Continuous User Authentication via PPG

PPG technology offers potentials not only in sign language interpretation, but also in user authentication. Traditional single-instance user authentication methods often result in a less-than-ideal user experience, especially in frequently used applications. This problem is notably acute in security-sensitive contexts, where unauthorized access could potentially follow a user’s initial login. In response to this, continuous user authentication (CA) has emerged as a compelling solution, offering seamless, low-effort authentication for users. In this chapter, we present a cost-effective system that leverages pulsatile signals captured by the photoplethysmography (PPG) sensor in commercially available wrist-worn wearables for CA. Our system stands out by eliminating the need for users’ participations and accommodating non-clinical PPG measurements that are prone to motion artifacts (MA), commonly encountered during daily activities. To address the challenges posed by MA, we delve into the unique characteristics of the human cardiac system and propose an MA filtering method that effectively mitigates the impact of everyday movements. Moreover, we identify key fiducial features and develop an adaptive classifier using the gradient boosting tree (GBT) method. Consequently, our system can continuously authenticate users based on their cardiac characteristics, requiring minimal training effort. We conduct experiments with our wrist-worn PPG sensing platform in practical scenarios. The results demonstrate the high accuracy of our system and a low false detection rate when detecting random attacks.