Pith. sign in

REVIEW 5 major objections 6 minor 17 references

A Privacy-Preserving Indoor Localization System based on Hierarchical Federated Learning

T0 review · 5 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A hierarchical federated learning system for indoor localization achieves nearly the same accuracy as a centralized model while keeping raw Wi-Fi data on each floor.

desk verdict The headline MAE numbers are unreliable because Eq. (3) zeroes out every sample with a misclassified building/floor, so the 3.44% FL-vs-CL claim rests on a flawed metric; the paper is an incremental hierarchical FL benchmark that needs a corrected evaluation before it can be trusted. read the letter →

arxiv 2507.01581 v1 pith:DPGPSUJ5 submitted 2025-07-02 cs.LG cs.CReess.SP

Masood Jan , Wafa Njima , Xun Zhang This is my paper · ORCID
classification cs.LGcs.CReess.SP
keywords hierarchicalfederatedlearningindoorlocalizationprivacypreservationdeepneuralnetworkUJIIndoorLocRSSIfingerprintingmeanabsoluteerrorcentralized
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to show that a hierarchical federated learning (FL) system can localize users indoors almost as accurately as a centralized deep neural network, while keeping raw Wi-Fi signal data on each floor and sharing only model weights. Using the UJIIndoorLoc public dataset, it reports a validation mean absolute error of 10.86 meters for FL versus 10.81 meters for centralized learning, a difference it frames as a marginal 3.44 percent increase. The point of the comparison is that the privacy, bandwidth, and server-reliability gains of FL can be bought with only a small accuracy cost, making FL a practical alternative for IoT location services. The paper also benchmarks against K-nearest neighbors and argues that machine-learning approaches beat it on the metrics that matter for real deployments.

What carries the argument

The load-bearing mechanism is the hierarchical aggregation of model weights rather than data. Each floor independently trains a DNN with three ReLU hidden layers, then each building combines its floors' weights into a regional model, and finally the global server aggregates the regional weights as a weighted average proportional to dataset sizes, following the update rule $W_{r+1} = \frac{1}{\sum_c |D_c|} \sum_c |D_c| W_c^r$. Two data transformations support the pipeline: deletion of access points present in more than 98 percent of missing measurements, leaving 248 of 520 original APs, and a powered RSSI normalization that maps signal strengths to positive values before training. The stated performance measure is a validation mean absolute error computed over the training loop, with the central model used as the direct benchmark under identical architecture and hyperparameters.

What would settle it

Re-run the same trained FL and centralized models on the UJIIndoorLoc validation set and compute the ordinary mean Euclidean distance between predicted and true longitude-latitude coordinates without any building or floor gates, then check whether the 10.86 versus 10.81 meter gap still stands and whether the 3.44 percent increase remains consistent when the metric is changed.

Watch

Extended reading notes

Core claim

The central claim is that a three-tier hierarchical federated learning scheme—floor models trained locally, building-level regional models aggregated from those floors, and a global model aggregated from the regions—achieves localization performance close to a centralized model on the same DNN architecture. When both models are trained for 1000 epochs on the UJIIndoorLoc data, the validation error of the FL central model is 10.86 meters and the centralized model is 10.81 meters, which the paper interprets as a 3.44 percent relative increase. Beyond the raw numbers, the paper argues that this small accuracy gap is accompanied by substantial practical benefits: raw RSSI measurements never leave the floor, only weight updates travel over the network, and no single server holds all the data. The study therefore asserts that privacy-preserving federated localization is not a compromise that sacrifices accuracy but a viable deployment option for real indoor environments.

Load-bearing premise

The reported accuracy numbers are trustworthy only if the metric in Eq. (3), which zeroes out the error whenever the predicted building or floor is wrong, is a valid way to measure localization error; if plain distance error is the intended measure, all the reported errors and the FL-versus-centralized comparison need to be recomputed.

