Movement based biometric authentication with smartphones

The widespread use of small, mobile computing devices such as smartphones increases the need to protect these devices and the sensitive data they contain against unauthorized use. Authentication solutions based on biometrics are a promising way to replace common mechanisms relying on personal identification numbers or passwords, which are often perceived as inconvenient by users. Fingerprint based authentication schemes have been introduced to the latest generations of smartphones. Fingerprints are easy to record and they show low intra-class variation. However, fingerprints or fingerprint templates must be safely stored on the phone and not transmitted to other data bases. Since fingerprints are linked to users for life, the damage caused by compromised fingerprint data is likely to be very significant and potentially permanent for users. Here we present a proof-of-principle demonstration for movement based biometrics with built-in smartphone motion sensors. Such dynamic biometric approaches have a lower potential for misuse than fingerprints while being fast and user friendly. Using an approach based on the implementation of a dynamic time warping algorithm we find that in a feasible application scenario a 0.02% false acceptance rate at a 10% false rejection rate can be achieved (equal error rate about 3%). We also investigate the threat posed by skilled forgeries, where access to a detailed video recording of the original user movement is provided. Furthermore, forgers were asked to practice the movement until they feel comfortable executing it. This lead to an increase in the equal error rate to 9%. A similarly detailed video recording of a user entering a password would likely result in a nearly 100% equal error rate. These proof-of-principle results demonstrate that movement based authentication with smartphones holds significant potential but must be improved further to serve as a stand-alone authentication method.