pith. sign in

arxiv: 2607.01882 · v1 · pith:XJJGJZV3new · submitted 2026-07-02 · 💻 cs.NI · eess.SP

CSI Simulation: Why Additive Noise Fails and How to Fix It

Aymen Bouferroum (FUN) , Ildi Alla (uni.lu) , Vincent Lenders (uni.lu) , Valeria Loscri (FUN) This is my paper

Pith reviewed 2026-07-03 04:30 UTC · model grok-4.3

classification 💻 cs.NI eess.SP
keywords channel state informationCSI simulationautomatic gain controladditive white Gaussian noisequantile mappingcopulajamming detectionwireless sensing
0
0 comments X

The pith

Automatic gain control multiplies CSI amplitudes so additive noise cannot reproduce observed distributions.

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

Standard practice assumes that adding noise to recorded channel estimates can simulate new conditions for training CSI models. This assumption fails because automatic gain control in the receiver multiplies amplitudes before they are digitized, creating distributions no additive model can match. The proposed M_QTC learns the actual transformation from measurements using quantile mapping and copulas to reorder subcarriers. It reduces amplitude error eight times and recovers 89 percent of the fidelity lost by additive methods. Classifiers trained on the resulting simulations reach 93 percent of the performance obtained with real data.

Core claim

Automatic gain control compresses the channel estimate multiplicatively before digitization, producing amplitude distributions that no additive noise variance can reproduce. To close the resulting fidelity gap, M_QTC learns the per-subcarrier distribution transformation through quantile mapping, temporal filtering, and copula-based cross-subcarrier reordering. This reduces amplitude error 8-fold and closes 89% of the aggregate fidelity gap across four dimensions. The improvement transfers to downstream tasks where classifiers recover 93% of real-data jamming detection performance.

What carries the argument

M_QTC, the measurement-calibrated model that applies quantile mapping, temporal filtering, and copula-based reordering to capture the multiplicative compression from automatic gain control.

If this is right

  • M_QTC reduces amplitude error 8-fold relative to AWGN simulation.
  • M_QTC closes 89% of the aggregate fidelity gap across four complementary dimensions.
  • Five classifiers from different families trained on M_QTC data recover 93% of real-data jamming detection performance.
  • AWGN-trained classifiers remain near random decision performance on the same task.

Where Pith is reading between the lines

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

  • The method may need per-hardware recalibration if AGC response curves differ across receiver models.
  • Similar multiplicative effects likely appear in other commodity-hardware sensing tasks that rely on amplitude-sensitive features.
  • The copula component could improve simulation of tasks that depend on subcarrier correlations beyond amplitude marginals.

Load-bearing premise

RF jamming on six commodity receivers in two indoor environments produces a representative sample of the multiplicative AGC transformation that holds for other perturbations, hardware, and environments.

What would settle it

Collect new CSI amplitude distributions under a different perturbation such as varying transmit power on the same hardware and test whether M_QTC parameters fitted on jamming data still match the observed distributions.

Figures

Figures reproduced from arXiv: 2607.01882 by Aymen Bouferroum (FUN), Ildi Alla (uni.lu), Valeria Loscri (FUN), Vincent Lenders (uni.lu).

Figure 1
Figure 1. Figure 1: The receiver chain between the antenna and the CSI output. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: MQTC pipeline. Three calibrated stages: quantile mapping (amplitude), AR(1) filtering (temporal), and Iman-Conover copula reordering (cross-subcarrier correlation). illustrated in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Experimental testbed. (a) Controlled room: single ESP32-C6, [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Mean amplitude across all 52 subcarriers at 20 dB. M [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Per-device aggregate fidelity in the laboratory at 20 dB (5 [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
read the original abstract

Channel State Information (CSI) has become a widely used wireless channel sensing modality for applications such as indoor localization, activity recognition, and respiration monitoring. Because collecting labeled data under every target condition is impractical, training CSI-based models often relies on simulated data produced by adding noise or perturbations to recorded channel estimates, most commonly additive white Gaussian noise (AWGN). This practice assumes that the receiver chain between the antenna and the channel estimator is linear and gain-invariant. We test this assumption empirically using RF jamming as a controlled perturbation on 6 commodity receivers across 2 indoor environments. The assumption does not hold. Automatic gain control compresses the channel estimate multiplicatively before digitization, producing amplitude distributions that no additive noise variance can reproduce. To close the resulting fidelity gap, we propose M_QTC, a measurement-calibrated model that learns the per-subcarrier distribution transformation through quantile mapping, temporal filtering, and copula-based cross-subcarrier reordering. M_QTC reduces amplitude error 8-fold and closes 89% of the aggregate fidelity gap across four complementary dimensions. The improvement transfers directly to downstream tasks, where 5 classifiers from different families trained on M_QTC-simulated data recover 93% of real-data jamming detection performance, while AWGN-trained classifiers remain near random decision.