Editorial extensions

If this is right

  • If the reported parity holds, a building or campus can run indoor localization without aggregating raw fingerprints, so user movement data is not exposed to a central authority.
  • The communication cost drops from transmitting full datasets to transmitting only weight vectors, which matters when many floors or buildings participate in a city-scale deployment.
  • Because the same DNN architecture serves as floor, regional, and global model, the approach extends naturally to other radio or optical signal sources, and the paper names visible-light communication as the next target.
  • The measured gap of about five centimeters of validation error suggests that privacy-preserving deployment choices do not force a meaningful accuracy sacrifice in this setting.
  • The comparison to KNN indicates that learned models, FL included, produce lower localization error than a classic fingerprint-matching baseline, reinforcing the motivation for federated training.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The evaluation metric in Eq. (3) multiplies the Euclidean distance error by Kronecker deltas for building and floor classification, so any fingerprint whose building or floor is mispredicted contributes zero to the sum; the reported 10.86 and 10.81 meter numbers are therefore not plain mean distance errors in the usual sense, and the gap between FL and centralized learning could change if the metr
  • The paper itself notes that its KNN result of 12.81 meters differs from a published UJIIndoorLoc KNN baseline near 8.5 meters, so the KNN column in the benchmark should be treated as implementation-specific rather than a fixed property of the algorithm.
  • A natural testable extension is to retrain the identical models on the same validation set while computing the ordinary mean Euclidean distance between predicted and true coordinates, and to report the error separately for correct-building, correct-floor, and misclassified samples, which would show whether the hierarchical FL advantage survives under stricter scoring.
  • The validation set was collected four months after the training set, giving a realistic temporal-displacement test, yet the paper reports only a single validation pass; a stronger claim would track validation error across multiple rounds and show that the 3.44 percent gap is stable rather than a snapshot.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper proposes a three-tier hierarchical federated learning (FL) scheme for Wi-Fi RSSI-based indoor localization. Floor-level DNNs are trained locally, aggregated into building-level regional models, and then into a global model, using the UJIIndoorLoc dataset. The authors report a validation MAE of 10.86 m for FL versus 10.81 m for a centralized DNN with the same architecture and 12.81 m for KNN, and conclude that FL performs comparably to centralized learning while preserving data privacy and reducing bandwidth. The central numerical claim is currently undermined by the evaluation metric in Eq. (3), which multiplies the Euclidean distance error by Kronecker deltas for building and floor, so samples with misclassified building or floor contribute zero error.

Significance. If the numerical results survive correction, the hierarchical FL design is a reasonable way to reduce communication and keep raw fingerprints local, and the use of a public benchmark with a temporally displaced validation set is a strength. The central quantitative claim, that FL is within 3.44% of centralized learning, is not supported as written because the reported errors are not true mean distances, the building/floor indicators in the metric are undefined for the stated architecture, and no uncertainty quantification is provided. The paper also does not currently provide enough algorithmic detail to reproduce the regional and global training procedure. The idea is worth pursuing, but the evidence presented here is insufficient for publication in its present form.

