pith. sign in

arxiv: 2505.01185 · v3 · submitted 2025-05-02 · 💻 cs.NI · cs.AI· cs.LG· eess.SP

Environment-Aware Indoor LoRaWAN Ranging Using Path Loss Model Inversion and Adaptive RSSI Filtering

Nahshon Mokua Obiri , Kristof Van Laerhoven This is my paper

Pith reviewed 2026-05-22 17:26 UTC · model grok-4.3

classification 💻 cs.NI cs.AIcs.LGeess.SP
keywords LoRaWANindoor rangingpath loss modelRSSI filteringenvironmental sensorsdistance estimationKalman filtersingle gateway
0
0 comments X

The pith

A calibrated path loss model with environmental sensors and RSSI Kalman filtering reaches 4.74 m mean error for single-gateway indoor LoRaWAN ranging.

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

The paper shows that indoor LoRaWAN distance estimates can be made accurate enough for many applications by folding real-time measurements of temperature, humidity, CO2, particulates, and pressure into a multi-wall path loss equation, then cleaning the raw RSSI values with a simple forward Kalman filter before inversion. On more than two million packets collected over a full year in one office, the resulting distances have 4.74 m mean absolute error and 6.76 m root-mean-square error. This is roughly half the error of a standard structure-only model and noticeably better than the same environment-aware model run without the filter. The method stays lightweight, needs only one gateway, and produces interpretable per-packet distances that can serve as a starting point for later multi-gateway localization.

Core claim

The authors invert a site-calibrated, multi-wall path loss model whose attenuation terms are expanded by frequency, SNR, and five co-located environmental variables; a forward-only innovation-driven Kalman prefilter is applied to the incoming RSSI stream before the inversion step. On a year-long single-gateway office trace of more than two million uplinks the pipeline yields 4.74 m MAE and 6.76 m RMSE, against 12.07 m MAE for the structure-only COST-231 baseline and 7.76 m MAE for the environment-augmented model without filtering. The filter itself cuts RSSI standard deviation from 10.33 dB to 5.43 dB and raises the path-loss fit from R^{2} = 0.82 to 0.89.

What carries the argument

Environment-augmented multi-wall path loss model inverted for distance after an innovation-driven Kalman prefilter on RSSI.

If this is right

  • Single-anchor LoRaWAN can produce sub-10 m ranging without dense gateway deployment.
  • RSSI volatility drops by nearly half and path-loss model fit improves from R^{2} 0.82 to 0.89.
  • The pipeline supplies an O(1) per-packet distance that can be fed directly into later multi-gateway localization algorithms.
  • The same environmental covariates can be reused for other indoor wireless technologies that rely on received-signal strength.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The method may generalize to buildings whose micro-climate changes more rapidly than the one-year office trace examined here.
  • Adding a second gateway with the same calibrated model could convert the per-packet distances into two-dimensional positions without new hardware.
  • Because the model remains fully interpretable, an operator can inspect which environmental variable most affects a given packet's error.

Load-bearing premise

The path loss equation must be calibrated at the deployment site with co-located environmental sensors whose readings remain stationary enough for deterministic inversion to give usable distances.

What would settle it

Re-run the identical one-year trace with the environmental covariates removed or with the Kalman filter disabled and measure whether the MAE stays below 7 m; if it rises back above 10 m the contribution of those two components is confirmed.

Figures

Figures reproduced from arXiv: 2505.01185 by Kristof Van Laerhoven, Nahshon Mokua Obiri.

Figure 1
Figure 1. Figure 1: Experimental layout (left) and end-node prototype (right). [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: Distance estimation error comparison: (a) RMSE, (b) MAE, (c) [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Relative error distributions per device for the MWM, the MWM-EP, [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
read the original abstract

Achieving sub-10 m indoor ranging with LoRaWAN is challenging because multipath, human blockage, and micro-climate dynamics induce non-stationary attenuation in received signal strength indicator (RSSI) measurements. We present a lightweight, interpretable, site-calibrated pipeline that couples an environment-aware multi-wall path loss model with a forward-only, innovation-driven Kalman prefilter for RSSI. The model augments distance and wall terms with frequency, signal-to-noise ratio (SNR), and co-located environmental covariates, including temperature, relative humidity, carbon dioxide, particulate matter, and barometric pressure, and is inverted deterministically for distance estimation. On a one-year single-gateway office dataset comprising over 2 million uplinks, the approach attains a mean absolute error (MAE) of 4.74 m and a root mean square error (RMSE) of 6.76 m in distance estimation, improving over a structure-only COST-231 multi-wall baseline with 12.07 m MAE and an environment-augmented variant without filtering with 7.76 m MAE. Filtering reduces RSSI volatility from 10.33 to 5.43 dB and lowers the path loss RMSE from 8.09 to 5.35 dB, while increasing R^2 from 0.82 to 0.89. The result is a single-anchor LoRaWAN ranging method with O(1) per-packet cost that is stable, interpretable, and practical within a calibrated indoor deployment, providing a useful building block for future multi-gateway localization and a benchmark for indoor LoRaWAN ranging.

