A study of touching behavior for authentication in touch screen smart devices

With the increased popularity of touch screen mobile phones, touch gesture behavior is becoming more and more important. Due to increasing demand for safer access in touch screen mobile phones, old strategies like pins, tokens, or passwords have failed to stay abreast of the challenges. However, we study user authentication scheme based on these touch dynamics features for accurate user authentication. We developed the software needed to collect readings from touch screen of mobile phone running the android operation system. Based on these preliminary experiments we concentrated on the Random Forest classifier to differentiate multiple users. Our results show that combining all features such as touch direction, finger pressure, finger size and acceleration correctly classified touch behavior on an android phone with 98.14% accuracy.