Biometric swiping on touchscreens

Abstract

Touchscreen devices have become very popular in the last decade and eased our modern life. It is now possible to automatically log in to any web page connected to our touchscreen phones, such as social networks, e-commerce sites and even mobile banking. Given these facts, the emerging touchscreen technology brings out a potential security issue: weakness of authentication protocols. Therefore, we put forward a biometric enhancement on “swiping” authentication, which is one of the options to log in a touchscreen phone however with the lowest security. We created a ghost password by extracting the features of coordinates and swipe durations to use them as the inputs of the Levenberg-Marquardt based neural network and adaptive neuro-fuzzy classifiers which both discriminate real attempts from fraud attacks after training.