pith. sign in

arxiv: 2604.15940 · v1 · submitted 2026-04-17 · 💻 cs.LG · stat.AP

(Weighted) Adaptive Radius Near Neighbor Search: Evaluation for WiFi Fingerprint-based Positioning

Khang Le , Joaqu\'in Torres-Sospedra , Philipp M\"uller This is my paper

Pith reviewed 2026-05-10 08:53 UTC · model grok-4.3

classification 💻 cs.LG stat.AP
keywords adaptivefrnnarnnnearneighborradiussearchdistance
0
0 comments X

The pith

Weighted Adaptive Radius Near Neighbor search ranks among the top performers and matches or exceeds kNN variants on WiFi fingerprint indoor positioning across 22 datasets.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Fixed radius nearest neighbor search predicts a label or position by considering every training example that falls inside a preset distance threshold. The authors note that a single fixed distance often fails to suit all queries, so they introduce an adaptive version where the radius changes for each new point. They further add weighting so that closer or more relevant neighbors contribute more to the final estimate. These methods are tested on WiFi fingerprinting, a regression task that uses received signal strengths from access points to estimate a device's location inside buildings. The evaluation runs on 22 separate datasets collected in different environments. Results show that plain fixed-radius and adaptive-radius versions performed poorly, while the weighted adaptive versions placed among the four strongest methods overall and were competitive with or better than many kNN variants.

Core claim

While the performances of the tested FRNN and ARNN versions were amongst the worse, three of the four best methods in the test were WARNN versions, indicating that using weights together with adaptive distances achieves performance comparable or even better than kNN variants.

Load-bearing premise

That the adaptive radius rule and weighting scheme can be defined and tuned in a way that generalizes across indoor environments without post-hoc adjustments that favor the proposed methods on the chosen datasets.

Figures

Figures reproduced from arXiv: 2604.15940 by Joaqu\'in Torres-Sospedra, Khang Le, Philipp M\"uller.

Figure 1
Figure 1. Figure 1: Illustration of challenge to choose suitable weights for the WARNN. Therefore, we propose an adaptive decay factor for the inverse distance weighting (IDW) function that reflects the distance between the test sample x and a neighbor yi as well as the relative position of the test sample in the sphere of yi defined by radius ri . The IDW function is defined by (1) wi = 1 d(yi , x) α , where α is the decay f… view at source ↗
Figure 2
Figure 2. Figure 2: 3D average positioning errors and corresponding coverage ratios of M23 for varying τϵ. If individual threshold values τϵ yielding the mean 3D positioning errors would be chosen for all dataset, the average 3D positioning error could be further reduced to 4.05 m with a coverage ratio of 95.81%. This suggests that using τϵ as a parameter, which should be optimized for each dataset individually, could be bene… view at source ↗
read the original abstract

Fixed Radius Near Neighbor (FRNN) search is an alternative to the widely used k Nearest Neighbors (kNN) search. Unlike kNN, FRNN determines a label or an estimate for a test sample based on all training samples within a predefined distance. While this approach is beneficial in certain scenarios, assuming a fixed maximum distance for all training samples can decrease the accuracy of the FRNN. Therefore, in this paper we propose the Adaptive Radius Near Neighbor (ARNN) and the Weighted ARNN (WARNN), which employ adaptive distances and in latter case weights. All three methods are compared to kNN and twelve of its variants for a regression problem, namely WiFi fingerprinting indoor positioning, using 22 different datasets to provide a comprehensive analysis. While the performances of the tested FRNN and ARNN versions were amongst the worse, three of the four best methods in the test were WARNN versions, indicating that using weights together with adaptive distances achieves performance comparable or even better than kNN variants.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper proposes Adaptive Radius Near Neighbor (ARNN) and Weighted ARNN (WARNN) as alternatives to fixed-radius NN and kNN for WiFi fingerprint-based indoor positioning regression. It reports a comparison of these methods against kNN and twelve variants across 22 datasets, with the central claim that three of the four best-performing methods were WARNN variants, indicating that weights combined with adaptive distances can achieve performance comparable or superior to kNN approaches.