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

2 major / 2 minor

Summary. The manuscript presents a site-calibrated pipeline for indoor LoRaWAN ranging that augments a multi-wall path loss model with frequency, SNR, and environmental covariates (temperature, relative humidity, CO2, particulate matter, barometric pressure) and inverts it deterministically for distance estimates, combined with a forward-only innovation-driven Kalman prefilter on RSSI. On a one-year single-gateway office dataset of over 2 million uplinks, it reports MAE of 4.74 m and RMSE of 6.76 m, outperforming a structure-only COST-231 baseline (12.07 m MAE) and an environment-augmented variant without filtering (7.76 m MAE). Additional claims include reduced RSSI volatility (10.33 to 5.43 dB), lower path loss RMSE (8.09 to 5.35 dB), and higher R² (0.82 to 0.89). The method is positioned as lightweight, interpretable, and a building block for multi-gateway localization.

Significance. If the reported gains prove robust to proper out-of-sample validation, the work would offer a practical, low-complexity benchmark for single-anchor indoor ranging in LoRaWAN, directly addressing non-stationary attenuation from micro-climate and multipath. The scale of the real-world dataset and explicit baseline comparisons are strengths. However, the central performance claims rest on site-specific calibration whose generalization is not yet demonstrated, limiting immediate impact until the validation procedure is clarified.

major comments (2)
  1. [Abstract] Abstract and Evaluation section: The environmental covariate coefficients (temperature, RH, CO2, PM, barometric pressure), SNR, frequency, and wall-loss terms are fitted on the same one-year dataset used to compute the reported 4.74 m MAE and 6.76 m RMSE. No temporal hold-out, deployment-period split, or k-fold procedure is described, so the improvement over the 7.76 m environment-augmented baseline may partly reflect in-sample absorption of dataset-specific correlations rather than robust inversion.
  2. [Abstract] Abstract: The Kalman prefilter is described as innovation-driven and forward-only, yet no information is given on whether its process/measurement noise parameters or innovation threshold were tuned on the evaluation data; this is load-bearing for the volatility reduction (10.33 dB to 5.43 dB) and the final ranging accuracy.
minor comments (2)
  1. [Abstract] Abstract: No error bars, confidence intervals, or statistical significance tests accompany the MAE/RMSE figures or the R² improvement.
  2. [Abstract] Abstract: The explicit functional form of the environment-augmented path-loss model (including how the covariates enter the equation) is omitted, impeding immediate reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback emphasizing validation rigor for our site-calibrated indoor LoRaWAN ranging pipeline. We address each major comment below, clarifying the design intent while committing to revisions that strengthen the out-of-sample aspects without altering the core claims.

read point-by-point responses

  1. Referee: [Abstract] Abstract and Evaluation section: The environmental covariate coefficients (temperature, RH, CO2, PM, barometric pressure), SNR, frequency, and wall-loss terms are fitted on the same one-year dataset used to compute the reported 4.74 m MAE and 6.76 m RMSE. No temporal hold-out, deployment-period split, or k-fold procedure is described, so the improvement over the 7.76 m environment-augmented baseline may partly reflect in-sample absorption of dataset-specific correlations rather than robust inversion.

    Authors: The pipeline is explicitly designed as site-calibrated, with coefficients fitted to the target deployment to capture local multipath and micro-climate effects; this is standard for practical indoor ranging systems. The environment-augmented baseline without filtering employs the identical fitting procedure on the same data, so the reported improvement (from 7.76 m to 4.74 m MAE) isolates the Kalman prefilter contribution rather than fitting artifacts. Nevertheless, we agree that explicit temporal validation would better demonstrate robustness. In the revised manuscript we will add a temporal hold-out experiment, fitting coefficients on the first six months and evaluating distance estimation on the final six months, and report the resulting MAE/RMSE to confirm the gains hold out-of-sample. revision: yes

  2. Referee: [Abstract] Abstract: The Kalman prefilter is described as innovation-driven and forward-only, yet no information is given on whether its process/measurement noise parameters or innovation threshold were tuned on the evaluation data; this is load-bearing for the volatility reduction (10.33 dB to 5.43 dB) and the final ranging accuracy.

