EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Fingerprinting Localization Based on Super Resolution CSI Features

Zekeriya Uykan, Hanan Al-Tous, Olav Tirkkonen, Riku Jäntti

IEEE Access · 2026 · 14 · 49750-49767

DOI: 10.1109/access.2026.3676530 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Zekeriya Uykan und 3 weiteren Autor:innen. Die Arbeit nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Research Council of Finland.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

We consider fingerprint localization based on super-resolution features for single point massive multiple-input multiple-output (MIMO) systems. We consider k-nearest neighbours (kNN), weighted k-nearest neighbors (W-kNN) as well as neural network based localization. The radio environment is characterized by several multipath components. We investigate several channel state information (CSI) features. We study the localization performance when the CSI feature includes absolute delays, relative delays, and no-delay information of multipath components. Simulation results based on the DeepMIMO data-set for a challenging non line-of-sight environment show that: 1) treating the angle and delay separately and formulating the normalized polar feature outperforms other CSI features from the literature; 2) the neural network based fingerprint localization is remarkably superior to the W-kNN based fingerprinting localization using the same features; and 3) the best localization performance is obtained when absolute delay information is available. Localization accuracy with few centimeters resolution is achieved by the proposed feature and based on neural network localization, less than 30 cm average localization error can be achieved based on relative delays with the angles and powers of multipath components, and less than 60 cm average localization error can be achieved for the no-delay case (i.e., power and angle of arrival of MPCs).

Förderung

  • Research Council of Finland

Zitieren

APA

Uykan, Z., Al-Tous, H., Tirkkonen, O., & Jäntti, R. (2026). Fingerprinting Localization Based on Super Resolution CSI Features. IEEE Access, 14, 49750-49767. https://doi.org/10.1109/access.2026.3676530

Vancouver

Uykan Z, Al-Tous H, Tirkkonen O, Jäntti R. Fingerprinting Localization Based on Super Resolution CSI Features. IEEE Access. 2026;14:49750-49767. doi:10.1109/access.2026.3676530

BibTeX
@article{Uykan2026,
  author = {Uykan, Zekeriya and Al-Tous, Hanan and Tirkkonen, Olav and Jäntti, Riku},
  title = {Fingerprinting Localization Based on Super Resolution CSI Features},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {49750--49767},
  doi = {10.1109/access.2026.3676530}
}
Kostenlos starten →