfor the same user but distinct among different people. Ourinnovative method successfully extracts the breathing patternof different people based on channel state information of WiFisignal to facilitate user verification. The prototype study usingtwo commodity WiFi devices can differentiate people with anaverage verification accuracy over 90%, suggesting that ourbreathing-based user verification system using commerical offthe-shelf (COTS) WiFi is promising to be one of the most criticalmethods in biometrics.