Behavioural and Face Biometric Based Authentication for Mobile Applications

Abstract

The usage of smartphones and other mobile devices like tablets are gradually increased over the past decades. With the privilege of using smartphones, reliable user authentication on mo-bile devices has become a crucial requirement. Many authentication strategies have been pro-posed as alternatives to traditional password protection approach. Shape-based authentication methods like password pattern and biometric based authentication methods such as face bio-metric and fingerprint are some of the popular alternatives among them. However password pattern approach is considered to be the most popular authentication method which is accepted by many users.In terms of security, password pattern is rather weak since the pattern can be easily stolen and reproduce. Thus, this project propose an implicit authentication approach that enhances the security of password pattern by adding additional security layer. the secondary layer will use a combination of face features and touch behaviours to identify the legitimate users. This study was conducted by using an Android application to collect data from user input on the touch screen. This data collection is done in two consecutive steps as face features and username is collected in the first step and touch behaviour in the second step.Data analysis and user verification are done by using a statistical method which uses confi-dence interval calculated by using the sample mean and standard deviation to identify legitimate users. The final decision is made based on a pre-calculated threshold values which are 0.7 as the face threshold and 0.46 as the touch threshold value. Finally the this method could achieve 76.11% accuracy rate with 91.67% success rate, 8.33% false rejection rate and 39.44% false acceptance rate.