View article

Transparent authentication on mobile phones suffers from privacy issues especially when biometric information is involved. In this paper, we propose a solution to address those two issues using the Biohashing algorithm on behavioral information extracted from a mobile phone. The authentication scenario is tested on a dataset composed of 100 users and shows promising results with a 10% EER in the worst case scenario (i.e when protection key is compromised) and a 1% EER in the best case one. In addition, privacy concerns are discussed and experimentally evaluated both in a quantitative and qualitative ways. This opens new perspectives concerning online authentication using smartphone sensing abilities.