Abstract
Smartwatches have been seen as the next generation personal devices, after smartphones, due to their use for wide range of applications, e.g., communication, financial applications, social networking, and fitness tracking, etc. These applications generate and store a lot of personal and private data which needs to be protected from unauthorized access. Several authentication schemes based on PIN/password, graphical password have been proposed, however, it is quite difficult to enter or sketch them because of small smartwatch screen. In addition, these schemes still suffer from unsolved security issues. In this paper, we present a novel movement-based behavioral biometric authentication solution for smartwatch, namely, Airsign. Airsign requires smartwatch worn user to write her name in the air. Airsign leverages the arm-movements while the user swings her finger in the air during the process of writing her name. More specifically, Airsign generates arm-movement fingerprint to profile and identify the user. We implemented and evaluated our scheme on Motorola Moto 3G smartwatch. We applied Dynamic Time Warping (DTW), in two settings, and One-class Multilayer Perceptron (OMLP) as classifiers. Airsign has the potential to replace the traditional authentication schemes and be accepted by users.