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 paper claims that AWGN-based CSI simulation fails because receiver automatic gain control applies multiplicative compression before digitization, producing amplitude distributions unreachable by additive noise. This is demonstrated via controlled RF jamming experiments on six commodity receivers across two indoor environments. The authors propose M_QTC, a measurement-calibrated model that learns per-subcarrier quantile mappings, applies temporal filtering, and uses copulas for cross-subcarrier reordering; it reduces amplitude error by 8x, closes 89% of the aggregate fidelity gap on four dimensions, and enables downstream classifiers to recover 93% of real-data jamming detection performance versus near-random results for AWGN-trained models.

Significance. If the central empirical findings hold, the work is significant for the CSI sensing community because it identifies a previously under-appreciated hardware nonlinearity that invalidates a standard simulation practice used for localization, activity recognition, and respiration monitoring. The controlled multi-receiver, multi-environment test plus direct transfer evaluation on five classifiers from different families constitute concrete, reproducible evidence of both the problem and a practical fix. The paper earns credit for grounding the critique in real hardware measurements rather than purely theoretical arguments.

major comments (2)
  1. [Abstract / M_QTC calibration] Abstract and methods description of M_QTC: the quantile mapping and copula parameters are fitted exclusively to the RF-jamming calibration set; no experiments evaluate whether these parameters transfer to other perturbations (small-scale fading, activity, respiration) that CSI simulators are intended to model. Because the central claim is that M_QTC 'fixes' CSI simulation in general, this untested generalization is load-bearing.
  2. [Downstream task results] Downstream task evaluation: the reported 93% recovery of real-data performance is presented without error bars, statistical significance tests, or details on data exclusion / train-test splits. Given that the abstract already notes the absence of these elements, the strength of the transfer claim cannot be fully assessed from the provided text.
minor comments (2)
  1. [Fidelity evaluation] The four fidelity dimensions used for the 89% aggregate gap closure are not enumerated or justified in the abstract; a short table or explicit list would improve clarity.
  2. [M_QTC description] Notation for the copula-based reordering step is introduced without a reference to the specific copula family or fitting procedure, which may hinder reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive review and for recognizing the significance of the empirical findings on AGC-induced nonlinearity in CSI simulation. We address each major comment below.

read point-by-point responses

  1. Referee: [Abstract / M_QTC calibration] Abstract and methods description of M_QTC: the quantile mapping and copula parameters are fitted exclusively to the RF-jamming calibration set; no experiments evaluate whether these parameters transfer to other perturbations (small-scale fading, activity, respiration) that CSI simulators are intended to model. Because the central claim is that M_QTC 'fixes' CSI simulation in general, this untested generalization is load-bearing.

    Authors: The M_QTC calibration uses the RF-jamming dataset because jamming provides a controlled, repeatable perturbation that isolates the multiplicative AGC compression effect. The quantile mapping and copula components are designed to capture this hardware nonlinearity, which is independent of the specific perturbation source. Nevertheless, we agree that the absence of explicit transfer experiments on other CSI perturbations (e.g., small-scale fading or respiration) limits the strength of the generalization claim. In the revised manuscript we will add a dedicated limitations subsection clarifying the calibration scope and discussing expected transferability, along with any feasible additional analysis using existing data. revision: partial

  2. Referee: [Downstream task results] Downstream task evaluation: the reported 93% recovery of real-data performance is presented without error bars, statistical significance tests, or details on data exclusion / train-test splits. Given that the abstract already notes the absence of these elements, the strength of the transfer claim cannot be fully assessed from the provided text.

    Authors: We acknowledge that the downstream evaluation would benefit from additional statistical detail. In the revised manuscript we will include error bars on the reported performance metrics, conduct and report appropriate statistical significance tests, and provide complete information on train-test splits, data exclusion criteria, and classifier training procedures. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical validation independent of fitted parameters

full rationale

The paper's central claims rest on direct empirical comparison between real RF-jamming CSI measurements and AWGN-simulated counterparts, followed by fitting M_QTC parameters to the same measurements and evaluating downstream classifier performance on held-out real data. No derivation reduces by construction to its inputs; the amplitude-error reduction and fidelity-gap closure are measured outcomes, not tautological renamings or self-referential predictions. No self-citations, uniqueness theorems, or ansatzes are invoked. The calibration step introduces data-driven fitting, but this does not collapse the reported results into the fitting procedure itself.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Review performed on abstract only; exact free parameters inside M_QTC (e.g., copula parameters, filter coefficients) are not stated. The ledger therefore records only the high-level modeling choices visible in the abstract.

