An implicit author verification system for text messages based on gesture typing biometrics

Abstract

Gesture typing is a popular text input method used on smartphones. Gesture keyboards are based on word gestures that subsequently trace all letters of a word on a virtual keyboard. Instead of tapping a word key by key, the user enters a word gesture with a single continuous stroke. In this paper, we introduce an implicit user verification approach for short text messages that are entered with a gesture keyboard. We utilize the way people interact with gesture keyboards to extract behavioral biometric features. We propose a proof-of-concept classification framework that learns the gesture typing behavior of a person and is able to decide whether a gestured message was written by the legitimate user or an imposter. Data collected from gesture keyboard users in a user study is used to assess the performance of the classification framework, demonstrating that the technique has considerable promise.

Improving Accuracy, Applicability and Usability of Keystroke Biometrics on Mobile Touchscreen Devices

“…In contrast to these projects, we include touch offsets and touch locations. Burgbacher and Hinrichs [11] trained Support Vector Machines (SVMs) in a "one-vs-rest" scheme to authenticate users via finger movement behaviour on gesture keyboards (see [36]). They reported 0% EER if five or more words in the message are known to the system.…”

“…However, research on keystroke biometrics has either ignored spatial touch-specific typing features on mobile devices [15,33,42,57], or only used such features on tabletops [29,39] or with gesture-keyboards [11].…”

“…The device itself may also be locked with a password or PIN to protect it in cases of loss or theft [25]. Utilising device sensors and assuming a one-toone relationship with personal devices, research proposed behavioural biometrics to enhance security [11,16,18,31]. For password entry, a second, implicit security layer can observe typing behaviour: If an attacker knows the password or PIN, for example due to shoulder surfing [48] or a smudge-attack [2,51], access can still be denied based on the fact that they do not type the password in the same way as the legitimate user (e.g.…”

“…Less work has been carried out on mobile devices with touch [32,42,46,59], and without investigating touch-specific behavioural features, such as those shown in Figure 1. Recent research has successfully used similar features with gesture keyboards [11], but the potential of touch biometrics is still unknown for typing by tapping [20,55]. Hence, to improve keystroke biometrics for the smartphone-era, this paper provides in-depth analyses of mobile-specific and touch-specific opportunities and challenges, leading to the following contributions:…”

Multi-Factor Authentication: A Survey

“…This approach could be easily combined with any text-input authentication methods as a typing pattern is unique for each person [119][120][121]. In case the MFA system is specifically developed for predefined gesture analysis [122], the user is required to replicate a previously learned movement while holding or wearing the sensing device [123][124][125].…”