Securing smartphones via typing heat maps

Abstract

Smartphones are the singular most utilised device able to track and monitor our activity continuously from which signatures such as our typing can be detecting. Various situations such as loss of phone, passing on to family members or friends require us to consider continuous modes of passive authentication. This paper explores how typing heat maps can improve user authentication for touch enabled devices. By revisiting the traditional typing digraph intervals, normally associated with typing biometric user authentication, this research investigates the strength of various features for user identification on smartphones. Considering the different methods of typing on a portable small device as opposed to the mechanical keyboard or pressure sensitive tablets, this research takes the touch screen data to present a new set of features which can be used as a passive mode of continuous authentication. Given the continuous use of the device by the owner, a multilayer perceptron neural network is believed to be a plausible and viable approach for this task, which is also understudied for smartphones. The findings in this research highlight the strength of various touch related features in typing signatures.