free parameters (1)
  • quantile mapping and copula parameters
    Learned from the calibration measurements on the six receivers to match observed amplitude distributions; exact values and fitting procedure not provided in abstract.
axioms (1)
  • domain assumption RF jamming constitutes a controlled perturbation whose effect on CSI is representative of the general AGC behavior under other real-world conditions.
    Invoked when the authors treat the jamming results as evidence that AWGN is fundamentally insufficient.

pith-pipeline@v0.9.1-grok · 5779 in / 1519 out tokens · 41762 ms · 2026-07-03T04:30:32.748339+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

38 extracted references · 1 canonical work pages · 1 internal anchor

  1. [1]

    A survey on Wi-Fi sensing generalizability: Taxonomy, techniques, datasets, and future research prospects,

    F. Wang, T. Zhang, W. Xi, H. Ding, G. Wang, D. Zhang, Y . Cui, F. Liu, J. Han, J. Xu, and T. X. Han, “A survey on Wi-Fi sensing generalizability: Taxonomy, techniques, datasets, and future research prospects,” IEEE Commun. Surveys Tuts. , 2026, early access

    2026

  2. [2]

    Sec5GLoc: Securing 5G indoor localization via adversary-resilient deep learning architecture,

    I. Alla and V . Loscri, “Sec5GLoc: Securing 5G indoor localization via adversary-resilient deep learning architecture,” in Proc. IEEE Conf. Commun. Netw. Security (CNS) , 2025, pp. 1–9

    2025

  3. [3]

    Tool release: Gath- ering 802.11n traces with channel state information,

    D. Halperin, W. Hu, A. Sheth, and D. Wetherall, “Tool release: Gath- ering 802.11n traces with channel state information,” ACM SIGCOMM Comput. Commun. Rev., vol. 41, no. 1, p. 53, 2011

    2011

  4. [4]

    Free your CSI: A channel state information extraction platform for modern Wi-Fi chipsets,

    F. Gringoli, M. Schulz, J. Link, and M. Hollick, “Free your CSI: A channel state information extraction platform for modern Wi-Fi chipsets,” inProc. 13th Int. Workshop Wireless Netw. Testbeds Exp. Eval. Characterization (WiNTECH), 2019

    2019

  5. [5]

    ESP-CSI: Wi-Fi channel state information application,

    Espressif Systems, “ESP-CSI: Wi-Fi channel state information application,” 2024, GitHub repository. [Online]. Available: https: //github.com/espressif/esp-csi

    2024

  6. [6]

    CrossSense: Towards cross-site and large-scale WiFi sensing,

    J. Zhang, Z. Tang, M. Li, D. Fang, P. Nurmi, and Z. Wang, “CrossSense: Towards cross-site and large-scale WiFi sensing,” in Proc. 24th Annu. Int. Conf. Mobile Comput. Netw. (MobiCom) , 2018, pp. 305–320

    2018

  7. [7]

    AirFi: Empowering WiFi-based passive human gesture recognition to unseen environment via domain generalization,

    D. Wang, J. Yang, W. Cui, L. Xie, and S. Sun, “AirFi: Empowering WiFi-based passive human gesture recognition to unseen environment via domain generalization,” IEEE Trans. Mobile Comput. , 2022

    2022

  8. [8]

    Data augmentation techniques for cross-domain WiFi CSI-based human activity recognition,

    J. Strohmayer and M. Kampel, “Data augmentation techniques for cross-domain WiFi CSI-based human activity recognition,” in Artificial Intelligence Applications and Innovations (AIAI 2024), IFIP AICT , vol. 711, 2024, pp. 44–57

    2024

  9. [9]

    Wireless channel aware data augmentation methods for deep learning-based indoor localization,

    O. G. Serbetci, D. Burghal, and A. F. Molisch, “Wireless channel aware data augmentation methods for deep learning-based indoor localization,” in Proc. IEEE Global Commun. Conf. (GLOBECOM) , 2023

    2023

  10. [10]

    Toward 5G NR high-precision indoor positioning via channel frequency response: A new paradigm and dataset generation method,

    K. Gao, H. Wang, H. Lv, and W. Liu, “Toward 5G NR high-precision indoor positioning via channel frequency response: A new paradigm and dataset generation method,” IEEE J. Sel. Areas Commun., vol. 40, no. 7, 2022

    2022

  11. [11]

