Itype: Using Eye Gaze to Enhance Typing Privacy

Zhenjiang Li,Mo Li,Prasant Mohapatra,Jinsong Han,Shuaiyu Chen

DOI: https://doi.org/10.1109/infocom.2017.8057233

2017-01-01

Abstract:This paper presents iType, a system that uses eye gaze for typing private information on commodity mobile platforms. The design combats three primary challenges: 1) relatively low accuracy of mobile gaze tracking; 2) difficulties in correcting input errors due to lacking the comparison with the true text-entry value; and 3) device motions and other noises that may interfere gaze tracking accuracy and thus the iType performance. We devise a set of effective techniques, including leveraging a collective behavior of the gaze tracking results, unique correlation of the typing error spatial distributions, and motion sensor hints from mobile devices, to address above challenges. A set of enhancement techniques are applied to further improve iType's robustness and reliability. We consolidate above designs and implement iType on iOS platform. Evaluations show that iType achieves high keystroke detection accuracy for the secure typing within a reasonable short latency.