EUVIMEDEUVIMED entdecken
Crossref

Fusing Quaternion Rotations and Star Operations: A TLSNet-Based High-Precision 3-D Indoor Visible Light Positioning System

Tianyi Cao, Lei Jing, Shaorui Sang

IEEE Transactions on Broadcasting · 2026-06 · 72 (2) · 434-448

DOI: 10.1109/tbc.2026.3664184 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Broadcasting, 2026-06. Verfasst von Tianyi Cao und 2 weiteren Autor:innen. Die Arbeit nennt 38 Literaturangaben. Gefördert von: National Natural Science Foundation (NSF) of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper presents a deep-learning-enhanced visible light positioning (VLP) system for accurate 3-D localization under arbitrary receiver orientations. In a 3-D VLP system with a single LED, multiple photodiodes, and an IMU, the commonly used Euler-angle representation is replaced with sign-consistent unit quaternions as inputs to deep-learning models. By further enforcing explicit constraints to achieve a one-to-one, order-invariant mapping, this reparameterization resolves ambiguities and rotation-order dependency, providing a more structured orientation input and improving the achievable performance of the learning-based pipeline under a fixed system configuration. Based on these processed inputs, a compact neural architecture, termed twinkle little star net (TLSNet), is developed to explicitly exploit multiplicative and high-order feature interactions through a structured three-stage design consisting of feature supply, feature resifting, and feature recombination. This architecture enhances nonlinear representational capacity while maintaining parameter efficiency and stable optimization behavior. Experimental results demonstrate that the proposed system achieves centimeter-level localization accuracy, attaining a mean absolute error (MAE) of 3.92 cm, outperforming several representative learning-based VLP models under the same experimental settings. TLSNet also effectively suppresses large-error outliers, with 95% of positioning errors remaining below 9.47 cm across receiver heights up to 1.2 m.

Förderung

  • National Natural Science Foundation (NSF) of China

Zitieren

APA

Cao, T., Jing, L., & Sang, S. (2026). Fusing Quaternion Rotations and Star Operations: A TLSNet-Based High-Precision 3-D Indoor Visible Light Positioning System. IEEE Transactions on Broadcasting, 72(2), 434-448. https://doi.org/10.1109/tbc.2026.3664184

Vancouver

Cao T, Jing L, Sang S. Fusing Quaternion Rotations and Star Operations: A TLSNet-Based High-Precision 3-D Indoor Visible Light Positioning System. IEEE Transactions on Broadcasting. 2026;72(2):434-448. doi:10.1109/tbc.2026.3664184

BibTeX
@article{Cao2026,
  author = {Cao, Tianyi and Jing, Lei and Sang, Shaorui},
  title = {Fusing Quaternion Rotations and Star Operations: A TLSNet-Based High-Precision 3-D Indoor Visible Light Positioning System},
  journal = {IEEE Transactions on Broadcasting},
  year = {2026},
  volume = {72},
  number = {2},
  pages = {434--448},
  doi = {10.1109/tbc.2026.3664184}
}
Kostenlos starten →