    Goldsmith, Wireless Communications

    A. Goldsmith, Wireless Communications. Cambridge Univ. Press, 2005

    2005

  12. [12]

    JamRF: Jamming RF framework,

    Technology Innovation Institute, “JamRF: Jamming RF framework,” 2022, GitHub repository (archived). [Online]. Available: https: //github.com/tiiuae/jamrf

    2022

  13. [13]

    CSI4Free: GAN-augmented mmWave CSI for improved pose classification,

    N. N. Bhat, R. Berkvens, and J. Famaey, “CSI4Free: GAN-augmented mmWave CSI for improved pose classification,” in Proc. IEEE 4th Int. Symp. Joint Commun. Sensing (JC&S) , 2024, pp. 1–6

    2024

  14. [14]

    Generative AI enabled robust data augmentation for wireless sensing in ISAC networks,

    J. Wang, C. Zhao, H. Du, G. Sun, J. Kang, S. Mao, D. Niyato, and D. I. Kim, “Generative AI enabled robust data augmentation for wireless sensing in ISAC networks,” IEEE J. Sel. Areas Commun. , 2025

    2025

  15. [15]

    RFBoost: Understanding and boosting deep WiFi sensing via physical data augmentation,

    W. Hou and C. Wu, “RFBoost: Understanding and boosting deep WiFi sensing via physical data augmentation,” Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., vol. 8, no. 2, 2024

    2024

  16. [16]

    T. S. Rappaport, Wireless Communications: Principles and Practice , 2nd ed. Prentice Hall, 2002

    2002

  17. [17]

    Optimal preprocessing of WiFi CSI for sensing applications,

    V . V . Ratnam, H. Chen, H.-H. Chang, A. Sehgal, and J. C. Zhang, “Optimal preprocessing of WiFi CSI for sensing applications,” IEEE Trans. Wireless Commun., vol. 23, no. 9, pp. 10 820–10 833, 2024

    2024

  18. [18]

    RSSI-CSI measurement and variation mitigation with commodity WiFi device,

    B. Wei, H. Song, J. Katto, and T. Kikkawa, “RSSI-CSI measurement and variation mitigation with commodity WiFi device,” IEEE Internet Things J., vol. 10, no. 7, pp. 6249–6258, 2023

    2023

  19. [19]

    Eliminating the barriers: Demystifying Wi-Fi baseband design and introducing the PicoScenes Wi-Fi sensing platform,

    Z. Jiang, T. H. Luan, X. Ren, D. Lv, H. Hao, J. Wang, K. Zhao, W. Xi, Y . Xu, and R. Li, “Eliminating the barriers: Demystifying Wi-Fi baseband design and introducing the PicoScenes Wi-Fi sensing platform,” IEEE Internet Things J. , vol. 9, no. 6, pp. 4476–4496, 2022

    2022

  20. [20]

    Same signal, different story: Demystifying receiver effects in Wi-Fi channel state information,

    F. Portner, F. Gringoli, M. Hollick, and A. Asadi, “Same signal, different story: Demystifying receiver effects in Wi-Fi channel state information,” IEEE Internet Things J. , 2026

    2026

  21. [21]

    The feasibility of launching and detecting jamming attacks in wireless networks,

    W. Xu, W. Trappe, Y . Zhang, and T. Wood, “The feasibility of launching and detecting jamming attacks in wireless networks,” in Proc. 6th ACM Int. Symp. Mobile Ad Hoc Netw. Comput. (MobiHoc) , 2005, pp. 46–57

    2005

  22. [22]

    Jamming attacks and anti-jamming strategies in wireless networks: A comprehensive survey,

    H. Pirayesh and H. Zeng, “Jamming attacks and anti-jamming strategies in wireless networks: A comprehensive survey,”IEEE Commun. Surveys Tuts., vol. 24, no. 2, pp. 767–809, 2022

    2022

  23. [23]

    Channel state information analysis for jamming attack detection in static and dynamic UA V networks – an experimental study,

    P. Mykytyn, R. Chitauro, Z. Dyka, and P. Langendoerfer, “Channel state information analysis for jamming attack detection in static and dynamic UA V networks – an experimental study,” inProc. 21st Int. Conf. Distrib. Comput. Smart Syst. Internet Things (DCOSS-IoT) , 2025, pp. 322–327

    2025

  24. [24]

    JamShield: A machine learning detection system for over-the-air jam- ming attacks,

    I. Panitsas, Y . Yigit, L. Tassiulas, L. A. Maglaras, and B. Canberk, “JamShield: A machine learning detection system for over-the-air jam- ming attacks,” inProc. IEEE Int. Conf. Commun. (ICC), 2025, pp. 1067– 1072

