User Authentication on Wearable Devices by Component Analysis of Heartbeat Signals


Authors:

Dmytro Progonov, Samsung Electronics & Igor Sikorsky Kyiv Polytechnic Institute; Oleksandra Sokol, Samsung Electronics

Abstract:

Heartbeat authentication technology is suitable for use in mass-market wearable devices. Nevertheless, practical usage of the technology is limited due to low-accuracy cheap electrocardiogram sensors, signal fluctuations caused by varying usage contexts and high computational complexity.

We propose to use an innovative component analysis method to address these issues by applying less resource-intensive algorithms. Our evaluation results show that the proposed processing pipelines, based on multifractal analysis and singular spectrum analysis, achieves close to state-of-the-art accuracy (0.01% False Acceptance Rate and 0.77% False Rejection Rate), and outperforms it in various usage contexts. Obtained results confirm the effectiveness of using specialized signal processing methods for heartbeat authentication on constrained devices such as wearables.

Download:
BibTeX:
@inproceedings{progonov-21-heartbeat,
    author = {Progonov, Dmytro and Sokol, Oleksandra},
    title = {{User Authentication on Wearable Devices by Component Analysis of Heartbeat Signals}},
    booktitle = {Who Are You?! Adventures in Authentication Workshop},
    year = {2021},
    series = {WAY~'21},
    pages = {1--5},
    address = {Virtual Conference},
    month = aug,
    publisher = {}
} % No publisher