University of Twente Student Theses
Leveraging Earable Sensors for Lightweight Gait-Based User Recognition
Georgieva, Gergana Nikolova (2025) Leveraging Earable Sensors for Lightweight Gait-Based User Recognition.
|
PDF
3MB |
| Abstract: | Wearable devices are evolving rapidly, with earables' popularity being on the rise as they gain new and sophisticated sensing capabilities. Their growing complexity, however, also poses heightened security risks. As these devices lack the interface to support traditional input-based authentication such as PIN or lock patterns, there is a call for new methods to provide reliable user verification. Gait-based behavioral biometrics, particularly in the context of leveraging IMU data, remain greatly underexplored for earables. This work investigates the feasibility of gait-based user recognition using IMU-equipped earable devices. The authors collect a new dataset consisting of 30 participants performing several different gait-related exercises at varying intensities. Traditional ML models (Random Forest, SVM, kNN, MLP) and a CNN-LSTM hybrid are benchmarked on authentication and identification tasks, in within-activity and across-activity scenarios, averaging at an EER below 2%. Feature selection and post-training quantization are shown to significantly reduce model size and inference cost without sacrificing accuracy. These findings confirm that gait-based user recognition using IMU-equipped earables is both feasible and practical, offering secure, unobtrusive verification on resource-constrained devices. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science MSc (60300) |
| Link to this item: | https://purl.utwente.nl/essays/105229 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page