    2025

  25. [25]

    CITADEL: CSI-Based Jamming Detection and Open-Set Classification for IIoT Networks

    A. Bouferroum, I. Alla, V . Loscri, A. Benslimane, and V . Lenders, “CITADEL: CSI-based jamming detection and open-set classification for IIoT networks,” arXiv preprint arXiv:2606.22939 , 2026. [Online]. Available: https://arxiv.org/abs/2606.22939

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  26. [26]

    Bridging the sim-to-real gap in RF localization with large-scale synthetic pretraining,

    A. Manukyan, R. Mkrtchyan, A. Saribekyan, T. P. Raptis, and H. Khachatrian, “Bridging the sim-to-real gap in RF localization with large-scale synthetic pretraining,” Inf. Fusion, vol. 130, p. 104104, 2026

    2026

  27. [27]

    Train- ing data augmentation for deep learning radio frequency systems,

    W. H. Clark, S. Hauser, W. C. Headley, and A. J. Michaels, “Train- ing data augmentation for deep learning radio frequency systems,” J. Defense Model. Simul. , vol. 18, no. 3, pp. 217–237, 2021

    2021

  28. [28]

    Finding a needle in a (spectrum) haystack: Multi-band multi-device radio finger- printing,

    I. Alla, M. Zhang, J. Ashdown, V . Loscri, and F. Restuccia, “Finding a needle in a (spectrum) haystack: Multi-band multi-device radio finger- printing,” Comput. Netw., vol. 280, p. 112148, 2026

    2026

  29. [29]

    Context-aware predictive coding: A representation learning framework for WiFi sens- ing,

    B. Barahimi, H. Tabassum, M. Omer, and O. Waqar, “Context-aware predictive coding: A representation learning framework for WiFi sens- ing,” IEEE Open J. Commun. Soc. , 2024

    2024

  30. [30]

    DF-WiSLR: Device-free Wi-Fi-based sign language recognition,

    H. F. Thariq Ahmed, H. Ahmad, K. Narasingamurthi, H. Harkat, and S. K. Phang, “DF-WiSLR: Device-free Wi-Fi-based sign language recognition,” Pervasive Mobile Comput., vol. 69, p. 101289, 2020

    2020

  31. [31]

    Exposing the CSI: A systematic investigation of CSI-based Wi-Fi sensing capabilities and limitations,

    M. Cominelli, F. Gringoli, and F. Restuccia, “Exposing the CSI: A systematic investigation of CSI-based Wi-Fi sensing capabilities and limitations,” in Proc. IEEE Int. Conf. Pervasive Comput. Commun. (PerCom), 2023, pp. 81–90

    2023

  32. [32]

    Bias correction of GCM precipitation by quantile mapping: How well do methods preserve changes in quantiles and extremes?

    A. J. Cannon, S. R. Sobie, and T. Q. Murdock, “Bias correction of GCM precipitation by quantile mapping: How well do methods preserve changes in quantiles and extremes?” J. Climate , vol. 28, no. 17, pp. 6938–6959, 2015

    2015

  33. [33]

    G. E. P. Box, G. M. Jenkins, G. C. Reinsel, and G. M. Ljung, Time Series Analysis: Forecasting and Control , 5th ed. Wiley, 2015

    2015

  34. [34]

    A distribution-free approach to inducing rank correlation among input variables,

    R. L. Iman and W. J. Conover, “A distribution-free approach to inducing rank correlation among input variables,” Commun. Statist. – Simul. Comput., vol. 11, no. 3, pp. 311–334, 1982

    1982

  35. [35]

    The earth mover’s distance as a metric for image retrieval,

    Y . Rubner, C. Tomasi, and L. J. Guibas, “The earth mover’s distance as a metric for image retrieval,” Int. J. Comput. Vis. , vol. 40, no. 2, pp. 99–121, 2000

    2000

  36. [36]

    K. V . Mardia and P. E. Jupp, Directional Statistics. Wiley, 1999

    1999

  37. [37]

    SignFi: Sign lan- guage recognition using WiFi,

    Y . Ma, G. Zhou, S. Wang, H. Zhao, and W. Jung, “SignFi: Sign lan- guage recognition using WiFi,” in Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., vol. 2, no. 1, 2018, pp. 1–21

    2018

  38. [38]

    Zero-effort cross-domain gesture recognition with Wi-Fi,

    Y . Zheng, Y . Zhang, K. Qian, G. Zhang, Y . Liu, C. Wu, and Z. Yang, “Zero-effort cross-domain gesture recognition with Wi-Fi,” in Proc. ACM MobiSys, 2019, pp. 313–325

    2019