EUVIMEDEUVIMED entdecken
Crossref

Mobility-Aware Asynchronous Federated Learning With Dynamic Sparsification

Jintao Yan, Tan Chen, Yuxuan Sun, Zhaojun Nan, Sheng Zhou, Zhisheng Niu

IEEE Transactions on Mobile Computing · 2026-11 · 25 (11) · 19508-19522

DOI: 10.1109/tmc.2026.3709197 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2026-11. Verfasst von Jintao Yan und 5 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 60 Literaturangaben. Gefördert von: China Postdoctoral Science Foundation, Open Fund of State Key Laboratory of Intelligent Green Vehicle and Mobility, Tsinghua University, State Key Laboratory of Internet of Things for Smart City, University of Macau.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Asynchronous Federated Learning (AFL) enables distributed model training across multiple mobile devices, allowing each device to independently update its local model without waiting for others. However, device mobility introduces intermittent connectivity, which necessitates gradient sparsification and leads to model staleness, jointly affecting AFL convergence. This paper develops a theoretical model to characterize the interplay among sparsification, staleness and mobility-induced contact patterns, and their joint impact on AFL convergence. Based on the analysis, we propose a mobility-aware dynamic sparsification (MADS) algorithm that optimizes the sparsification degree and transmission power based on contact time and staleness. A long-term optimization problem is formulated and the Lyapunov optimization method is used to convert it to an online non-convex optimization problem. This problem is further converted to a convex optimization problem and closed-form solutions are obtained. Experimental results validate the theoretical findings. Compared with the state-of-the-art benchmarks, the MADS algorithm increases the image classification accuracy on the CIFAR-10 dataset by 8.76% and reduces the average displacement error in the Argoverse trajectory prediction dataset by 9.46%.

Förderung

  • China Postdoctoral Science Foundation
  • Open Fund of State Key Laboratory of Intelligent Green Vehicle and Mobility, Tsinghua University
  • State Key Laboratory of Internet of Things for Smart City, University of Macau
  • Project of Tsinghua University-Toyota Joint Research Center for AI Technology of Automated Vehicle
  • National Natural Science Foundation of China
  • Young Elite Scientists Sponsorship Program by China Association for Science and Technology

Zitieren

APA

Yan, J., Chen, T., Sun, Y., Nan, Z., Zhou, S., & Niu, Z. (2026). Mobility-Aware Asynchronous Federated Learning With Dynamic Sparsification. IEEE Transactions on Mobile Computing, 25(11), 19508-19522. https://doi.org/10.1109/tmc.2026.3709197

Vancouver

Yan J, Chen T, Sun Y, Nan Z, Zhou S, Niu Z. Mobility-Aware Asynchronous Federated Learning With Dynamic Sparsification. IEEE Transactions on Mobile Computing. 2026;25(11):19508-19522. doi:10.1109/tmc.2026.3709197

BibTeX
@article{Yan2026,
  author = {Yan, Jintao and Chen, Tan and Sun, Yuxuan and Nan, Zhaojun and Zhou, Sheng and Niu, Zhisheng},
  title = {Mobility-Aware Asynchronous Federated Learning With Dynamic Sparsification},
  journal = {IEEE Transactions on Mobile Computing},
  year = {2026},
  volume = {25},
  number = {11},
  pages = {19508--19522},
  doi = {10.1109/tmc.2026.3709197}
}
Kostenlos starten →