Cross-device tracking through identification of user typing behaviours
A novel method of cross-device tracking based on user typing behaviours is presented. Compared with existing methods, typing behaviours can offer greater security and efficiency. When people type on their devices, a number of different factors may be considered to identify users, such as the angle and distance of contact point to the centre of the target character, the time elapsed between two typing actions and the physical force exerted on the device (which can be measured by an accelerometer). An experiment was conducted to validate the proposed model; those data are collected through an Android App developed for the purpose of this study. By collecting a reasonable amount of this type of data, it is shown that machine learning algorithms can be employed to first classify different users and subsequently authenticate users across devices.