major comments (5)
  1. [§III-A, Eq. (3)] Equation (3) defines the reported error as (1/N) Σ δ(B_i, B̂_i) δ(F_i, F̂_i) √((x_i−x̂_i)²+(y_i−ŷ_i)²), so any sample whose predicted building or floor does not match the ground truth contributes zero error to the sum while still counting in N. The prose immediately after the equation states that the MAE 'reflects the average distance between the predicted and actual geographic coordinates,' which is not what the formula computes. Moreover, the architecture described in §III-B outputs only a two-dimensional coordinate vector and has no building/floor classification head, so B̂_i and F̂_i are not defined by the stated model. Because Table IV and the 3.44% conclusion in §VI are based on this metric, the main quantitative claim is unsupported. Please report plain mean Euclidean distance error, with building/floor misclassification rates given separately, and re-run the FL/CL comparison.
  2. [§V, Table IV] The reported validation errors of 10.86 m and 10.81 m come from single runs; no seeds, repeated trials, or confidence intervals are given. The 0.05 m gap underlying the '3.44% increase' claim is far smaller than the run-to-run variability typical of DNN training, so it cannot be interpreted without repeated runs and a significance test. Please provide results over multiple random seeds and report mean ± standard deviation or confidence intervals.
  3. [§III-B, Algorithm 1] Algorithm 1 is ambiguous about what happens at the regional and global levels. Lines 12–16 state that the building model is trained using the aggregated floor weights, and lines 17–20 state that the global model is trained using the aggregated regional weights; standard FederatedAveraging performs only weighted averaging and does not introduce an additional training pass on aggregated weights. If additional training is performed, the data used for that training and the privacy guarantees need to be specified. This is essential for reproducibility and for the claim that raw data never leaves the floors.
  4. [§II and §III-A] The system treats each floor as a single FL user ('We considered each floor as a single user due to the limitation of data'), so training data for an entire floor is pooled before the first local update. The privacy guarantee is therefore at the floor level, not the device level: an honest-but-curious floor server could see all individual fingerprints on that floor. The abstract's statement that 'the data remains localized and confined within each device' and the related privacy framing should be revised to match the actual system, or the system should be evaluated with per-device clients.
  5. [§IV, Table IV] The KNN comparison is not trustworthy as reported. The text states that the KNN baseline for UJIIndoorLoc is about 8.5 m in the literature while their implementation produces 12.81 m, and attributes the discrepancy to implementation differences. Without a precise description of the KNN configuration (number of neighbors, distance measure, preprocessing, and whether building/floor classification is used), the benchmark does not support the claim that FL outperforms KNN. Either align the KNN setup with the published baseline or report full configuration details and re-run.
minor comments (6)
  1. [§III-A, Table III] Table III lists Adam exponential decay rates β1 = 0.1 and β2 = 0.99; the standard Adam defaults are 0.9 and 0.999. Please clarify whether this is a typo or an intentional non-standard configuration.
  2. [Table IV] The table header uses 'MDE' while the text consistently uses 'MAE'; please make the terminology uniform and define the acronym at first use.
  3. [§III-A] The sentence 'deleting all those columns where there is no data recorded at all i-e; cell values with 100 dBm' is ambiguous: it could mean columns where every value is 100 dBm, or columns containing any 100 dBm value. Please specify precisely the deletion criterion.
  4. [§III-A] The choices τ = 0.98, β = e, and minRSSI = −105 dBm are introduced without sensitivity analysis or citation beyond the general method in [15]. Since these choices affect all models equally, this is not a fatal issue, but a brief justification or a reference to a specific configuration would improve reproducibility.
  5. [§V, Fig. 3] Figure 3 should include axis labels and a legend so the reader can identify which curve corresponds to FL and which to CL; the current description does not make this clear.
  6. [§III-B, Algorithm 1] The notation W^c_{r,k}, β_c, and B_c is used in the algorithm but not defined; in particular, B_c appears to be a data batch rather than a building, which is confusing given that B is also used for buildings.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the paper reports an empirical comparison on a public dataset; no fitted parameter is renamed as a prediction, and the only self-citation is non-load-bearing related work.

full rationale

The paper does not present a derivation chain in which a predicted quantity is defined in terms of itself or in which a fitted parameter is renamed as a prediction. The proposed hierarchical FL architecture (Algorithm 1, Eq. 4) and the centralized comparison are described explicitly, and the reported validation MAE values (10.86 m for FL, 10.81 m for CL, 12.81 m for KNN in Table IV) are empirical results on the public UJIIndoorLoc dataset, not outputs of a self-consistent mathematical derivation. The preprocessing choices, beta=e in Eq. (2) and the AP threshold tau=0.98, are either cited to prior external work [15] or stated as hand-set thresholds; they are not fitted to minimize the reported validation MAE, so no fitted-input-called-prediction pattern is present. The only self-citation, [12] (Etiabi, Njima, Amhoud), is used to describe a different existing FL approach and is not load-bearing: the present paper's architecture, aggregation rule, and experimental settings are self-contained. The reviewer-identified concern about Eq. (3), in which building/floor Kronecker deltas zero out the distance error for misclassified samples, is a correctness and metric-validity threat that affects whether the 3.44% comparison is meaningful, but it is not circularity under the defined patterns: the metric is not an input used to derive a predicted output, and no claim reduces to its own definition. Therefore the correct circularity finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The paper introduces no new physical or mathematical entities. It relies on hand-chosen preprocessing thresholds and on the standard FedAvg assumption that weight averaging produces a converged global model.