Significance. If the results hold under scrutiny, the multi-dataset evaluation provides useful empirical evidence that adaptive-radius weighting can be competitive in positioning tasks. The breadth of 22 datasets is a positive aspect for assessing generalizability in indoor environments. However, the lack of implementation details, statistical validation, and discussion of dataset selection limits the strength of the conclusions and their immediate utility for the field.

major comments (3)
  1. [Abstract] Abstract: The claim that 'three of the four best methods in the test were WARNN versions' is presented without error bars, confidence intervals, or any statistical significance tests across the 22 datasets. This is load-bearing for the headline performance comparison, as it leaves open whether the ranking reflects reliable superiority or dataset-specific variation.
  2. [Methods] Methods section (implementation of ARNN/WARNN): No explicit formulas, parameter values, or tuning procedure are provided for the adaptive radius rule or the weighting scheme. This omission is load-bearing because the central claim that these methods generalize and outperform kNN variants depends on reproducible definitions that do not require post-hoc dataset-specific adjustments.
  3. [Experiments] Experimental evaluation (dataset selection and analysis): There is no discussion of how the 22 datasets were chosen or any analysis of potential selection effects, nor are there breakdowns by environment type (e.g., building size or signal characteristics). This undermines confidence that the reported superiority of WARNN holds beyond the chosen collection.
minor comments (2)
  1. [Abstract] Abstract: The phrase 'twelve of its variants' for kNN is vague; listing the specific variants (e.g., weighted kNN, distance-weighted, etc.) would improve clarity for readers.
  2. [Throughout] Throughout: Adding pseudocode or a clear algorithmic outline for the adaptive radius computation and weighting would aid reproducibility, especially given the empirical focus.

Axiom & Free-Parameter Ledger

2 free parameters · 0 axioms · 0 invented entities

The central claim rests on the specific definition of the adaptive radius rule and the weighting function, both of which are introduced without upstream justification in the abstract and likely involve tunable elements.

free parameters (2)
  • adaptive radius rule parameters
    The mechanism that sets the per-query radius is not specified and must involve at least one tunable or data-dependent quantity.
  • weighting function parameters
    Weights assigned to neighbors within the adaptive radius require a functional form and scaling constants that are not detailed.

pith-pipeline@v0.9.0 · 5482 in / 1167 out tokens · 28607 ms · 2026-05-10T08:53:44.415342+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

31 extracted references · 31 canonical work pages

  1. [1]

    Discriminatory analysis. nonparametric discrimination: Consistency properties,

    E. Fix and J.L. Hodges, “Discriminatory analysis. nonparametric discrimination: Consistency properties,” Int. Stat. Rev., 57(3), 238–247, 1989

    work page 1989

  2. [2]

    Let’s Talk about k-NN for Indoor Positioning: Myths and Facts in RF-based Fingerprinting,

    J. Torres-Sospedra, C. Pend˜ ao, I. Silva, et al., “Let’s Talk about k-NN for Indoor Positioning: Myths and Facts in RF-based Fingerprinting,” IPIN 2023, Sept. 2023

    work page 2023

  3. [3]

    A survey of techniques for fixed radius near neighbor searching,

    J.L. Bentley, “A survey of techniques for fixed radius near neighbor searching,” Technical Report SLAC-186 and STAN-CS-75-513, Stanford Linear Accelerator Center, August 1975

    work page 1975

  4. [4]

    Molecular model-building by computer,

    C. Levinthal, “Molecular model-building by computer,” Scientific American, 214, 42–52, 1966

    work page 1966

  5. [5]

    Comparative performance analysis of K-nearest neighbour (KNN) algorithm and its different variants for disease prediction,

    S. Uddin, I. Haque, H. Lu, et al., “Comparative performance analysis of K-nearest neighbour (KNN) algorithm and its different variants for disease prediction,” Scientific Reports, 12(6256), 2022

    work page 2022

  6. [6]

    RADAR: An in-building RF-based user location and tracking system,

    P. Bahl and V. Padmanabhan, “RADAR: An in-building RF-based user location and tracking system,” Proceedings IEEE INFOCOM, 2000

    work page 2000

  7. [7]

