aalto1 untyped-item.component.html
Channel Covariance CSI-based Indoor Localization: Machine Learning versus Neighbourhood Geometric Approaches
Loading...
Access rights
openAccess
acceptedVersion
URL
Journal Title
Journal ISSN
Volume Title
A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
This publication is imported from Aalto University research portal.
View publication in the Research portal (opens in new window)
View/Open full text file from the Research portal (opens in new window)
Other link related to publication (opens in new window)
View publication in the Research portal (opens in new window)
View/Open full text file from the Research portal (opens in new window)
Other link related to publication (opens in new window)
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Date
Major/Subject
Mcode
Degree programme
Language
en
Pages
16
Series
IEEE Transactions on Vehicular Technology, pp. 1-16
Abstract
We consider fingerprinting-based localization in highly cluttered Non-Line-of-Sight (NLoS) multipath environments, typical of indoor scenarios. Channel covariances of multiple Base Stations (BSs) are utilized as Channel State Information (CSI) fingerprints, capturing large-scale features such as power and angle information of multipath components. Several algorithms are analysed, including Weighted K-Nearest Neighbour (WKNN) regression, parametric regression models, and a novel neighbourhood geometric localization method. For WKNN regression, no weighting function can achieve accurate localization for points outside the convex hull of their feature neighbours. In this regard, we propose a neighbour selection method to expand the convex hull. WKNN localization is further enhanced using optimized weighting functions based on Laguerre polynomials and Neural Networks (NNs). Parametric regression models based on kernel methods and NNs are investigated. Furthermore, a feature distance-based neighbourhood geometric framework is developed for NLoS localization, which is extended from conventional geometric positioning models applied in line-of-sight environments. Simulation results show that enhanced WKNN localization with optimum weighting functions and neighbour selection outperforms parametric regression methods. The neighbourhood geometric localization method has comparable localization performance to enhanced WKNN.
Description
Other note
Citation
Li, X, Al-Tous, H, Hajri, S E & Tirkkonen, O 2026, 'Channel Covariance CSI-based Indoor Localization: Machine Learning versus Neighbourhood Geometric Approaches', IEEE Transactions on Vehicular Technology, pp. 1-16. https://doi.org/10.1109/TVT.2026.3674998