CueAuth

Abstract

Secure authentication on situated displays (e.g., to access sensitive information or to make purchases) is becoming increasingly important. A promising approach to resist shoulder surfing attacks is to employ cues that users respond to while authenticating; this overwhelms observers by requiring them to observe both the cue itself as well as users' response to the cue. Although previous work proposed a variety of modalities, such as gaze and mid-air gestures, to further improve security, an understanding of how they compare with regard to usability and security is still missing as of today. In this paper, we rigorously compare modalities for cue-based authentication on situated displays. In particular, we provide the first comparison between touch, mid-air gestures, and calibration-free gaze using a state-of-the-art authentication concept. In two in-depth user studies (N=37) we found that the choice of touch or gaze presents a clear trade-off between usability and security. For example, while gaze input is more secure, it is also more demanding and requires longer authentication times. Mid-air gestures are slightly slower and more secure than touch but users hesitate to use them in public. We conclude with three significant design implications for authentication using touch, mid-air gestures, and gaze and discuss how the choice of modality creates opportunities and challenges for improved authentication in public. CCS Concepts: • Security and privacy → Authentication; • Human-centered computing → Human computer interaction (HCI); Interaction techniques; Interaction design;

The Role of Eye Gaze in Security and Privacy Applications: Survey and Future HCI Research Directions

“…Several works used smooth pursuit eye movements to allow users to enter 4-digit PINs on a public display [27,80,83]. Other researchers explored the augmentation of graphical password schemes by using gaze input.…”

“…This method has been adopted by several unimodal authentication schemes [20,33,44]. An alternative is to detect certain gaze behaviors that would indicate input, such as gaze gestures [36] as done in EyePassShapes [32], or smooth pursuit eye movements [27,83,126,127,164].…”

“…The user's task is to enter a password provided by the experimenter. The password could be verbally communicated to the user [79] or read by a text-to-speech system [83]. Requiring the participant to read passwords from a piece of paper or a screen [82] is not recommended for gazebased authentication schemes as it might impact the tracked gaze behavior.…”

Just gaze and wave

“…To evaluate the observation resistance, we measured the Levenshtein distance between the guesses and the correct password to analyze how close the guess is to the correct password. The Levenshtein distance refers to the distance between the attackers' guesses and the correct password; it is a commonly used metric in security analysis that reveals how close a guess is to the original password [13,21,25]. Thus, Levenshtein distance was the dependent variable.…”

“…While recent work compared multiple modalities for cue-based authentication [25], a comparison of multimodal authentication approaches is missing. To close this gap, we report on 6 concepts:…”