    Performance analysis of adaptive K for weighted K-nearest neighbor based indoor positioning

    S. Liu, R. Lacerda, and J. Fiorina, “Performance analysis of adaptive K for weighted K-nearest neighbor based indoor positioning”, VTC2022-Spring, June 2022

    work page 2022

  8. [8]

    Experimental analysis on weight K -nearest neighbor indoor fingerprint positioning,

    J. Hu, D. Liu, Z. Yan, H. Liu , “Experimental analysis on weight K -nearest neighbor indoor fingerprint positioning,” IEEE Internet of Things Journal, 6(1), 891–897, 2019

    work page 2019

  9. [9]

    Flexible K Nearest Neighbors Classifier: Derivation and Application for Ion-mobility Spectrometry-based Indoor Localization,

    P. M¨ uller, “Flexible K Nearest Neighbors Classifier: Derivation and Application for Ion-mobility Spectrometry-based Indoor Localization,” IPIN 2023, Sept. 2023

    work page 2023

  10. [10]

    Locally adaptive nearest neighbor algorithms,

    D. Wettschereck and T.G. Dietterich, “Locally adaptive nearest neighbor algorithms,” Proceedings of NIPS’93, pp. 184–191, November 1993

    work page 1993

  11. [11]

    An adaptive k-nearest neighbor algorithm,

    S. Sun and R. Huang, “An adaptive k-nearest neighbor algorithm,” 7th Int. Conf. on Fuzzy Systems & Knowledge Discovery, 91–94, 2010

    work page 2010

  12. [12]

    The Distance-Weighted k-Nearest-Neighbor Rule,

    S.A. Dudani, “The Distance-Weighted k-Nearest-Neighbor Rule,” IEEE Trans. SMC, 6(4), 325–327, 1976

    work page 1976

  13. [13]

    Comparative evaluation of spatial prediction methods in a field experiment for mapping soil potassium,

    A. Bekele, R. Downer, M. Wolcott, et al., “Comparative evaluation of spatial prediction methods in a field experiment for mapping soil potassium,” Soil Science, 168, 15–28, 2003

    work page 2003

  14. [14]

    Exploring spatial dependence of cotton yield using global and local autocorrelation statistics,

    J. Ping, C. Green, R. Zartman, and K. Bronson, “Exploring spatial dependence of cotton yield using global and local autocorrelation statistics,” Field Crops Research, 89, 219–236, 2004

    work page 2004

  15. [15]

    Assessing the Effect of Integrating Elevation Data Into the Estimation of Monthly Precipitation in Great Britain,

    C. Lloyd, “Assessing the Effect of Integrating Elevation Data Into the Estimation of Monthly Precipitation in Great Britain,” J of Hydrology, 308, 128–150, 2005

    work page 2005

  16. [16]

    An adaptive inverse-distance weighting spatial interpolation technique,

    G. Lu and D. Wong, “An adaptive inverse-distance weighting spatial interpolation technique,” Computers Geosciences, 34, pp. 1044–1055, 2008

    work page 2008

  17. [17]

    The DSI dataset for Wi-FI fingerprinting using mobile devices,

    A. Moreira, I. Silva, J. Torres-Sospedra, “The DSI dataset for Wi-FI fingerprinting using mobile devices,” Version 1.0, Zenodo, 2020

    work page 2020

  18. [18]

    Long-term WiFi fingerprinting dataset for research on robust indoor positioning,

    G.M. Mendoza-Silva, P. Richter, J. Torres-Sospedra, et al., “Long-term WiFi fingerprinting dataset for research on robust indoor positioning,” Data, 3(1), 3, 2018. (WEIGHTED) ADAPTIVE RADIUS NEAR NEIGHBOR SEARCH 11

    work page 2018

  19. [19]

    Deconvolution-based indoor localization with WLAN signals and unknown access point locations,