free parameters (4)
  • minRSSI = -105 dBm
    Chosen as a floor slightly below the observed minimum RSSI of -104 dBm to represent missing '100 dBm' values; used in Eq. (2) transformation.
  • tau = 0.98
    Hand-picked threshold for discarding access points missing in over 98% of measurements; reduces the feature set from 465 to 248 APs.
  • beta = e
    Exponent in the powered RSSI representation Eq. (2), taken from prior work [15]; not tuned on the data, but controls the transformation and affects DNN inputs.
  • Adam learning rate = 0.0005
    Optimizer step size chosen without reported tuning; standard hyperparameter affecting convergence and final error.
assumptions (3)
  • domain assumption Each floor is treated as a single federated client with its full local dataset.
    Section II states 'We considered each floor as a single user due to the limitation of data.' This turns the 12-floor setup into 12 clients and is central to the FL framing.
  • domain assumption The UJIIndoorLoc training set and the validation set recorded four months later are representative of real-world indoor Wi-Fi conditions.
    Section III-A describes the validation set as measured 4 months after training, which the paper uses to claim generalization.
  • domain assumption Federated averaging of model weights (Eq. 4) converges to a good global model in this setting.
    The paper assumes FedAvg weight averaging across floor and building models yields a usable global model without raw data sharing.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Privacy-Preserving Indoor Localization System based on Hierarchical Federated Learning." pith.science (2026). https://pith.science/paper/DPGPSUJ5

@misc{pith2026250701581,
  author       = {Pith},
  title        = {Pith review of: A Privacy-Preserving Indoor Localization System based on Hierarchical Federated Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DPGPSUJ5}},
  note         = {Machine review of arXiv:2507.01581}
}
read the original abstract

Location information serves as the fundamental element for numerous Internet of Things (IoT) applications. Traditional indoor localization techniques often produce significant errors and raise privacy concerns due to centralized data collection. In response, Machine Learning (ML) techniques offer promising solutions by capturing indoor environment variations. However, they typically require central data aggregation, leading to privacy, bandwidth, and server reliability issues. To overcome these challenges, in this paper, we propose a Federated Learning (FL)-based approach for dynamic indoor localization using a Deep Neural Network (DNN) model. Experimental results show that FL has the nearby performance to Centralized Model (CL) while keeping the data privacy, bandwidth efficiency and server reliability. This research demonstrates that our proposed FL approach provides a viable solution for privacy-enhanced indoor localization, paving the way for advancements in secure and efficient indoor localization systems.

Figures

Figures reproduced from arXiv: 2507.01581 by the authors.

Figure 1
Figure 1. Hierarchical representation of FL based localization process. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. illustrates the outcomes of our code execution for training the Building Models, which comprised 100 rounds and 10 epochs. Each epoch ran for 10 iterations per floor before advancing to the subsequent round. The total number of epochs will be; Training Loss: 10 epochs per floor * 4 floors per round * 100 rounds = 4000 epochs [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Comparison of FL & CL Central Models. The reason behind the CL central model’s superior perfor￾mance compared to the FL central model lies in its centralized learning approach. In CL, all data is aggregated and processed on a single central server, allowing the model to access a comprehensive dataset, which leads to better model conver￾gence and more robust generalization. On the other hand, the FL approach, employe… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 16 canonical work pages

  1. [12]

    Federated Learning based Hierarchical 3D Indoor Localization,

    Y . Etiabi, W. Njima, and E. M. Amhoud, “Federated Learning based Hierarchical 3D Indoor Localization,” in 2023 IEEE Wireless Commu- nications and Networking Conference (WCNC) . IEEE, 2023, pp. 1–6

    work page 2023

  2. [1]

    Precise Posi- tioning of Marketing and Behavior Intentions of Location-based Mobile Commerce in the Internet of Things,

    Y .-T. Tsai, S.-C. Wang, K.-Q. Yan, and C.-M. Chang, “Precise Posi- tioning of Marketing and Behavior Intentions of Location-based Mobile Commerce in the Internet of Things,” Symmetry, vol. 9, no. 8, p. 139, 2017

    work page 2017

  3. [2]

    Hinch, Outdoor Navigation with GPS

    S. Hinch, Outdoor Navigation with GPS . Wilderness press, 2010

    work page 2010

  4. [3]

    Deep Learning based Data Recovery for Localization,

    W. Njima, M. Chafii, A. Nimr, and G. Fettweis, “Deep Learning based Data Recovery for Localization,” IEEE Access , vol. 8, pp. 175 741– 175 752, 2020

    work page 2020

  5. [4]

    Mobile Positioning using Wireless Networks: Possibilities and Fundamental Limitations based on Available Wireless Network Measurements,

    F. Gustafsson and F. Gunnarsson, “Mobile Positioning using Wireless Networks: Possibilities and Fundamental Limitations based on Available Wireless Network Measurements,” IEEE Signal processing magazine , vol. 22, no. 4, pp. 41–53, 2005

    work page 2005

  6. [5]

    Network-based Wireless Location: Challenges Faced in Developing Techniques for Accurate Wireless Location Information,

