Improving the Security of Smartwatch Payment with Deep Learning

TL;DR

This paper investigates deep learning techniques to enhance smartwatch payment security by reducing user enrollment gestures and improving authentication accuracy, making the process more user-friendly and reliable.

Contribution

It introduces a deep learning-based authentication system that outperforms existing methods and employs synthetic gesture generation to decrease enrollment gestures without increasing error rates.

Findings

Deep learning improves authentication accuracy with fewer gestures.

Synthetic gestures enhance classifier training and performance.

Reduced enrollment gestures do not compromise security or accuracy.

Abstract

Making contactless payments using a smartwatch is increasingly popular, but this payment medium lacks traditional biometric security measures such as facial or fingerprint recognition. In 2022, Sturgess et al. proposed WatchAuth, a system for authenticating smartwatch payments using the physical gesture of reaching towards a payment terminal. While effective, the system requires the user to undergo a burdensome enrolment period to achieve acceptable error levels. In this dissertation, we explore whether applications of deep learning can reduce the number of gestures a user must provide to enrol into an authentication system for smartwatch payment. We firstly construct a deep-learned authentication system that outperforms the current state-of-the-art, including in a scenario where the target user has provided a limited number of gestures. We then develop a regularised autoencoder model for generating synthetic user-specific gestures. We show that using these gestures in training improves classification ability for an authentication system. Through this technique we can reduce the number of gestures required to enrol a user into a WatchAuth-like system without negatively impacting its error rates.