Harnessing Meta-Reinforcement Learning for Enhanced Tracking in Geofencing Systems
dc.contributor.author | Famili, Alireza | en |
dc.contributor.author | Sun, Shihua | en |
dc.contributor.author | Atalay, Tolga | en |
dc.contributor.author | Stavrou, Angelos | en |
dc.date.accessioned | 2025-03-10T17:46:54Z | en |
dc.date.available | 2025-03-10T17:46:54Z | en |
dc.date.issued | 2025-01-20 | en |
dc.description.abstract | Geofencing technologies have become pivotal in creating virtual boundaries for both real and virtual environments, offering a secure means to control and monitor designated areas. They are now considered essential tools for defining and controlling boundaries across various applications, from aviation safety in drone management to access control within mixed reality platforms like the metaverse. Effective geofencing relies heavily on precise tracking capabilities, a critical component for maintaining the integrity and functionality of these systems. Leveraging the advantages of 5G technology, including its large bandwidth and extensive accessibility, presents a promising solution to enhance geofencing performance. In this paper, we introduce MetaFence: Meta-Reinforcement Learning for Geofencing Enhancement, a novel approach for precise geofencing utilizing indoor 5G small cells, termed "5G Points", which are optimally deployed using a meta-reinforcement learning (meta-RL) framework. Our proposed meta-RL method addresses the NP-hard problem of determining an optimal placement of 5G Points to minimize spatial geometry-induced errors. Moreover, the meta-training approach enables the learned policy to quickly adapt to diverse new environments. We devised a comprehensive test campaign to evaluate the performance of MetaFence. Our results demonstrate that this strategic placement significantly improves tracking accuracy compared to traditional methods. Furthermore, we show that the meta-training strategy enables the learned policy to generalize effectively and perform efficiently when faced with new environments. | en |
dc.description.version | Published version | en |
dc.format.extent | Pages 944-960 | en |
dc.format.extent | 17 page(s) | en |
dc.format.mimetype | application/pdf | en |
dc.identifier.doi | https://doi.org/10.1109/OJCOMS.2025.3531318 | en |
dc.identifier.eissn | 2644-125X | en |
dc.identifier.issn | 2644-125X | en |
dc.identifier.orcid | Stavrou, Angelos [0000-0001-9888-0592] | en |
dc.identifier.uri | https://hdl.handle.net/10919/124838 | en |
dc.identifier.volume | 6 | en |
dc.language.iso | en | en |
dc.publisher | IEEE | en |
dc.rights | Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International | en |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/4.0/ | en |
dc.subject | 5G mobile communication | en |
dc.subject | Accuracy | en |
dc.subject | Three-dimensional displays | en |
dc.subject | Geometry | en |
dc.subject | Drones | en |
dc.subject | Distance measurement | en |
dc.subject | Wireless fidelity | en |
dc.subject | NP-hard problem | en |
dc.subject | Mixed reality | en |
dc.subject | Metaverse | en |
dc.subject | Geofencing | en |
dc.subject | tracking | en |
dc.subject | meta-RL | en |
dc.subject | sensor placement | en |
dc.subject | 5G networks | en |
dc.title | Harnessing Meta-Reinforcement Learning for Enhanced Tracking in Geofencing Systems | en |
dc.title.serial | IEEE Open Journal of the Communications Society | en |
dc.type | Article - Refereed | en |
dc.type.dcmitype | Text | en |
dc.type.other | Article | en |
dc.type.other | Journal | en |
pubs.organisational-group | Virginia Tech | en |
pubs.organisational-group | Virginia Tech/All T&R Faculty | en |
pubs.organisational-group | Virginia Tech/Innovation Campus | en |
Files
Original bundle
1 - 1 of 1
Loading...
- Name:
- Harnessing_Meta-Reinforcement_Learning_for_Enhanced_Tracking_in_Geofencing_Systems.pdf
- Size:
- 6.26 MB
- Format:
- Adobe Portable Document Format
- Description:
- Published version
License bundle
1 - 1 of 1