Comparison of PIN- and pattern-based behavioral biometric authentication on mobile devices

Personal identification numbers (PIN) and unlock patterns are highly popular authentication mechanisms on smart mobile devices but they are not sufficiently secure. PIN or pattern mechanisms enhanced by additional, implicit behavioral biometric authentication can offer stronger authentication assurance while preserving usability, therefore becoming very attractive. Individual studies on PIN- and pattern-based behavioral biometric authentication on smartphones were conducted but their results cannot be directly compared. In this work, we present a comparison study on the authentication accuracy between PIN-based and pattern-based behavioral biometric authentication using both smartphone and tablet. We developed a uniform framework for both PIN-based and pattern-based schemes and used two representative methods-Histogram and DTW-for user verification. We recruited 15 users and collected behavioral biometric data for both simple and complex PINs and patterns. Our experimental results show that PIN-based and pattern-based behavioral biometric authentication schemes can achieve about the same level of accuracy but not all verification methods are equal. The Histogram method can achieve more consistent results and handle template aging better than the DTW method based on our results. Our findings are expected to shed light on the exploration and analysis of effective behavioral biometric verification methods and facilitate more comprehensive investigation on behavioral biometric authentication for mobile devices.