Human Activity Recognition in Mobile Edge Computing: A Low-Cost and High-Fidelity Digital Twin Approach With Deep Reinforcement Learning
IEEE Transactions on Consumer Electronics · 2025-05 · 71 (2) · 6451-6459
DOI: 10.1109/tce.2024.3375859 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Consumer Electronics, 2025-05. Verfasst von Chenyu Wang und 2 weiteren Autor:innen. Die Arbeit wurde bisher 18-mal zitiert (Crossref) und nennt 44 Literaturangaben. Gefördert von: National Science Foundation (NSF), USA.
Abstract (Originalsprache)
The presence of digital twins (DTs) has expanded within the consumer electronics area due to its inherent benefits of high-fidelity modeling and predictive insight. However, DT deployment remains costly and constrained in providing the fundamental functionalities required by consumer electronics systems due to massive computation and communication. The emergence of mobile edge computing (MEC) has made DT deployment feasible to achieve data-driven modeling and consumer-centric concurrently with low communication costs. This article proposes an MEC-based DT deployment scheme in the smart home domain. The cloud service can predict subsequent sensor updates leveraging the MEC platform’s human activity recognition result of residential environments. In addition, with deep reinforcement learning (DRL), it can only track essential data to maintain sensor update consistency between both physical and virtual sides. We implement experimental evaluation with two real-world residential datasets, and the results demonstrate that our scheme maintains high-level sensor update consistency while being energy-efficient in different residential environments, which illuminates the promising prospects of DT implementation in MEC and consumer electronics.
Förderung
- National Science Foundation (NSF), USA
Zitieren
APA
Wang, C., Cai, Z., & Li, Y. (2025). Human Activity Recognition in Mobile Edge Computing: A Low-Cost and High-Fidelity Digital Twin Approach With Deep Reinforcement Learning. IEEE Transactions on Consumer Electronics, 71(2), 6451-6459. https://doi.org/10.1109/tce.2024.3375859
Vancouver
Wang C, Cai Z, Li Y. Human Activity Recognition in Mobile Edge Computing: A Low-Cost and High-Fidelity Digital Twin Approach With Deep Reinforcement Learning. IEEE Transactions on Consumer Electronics. 2025;71(2):6451-6459. doi:10.1109/tce.2024.3375859
BibTeX
@article{Wang2025,
author = {Wang, Chenyu and Cai, Zhipeng and Li, Yingshu},
title = {Human Activity Recognition in Mobile Edge Computing: A Low-Cost and High-Fidelity Digital Twin Approach With Deep Reinforcement Learning},
journal = {IEEE Transactions on Consumer Electronics},
year = {2025},
volume = {71},
number = {2},
pages = {6451--6459},
doi = {10.1109/tce.2024.3375859}
}
EUVIMED