    S. Shreshta, J. Talvitie, and E.S. Lohan, “Deconvolution-based indoor localization with WLAN signals and unknown access point locations,” [online] http: //www.cs.tut.fi/tlt/pos/MEASUREMENTS WLAN FOR WEB.zip, 2013

    work page 2013

  20. [20]

    K-means fingerprint clustering for low-complexity floor estima- tion in indoor mobile localization,

    A. Razavi, M. Valkama, and E.S. Lohan, “K-means fingerprint clustering for low-complexity floor estima- tion in indoor mobile localization,” IEEE Globecom Workshops, 2015

    work page 2015

  21. [21]

    Clustering benefits in mobile-centric WiFi positioning in multi-floor buildings,

    A. Cramariuc, H. Huttunen, and E.S. Lohan, “Clustering benefits in mobile-centric WiFi positioning in multi-floor buildings,” ICL-GNSS, 2016

    work page 2016

  22. [22]

    Wi-Fi crowdsourced fingerprinting dataset for indoor positioning,

    E.S. Lohan, J. Torres-Sospedra, H. Lepp¨ akoski, et al., “Wi-Fi crowdsourced fingerprinting dataset for indoor positioning,” Data, 2(4), 2017

    work page 2017

  23. [23]

    WLAN (WiFi) RSS database for fingerprinting positioning

    P. Richter, E.S. Lohan, andJ. Talvitie, “WLAN (WiFi) RSS database for fingerprinting positioning”, Version 1.0.0, Zenodo, 2018

    work page 2018

  24. [24]

    Additional TAU datasets for Wi-Fi fingerprinting-based positioning,

    E.S. Lohan, “Additional TAU datasets for Wi-Fi fingerprinting-based positioning,” Zenodo, 2020

    work page 2020

  25. [25]

    UJIIndoorLoc: A new multi-building and multi- floor database for WLAN fingerprint-based indoor localization problems,

    J. Torres-Sospedra, R. Montoliu, A. Mart´ ınez-Us´ o, et al., “UJIIndoorLoc: A new multi-building and multi- floor database for WLAN fingerprint-based indoor localization problems,” IPIN 2014, Sept. 2014

    work page 2014

  26. [26]

    Supplementary open dataset for WiFi indoor localization based on received signal strength,

    J. Bi, Y. Wang, B. Yu, et al., “Supplementary open dataset for WiFi indoor localization based on received signal strength,” Satell Navig, 3(1), 25, 2022

    work page 2022

  27. [27]

    Supplementary Materials for<<Let’s Talk about k-NN for Indoor Positioning: Myths and Facts in RF-based Fingerprinting>>,

    J. Torres-Sospedra, C. Pend˜ ao, I. Silva, et al., “Supplementary Materials for<<Let’s Talk about k-NN for Indoor Positioning: Myths and Facts in RF-based Fingerprinting>>,” Version 1.0, Zenodo, 2023

    work page 2023

  28. [28]

    Towards reproducible indoor positioning research,

    G.G. Anagnostopoulos and A. Kalousis, “Towards reproducible indoor positioning research,” IPIN 2021, Nov. 2021

    work page 2021

  29. [29]

    Fingerprint-based location positioning using improved KNN,

    X. Liang, X. Gou, and Y. Liu, “Fingerprint-based location positioning using improved KNN,” 3rd IEEE Int. Conf. on Network Infrastr & Digital Content, 57–61, 2012

    work page 2012

  30. [30]

    WinIPS: WiFi-based non-intrusive indoor positioning system with online radio map construction and adaptation,

    H. Zou, M. Jin, H. Jiang, et al., “WinIPS: WiFi-based non-intrusive indoor positioning system with online radio map construction and adaptation,” IEEE Trans. Wireless Communications, 16(12), 8118–8130, 2017

    work page 2017

  31. [31]

    Adaptive Residual Weighted K-Nearest Neighbor Fingerprint Positioning Algorithm Based on Visible Light Communication,

    S. Xu, C.-C. Chen, Y. Wu, et al., “Adaptive Residual Weighted K-Nearest Neighbor Fingerprint Positioning Algorithm Based on Visible Light Communication,” Sensors, 20(16), 4432, 2020

    work page 2020