EUVIMEDEUVIMED entdecken
Crossref

LDA-FedHAR: Federated Human Activity Recognition for Wearable Devices Through Local HAR Data Alignment

Minda Yao, Wei Chen, Zehua Wang, Minglong Cheng, Chuanlei Zhang, Fei Richard Yu, Victor C. M. Leung

IEEE Internet of Things Journal · 01.12.2025 · 12 (23) · 49442-49455

DOI: 10.1109/jiot.2025.3602828 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Internet of Things Journal, erschienen am 01.12.2025. Verfasst von Minda Yao und 6 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 42 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Jiangsu Distinguished Professor project in Jiangsu Province, Graduate Innovation Program of China University of Mining and Technology.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Wearable device-based human activity recognition (HAR) has attracted considerable interest with the rapid development of the Internet of Things (IoT), and federated learning (FL) has been widely adopted in this domain for its ability to collaboratively train models across decentralized devices while preserving privacy. However, its performance is hindered by data heterogeneity arising from variations in the placement of the wearable devices, user behaviors, and physiological characteristics. In this work, we present local HAR data alignment (LDA)-FedHAR, a federated HAR framework designed for wearable devices by capturing more common knowledge from aligned client HAR data. It performs LDA on each client, which is an entirely on-device alignment method that operates independently on local HAR data. By computing the transformation matrix solely from local HAR data and applying it to the data itself, LDA projects heterogeneous client data into a unified space, thereby reducing interclient discrepancies at the source. To further enhance efficiency and robustness, we propose two inertial measurement unit (IMU)-specific variants, LDA(S-IMU) and LDA(C-IMU), which explore intra- and inter-IMU correlations based on practical placements of wearable devices. Experiments are conducted on 4 public HAR datasets: heterogeneity HAR (HHAR), Shoaib2014, OPPORTUNITY++, and PAMAP2. The results show that LDA effectively reduces interclient discrepancies, and LDA-FedHAR along with its variants consistently outperforms state-of-the-art FL methods. Moreover, the improvements achieved by integrating LDA into other FL methods highlight its applicability.

Förderung

  • National Natural Science Foundation of China
  • Jiangsu Distinguished Professor project in Jiangsu Province
  • Graduate Innovation Program of China University of Mining and Technology

Zitieren

APA

Yao, M., Chen, W., Wang, Z., Cheng, M., Zhang, C., Yu, F. R., & Leung, V. C. M. (2025). LDA-FedHAR: Federated Human Activity Recognition for Wearable Devices Through Local HAR Data Alignment. IEEE Internet of Things Journal, 12(23), 49442-49455. https://doi.org/10.1109/jiot.2025.3602828

Vancouver

Yao M, Chen W, Wang Z, Cheng M, Zhang C, Yu FR, et al. LDA-FedHAR: Federated Human Activity Recognition for Wearable Devices Through Local HAR Data Alignment. IEEE Internet of Things Journal. 2025;12(23):49442-49455. doi:10.1109/jiot.2025.3602828

BibTeX
@article{Yao2025,
  author = {Yao, Minda and Chen, Wei and Wang, Zehua and Cheng, Minglong and Zhang, Chuanlei and Yu, Fei Richard and Leung, Victor C. M.},
  title = {LDA-FedHAR: Federated Human Activity Recognition for Wearable Devices Through Local HAR Data Alignment},
  journal = {IEEE Internet of Things Journal},
  year = {2025},
  volume = {12},
  number = {23},
  pages = {49442--49455},
  doi = {10.1109/jiot.2025.3602828}
}
Kostenlos starten →