    A. H. Sayed, A. Tarighat, and N. Khajehnouri, “Network-based Wireless Location: Challenges Faced in Developing Techniques for Accurate Wireless Location Information,” IEEE signal processing magazine , vol. 22, no. 4, pp. 24–40, 2005

    work page 2005

  7. [6]

    FedLoc: Federated Learning Framework for Data-Driven Cooperative Localization and Location Data Processing,

    F. Yin, Z. Lin, Q. Kong, Y . Xu, D. Li, S. Theodoridis, and S. R. Cui, “FedLoc: Federated Learning Framework for Data-Driven Cooperative Localization and Location Data Processing,” IEEE Open Journal of Signal Processing, vol. 1, pp. 187–215, 2020

    work page 2020

  8. [7]

    A Survey of Machine Learning for Indoor Positioning,

    A. Nessa, B. Adhikari, F. Hussain, and X. N. Fernando, “A Survey of Machine Learning for Indoor Positioning,” IEEE access , vol. 8, pp. 214 945–214 965, 2020

    work page 2020

Show all 17 references
  1. [8]

    Toward Smart Security Enhancement of Federated Learning Networks,

    J. Tan, Y .-C. Liang, N. C. Luong, and D. Niyato, “Toward Smart Security Enhancement of Federated Learning Networks,” IEEE Network, vol. 35, no. 1, pp. 340–347, 2020

    2020

  2. [9]

    How Apple and Google are Enabling Covid-19 Blue- tooth Contact-Tracing,

    A. Greenberg, “How Apple and Google are Enabling Covid-19 Blue- tooth Contact-Tracing,” Wired, 2020

    2020

  3. [10]

    Federated learning for internet of things: A comprehensive survey,

    D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, and H. V . Poor, “Federated learning for internet of things: A comprehensive survey,” IEEE Communications Surveys & Tutorials , vol. 23, no. 3, pp. 1622–1658, 2021

    2021

  4. [11]

    Confidential- ity Preserved Federated Learning for Indoor Localization Using Wi-Fi Fingerprinting,

    R. Kumar, R. Popli, V . Khullar, I. Kansal, and A. Sharma, “Confidential- ity Preserved Federated Learning for Indoor Localization Using Wi-Fi Fingerprinting,” Buildings, vol. 13, no. 8, p. 2048, 2023

    2023

  5. [13]

    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, J. P. Avariento, T. J. Arnau, M. Benedito-Bordonau, and J. Huerta, “UJIIndoorLoc: A New Multi-Building and Multi-Floor Database for WLAN Fingerprint-based Indoor Localization Problems,” in 2014 international conference on in...

    2014

  6. [14]

    The Future of Digital Health with Federated Learning,

    N. Rieke, J. Hancox, W. Li, F. Milletari, H. R. Roth, S. Albarqouni, S. Bakas, M. N. Galtier, B. A. Landman, K. Maier-Hein et al. , “The Future of Digital Health with Federated Learning,” NPJ digital medicine, vol. 3, no. 1, pp. 1–7, 2020

    2020

  7. [15]

    Comprehensive Analysis of Distance and Similarity Measures for Wi- Fi Fingerprinting Indoor Positioning Systems,

    J. Torres-Sospedra, R. Montoliu, S. Trilles, ´O. Belmonte, and J. Huerta, “Comprehensive Analysis of Distance and Similarity Measures for Wi- Fi Fingerprinting Indoor Positioning Systems,” Expert Systems with Applications, vol. 42, no. 23, pp. 9263–9278, 2015

    2015

  8. [16]

    A Comprehensive Survey of Deep Learning Models based on Keras Framework,

    B. T. Chicho and A. B. Sallow, “A Comprehensive Survey of Deep Learning Models based on Keras Framework,” Journal of Soft Comput- ing and Data Mining , vol. 2, no. 2, pp. 49–62, 2021

    2021

  9. [17]

    A realistic evaluation of indoor positioning systems based on wi-fi fingerprinting: The 2015 evaal–etri competition,

    J. Torres-Sospedra, A. Moreira, S. Knauth, R. Berkvens, R. Montoliu, O. Belmonte, S. Trilles, M. Joao Nicolau, F. Meneses, A. Costa et al. , “A realistic evaluation of indoor positioning systems based on wi-fi fingerprinting: The 2015 evaal–etri competition,” Journal of ambien...

    2015

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.