Recent advances in biometrics-based user authentication for wearable devices: A contemporary survey

• Published in: Digital signal processing 2022-06, Vol.125, p.103120, Article 103120
• Main Authors: Liu, Shuqi, Shao, Wei, Li, Tan, Xu, Weitao, Song, Linqi
• Format: Article
• Language: English
• Subjects: Machine learning; Signal processing; User authentication; Wearable devices
• ISSN: 1051-2004; 1095-4333
• DOI: 10.1016/j.dsp.2021.103120

In recent years, wearable technology is interwoven with our everyday lives because of its commoditization and comfort. Security and privacy become a big concern as many user-sensitive data have been stored in such devices, such as personal emails and bank accounts. Traditional user authentication techniques like PIN entry are unfriendly and vulnerable to shoulder surfing attacks. To address these problems, a number of new authentication methods have been proposed. In this survey, we review and categorize recent advances in user authentication for wearable devices. We classify existing studies into physiological biometrics based and behavioral biometrics based methods. For each category, we review how signal processing techniques have been used to extract features in various wearable devices. Leveraging these extracted features, specifically designed classification methods can be used to realize user authentication. Finally, we review evaluation metrics for user authentication in wearable devices. Overall, in this survey, we systematically study assorted state-of-the-art user authentication methods for wearable devices, aiming to provide guidance and directions for future research in this area.