    Authors: The Kalman filter parameters were selected from standard RSSI noise models reported in prior LoRaWAN literature and from variance statistics computed on an initial exploratory subset of the dataset; they were not adjusted to optimize the final MAE, RMSE, or volatility figures. The innovation threshold was fixed at a conservative level to reject only gross outliers while preserving causality. To eliminate ambiguity we will insert a short subsection in the revised manuscript that states the exact parameter values, the selection rationale, and confirms that no performance-driven tuning on the reported metrics was performed. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical performance metrics are independent of derivation

full rationale

The paper defines an environment-augmented multi-wall path loss model incorporating frequency, SNR, and environmental covariates, inverts it deterministically for distance, and applies a forward Kalman filter to RSSI. Reported MAE/RMSE values are computed as empirical errors on a large one-year dataset against published baselines (COST-231 and environment-augmented without filtering). No quoted equation or step reduces the claimed performance metric to a fitted parameter by construction, nor does any self-citation or ansatz serve as the load-bearing justification for the central result. The derivation chain remains self-contained as a standard site-calibrated inversion pipeline whose outputs are externally falsifiable distance errors.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on site-specific fitting of environmental coefficients and the assumption that the augmented path loss equation remains invertible without additional constraints.

free parameters (1)
  • environmental covariate coefficients
    Coefficients for temperature, humidity, CO2, PM, and pressure terms in the path loss model are calibrated to the deployment site.
axioms (1)
  • domain assumption The augmented multi-wall path loss model can be inverted deterministically to solve for distance given RSSI and environmental covariates
    Invoked when the model is inverted for distance estimation without stated safeguards against multiple solutions or instability.

pith-pipeline@v0.9.0 · 5843 in / 1355 out tokens · 28866 ms · 2026-05-22T17:26:07.680440+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

20 extracted references · 20 canonical work pages

  1. [1]

    RTPL: A Real-Time Communication Protocol for LoRa Network,

    S. Fahmida, A. Jain, V . P. Modekurthy, D. Ismail, and A. Saifullah, “RTPL: A Real-Time Communication Protocol for LoRa Network,” ACM Trans. Embed. Comput. Syst., vol. 24, no. 1, pp. 17:1–17:31, Dec. 2024

    work page 2024

  2. [2]

    A Survey of LoRaW AN- Integrated Wearable Sensor Networks for Human Activity Recognition: Applications, Challenges and Possible Solutions,

    N. M. Obiri and K. Van Laerhoven, “A Survey of LoRaW AN- Integrated Wearable Sensor Networks for Human Activity Recognition: Applications, Challenges and Possible Solutions,”IEEE Open Journal of the Communications Society, vol. 5, pp. 6713–6735, 2024

    work page 2024

  3. [3]

    ILLOC: In-Hall Localization with Standard LoRaW AN Uplink Frames,

    D. Guo, C. Gu, L. Jiang, W. Luo, and R. Tan, “ILLOC: In-Hall Localization with Standard LoRaW AN Uplink Frames,”Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., vol. 6, no. 1, pp. 13:1– 13:26, Mar. 2022

    work page 2022

  4. [4]

    Recent Advances in WSN-Based Indoor Lo- calization: A Systematic Review of Emerging Technologies, Methods, Challenges, and Trends,

    N. Syazreen Ahmad, “Recent Advances in WSN-Based Indoor Lo- calization: A Systematic Review of Emerging Technologies, Methods, Challenges, and Trends,”IEEE Access, vol. 12, pp. 180 674–180 714, 2024

    work page 2024

  5. [5]

    RSSI-based location finger- print method for RFID indoor positioning: a review,

    Z. Wei, J. Chen, H. Tang, and H. Zhang, “RSSI-based location finger- print method for RFID indoor positioning: a review,”Nondestructive Testing and Evaluation, vol. 39, no. 1, pp. 3–31, Jan. 2024

    work page 2024

  6. [6]

    RSSI and Machine Learning-Based Indoor Localization Systems for Smart Cities,

    R. M. M. R. Rathnayake, M. W. P. Maduranga, V . Tilwari, and M. B. Dissanayake, “RSSI and Machine Learning-Based Indoor Localization Systems for Smart Cities,”Eng, vol. 4, no. 2, pp. 1468–1494, Jun. 2023

    work page 2023

  7. [7]

    Digital mobile radio towards future generation systems,

    European Commission, “Digital mobile radio towards future generation systems,” European Commission, Directorate-General for the Informa- tion Society and Media, Luxembourg, Final Report, 1999

    work page 1999

  8. [8]

    A Critical Review of the Propagation Models Employed in LoRa Systems,

    J. A. Azevedo and F. Mendonc ¸a, “A Critical Review of the Propagation Models Employed in LoRa Systems,”Sensors, vol. 24, no. 12, p. 3877, Jan. 2024

    work page 2024

  9. [9]

    An Experimental Evaluation of the Reliability of LoRa Long-Range Low-Power Wireless Commu- nication,

    M. Cattani, C. A. Boano, and K. R ¨omer, “An Experimental Evaluation of the Reliability of LoRa Long-Range Low-Power Wireless Commu- nication,”Journal of Sensor and Actuator Networks, vol. 6, no. 2, p. 7, Jun. 2017

    work page 2017

  10. [10]

