Fine-Grained and Context-Aware Behavioral Biometrics for Pattern Lock on Smartphones

Abstract

Pattern lock has been widely used in smartphones as a simple and effective authentication mechanism, which however is shown to be vulnerable to various attacks. In this paper, we design a novel authentication system for more secure pattern unlocking on smartphones. The basic idea is to utilize various behavior information of the user during pattern unlocking as additional authentication fingerprints, so that even if the pattern password is leaked to an attacker, the system remains safe and protected. To accommodate a variety of user contexts by our system, a context-aware module is proposed to distinguish any of such contexts (e.g., body postures when drawing the pattern) and use it to guide the authentication. Moreover, we design a polyline weighted strategy with overlapping based on the consistency of pattern lock, which analyzes the behavior information of the user during the unlock process in a fine-grained manner and takes an overall consideration the results of different polylines. Based on 14,850 samples collected from 77 participants, we have extensively evaluated the proposed system. The results demonstrate that it outperforms state-of-the-art implicit authentication based pattern lock approaches, and that each key module in our system is effective.

PresSafe: Barometer-Based On-Screen Pressure-Assisted Implicit Authentication for Smartphones

“…Wang et al [40] proposed a context-aware module to classify the application scenario, and send the weighted user data to the corresponding authentication module. Shi et al [41] further adopted a polygonal line weighted strategy. This method takes consistency of unlocking patterns into account, then analyzes the patterns with different grains, and enhances the information stored in the patterns' principal parts though weighting.…”

“…This requirement diminishes the value of authentication in real world applications. Some recent studies [13], [40], [41] applied conventional binary classifiers to address this problem. But this approach relies heavily on manually extracted features from the raw data.…”

“…Patterns. The pattern unlocking tech was first introduced by Google [44] to enhance the security level of Android smartphones in 2008, and recent studies [41] have proved that the following perspectives on pattern unlocking. Screen-related Info.…”

Stepping Into the Next Decade of Ubiquitous and Pervasive Computing: UbiComp and ISWC 2021

“…Krasovec et al 88 proposed an IoT testbed to gather behavior data, such as a person's movement in space, interaction with certain physical objects, PC terminal usage, keyboard typing etc., which further can be analyzed for authentication purpose. In another paper, a behavior biometrics-based authentication system 89 can be deployed in a smartphone's pattern lock. Similarly, Kumar et al 90 proposed Aquilis, a smartphone-based privacy protection system on this front.…”

“…Similarly, Kumar et al 90 proposed Aquilis, a smartphone-based privacy protection system on this front. On the other hand, the "Privacy" session discusses different aspects of privacy concerns in this domain, including privacy-preserving customization framework for edge computing, 91 privacy implications in location tracking, 89 spatial privacy risks in mobilemixed reality data, 92 location privacy in crowd-sensing photos 93 and privacy-preserving federated learning. 94 In the Q&A session, the audience was interested in "thread modeling," "generalization of Aquilis modeling,' 'and "adversarial geo-identification" concerns regarding location images and "one's privacy preferences."…”