    The Effect of Temperature and Humidity on Indoor LoRa Propagation Model,

    S. Lavdas, L. Zacharioudakis, A. Khalifeh, and Z. Zinonos, “The Effect of Temperature and Humidity on Indoor LoRa Propagation Model,” in2021 17th International Conference on Distributed Computing in Sensor Systems (DCOSS). Online: IEEE, Jul. 2021, pp. 374–379

    work page 2021

  11. [11]

    Dense Indoor Sensor Net- works: Towards passively sensing human presence with LoRaW AN,

    J. Gr ¨ubel, T. Thrash, L. Aguilar, M. Gath-Morad, D. H ´elal, R. W. Sumner, C. H ¨olscher, and V . R. Schinazi, “Dense Indoor Sensor Net- works: Towards passively sensing human presence with LoRaW AN,” Pervasive and Mobile Computing, vol. 84, p. 101640, Aug. 2022

    work page 2022

  12. [12]

    Advance Path Loss Model for Distance Estimation Using LoRaW AN Network’s Received Signal Strength Indicator (RSSI),

    H. V o, V . Hoang Long Nguyen, V . L. Tran, F. Ferrero, F.-Y . Lee, and M.-H. Tsai, “Advance Path Loss Model for Distance Estimation Using LoRaW AN Network’s Received Signal Strength Indicator (RSSI),” IEEE Access, vol. 12, pp. 83 205–83 216, 2024

    work page 2024

  13. [13]

    A New Approach to Linear Filtering and Prediction Problems,

    R. E. Kalman, “A New Approach to Linear Filtering and Prediction Problems,”Journal of Basic Engineering, vol. 82, no. 1, pp. 35–45, Mar. 1960

    work page 1960

  14. [14]

    A Dynamic Self- Tuning Maximum Correntropy Kalman Filter for Wireless Sensors Networks Positioning Systems,

    T. Liao, K. Hirota, X. Wu, S. Shao, and Y . Dai, “A Dynamic Self- Tuning Maximum Correntropy Kalman Filter for Wireless Sensors Networks Positioning Systems,”Remote Sensing, vol. 14, no. 17, p. 4345, Jan. 2022

    work page 2022

  15. [15]

    Kalman filters explained: Removing noise from RSSI signals,

    W. Bulten, “Kalman filters explained: Removing noise from RSSI signals,” https://www.wouterbulten.nl/posts/kalman-filters-explained- removing-noise-from-rssi-signals/, Oct. 2015

    work page 2015

  16. [16]

    A Comprehensive Data De- scription for LoRaW AN Path Loss Measurements in an Indoor Office Setting: Effects of Environmental Factors,

    N. M. Obiri and K. Van Laerhoven, “A Comprehensive Data De- scription for LoRaW AN Path Loss Measurements in an Indoor Office Setting: Effects of Environmental Factors,”IEEE Access, vol. 13, pp. 83 148–83 170, 2025

    work page 2025

  17. [17]

    Machine-Learning-Based Combined Path Loss and Shadowing Model in LoRaW AN for Energy Efficiency Enhancement,

    M. Gonz ´alez-Palacio, D. Tob ´on-Vallejo, L. M. Sep ´ulveda-Cano, S. R ´ua, and L. B. Le, “Machine-Learning-Based Combined Path Loss and Shadowing Model in LoRaW AN for Energy Efficiency Enhancement,”IEEE Internet of Things Journal, vol. 10, no. 12, pp. 10 725–10 739, Jun. 2023

    work page 2023

  18. [18]

    A Statistical Evaluation of Indoor LoRaW AN Environment-Aware Propagation for 6G: MLR, ANOV A, and Residual Distribution Analysis,

    N. M. Obiri and K. Van Laerhoven, “A Statistical Evaluation of Indoor LoRaW AN Environment-Aware Propagation for 6G: MLR, ANOV A, and Residual Distribution Analysis,” in2025 Joint European Confer- ence on Networks and Communications & 6G Summit (EuCNC/6G Summit), Jun. 2025, pp. 494–499

    work page 2025

  19. [19]

    A weighted hybrid indoor positioning method based on path loss exponent estimation,

    Y . Wang, J. Fu, and Y . Cao, “A weighted hybrid indoor positioning method based on path loss exponent estimation,”Ad Hoc Networks, vol. 166, p. 103684, Jan. 2025

    work page 2025

  20. [20]

    Environment-Aware Indoor Lo- RaW AN Path Loss: Parametric Regression Comparisons, Shadow Fading, and Calibrated Fade Margins,

    N. M. Obiri and K. V . Laerhoven, “Environment-Aware Indoor Lo- RaW AN Path Loss: Parametric Regression Comparisons, Shadow Fading, and Calibrated Fade Margins,” Oct. 2025

    work page 2025