Pith. sign in

REVIEW 3 major objections 8 minor 16 references

WiFi-based Real-time Breathing and Heart Rate Monitoring during Sleep

T0 review · 3 major / 8 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A pair of commodity WiFi devices can measure breathing and heart rate live during sleep, reporting average errors of 0.575 bpm for breathing and 3.9 bpm for heart rate.

desk verdict Plausible and honest WiFi sensing work, but the reported accuracy numbers are lab-specific; the transferable claim needs much stronger evidence. read the letter →

arxiv 1908.05108 v1 pith:JOAFEHW7 submitted 2019-08-14 cs.HC eess.SP

Yu Gu , Xiang Zhang , Zhi Liu , Fuji Ren This is my paper · ORCID
classification cs.HCeess.SP
keywords WiFisensingchannelstateinformationbreathingrateheartsleepmonitoringFresnelzonereal-timevitalsigns
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

This paper reports that one transmitting and one receiving WiFi device, both commodity hardware with omnidirectional antennas, can monitor an individual's breathing rate and heart rate continuously during sleep, in real time and across four sleeping postures. The claim is that the tiny chest and abdomen motions from breathing and heartbeat visibly modulate WiFi channel state information, and that careful antenna placement informed by Fresnel-zone theory plus a manual selection of the best receive stream makes those modulations strong enough to read. In tests with five participants, the reported average error is 0.575 bpm for breathing and 3.9 bpm for heart rate, corresponding to overall accuracy figures of 96.636% and 94.215%. If the result holds, it would offer a low-cost, contactless way to screen sleep-related vital signs at home, replacing expensive polysomnography equipment or body-worn sensors.

What carries the argument

The load-bearing mechanism is the Fresnel-zone geometry of the antenna pair plus an empirical stream-selection rule. Fresnel zones are the concentric ellipsoids between transmitter and receiver on which a reflected path's phase shift is an integer multiple of half the wavelength; breathing and heartbeat motions change the reflected path length, and placing the body inside a sensitive zone is supposed to maximize the resulting channel response. The paper finds that the model is only a rough guide: in its own preliminary experiments the theoretically favored stream is not always the best one, and obstacles such as a plastic plate can make another stream more sensitive. The prototype therefore places the bed just outside the second Fresnel zone of the chosen T1-R3 link, puts a lead sheet under the transmitter to boost sensitivity, and manually selects R3 as the working stream, while the other two antennas are kept to prevent R3 from having the highest SNR. The processing chain then runs subcarrier selection by variance, Hampel outlier filtering, Butterworth bandpass separation of the breathing and heartbeat bands, and FFT-based rate extraction.

What would settle it

Run the same two-device setup in a different room with no plastic plate between the antennas, no lead sheet, and no manual stream search, and compare the accuracy of the automatically chosen best stream with the reported 96.636% breathing and 94.215% heart-rate accuracy; if the average errors rise well above 0.575 bpm and 3.9 bpm, the claimed general capability does not transfer as stated.

Watch

Extended reading notes

Core claim

The paper's central claim is that a single pair of WiFi transceivers with omnidirectional antennas suffices for real-time individual breathing-rate and heart-rate monitoring during sleep, in supine, prone, and left/right recumbent postures. The authors position this as the first demonstration of that capability: earlier WiFi sleep-monitoring work either needed multiple transmitters and receivers, restricted heart-rate detection to a supine posture with a directional antenna in line of sight, or used a metronome to pace the subject's breathing. The system extracts channel state information from an Intel 5300 NIC, selects the most sensitive subcarrier, filters the signal into a breathing band (0.25-0.5 Hz) and a heartbeat band (1-2 Hz), and derives rates by FFT. Ground truth came from an abdomen-worn accelerometer for breathing and a fingertip pulse oximeter for heart rate; the reported average errors are 0.575 bpm and 3.9 bpm. The authors also report that the best receiving antenna is not the one with the highest SNR, and that after each restart the best data stream must be rediscovered.

Load-bearing premise

The claim rests on the hand-tuned antenna configuration—the chosen stream, the lead sheet, and even the plastic plate that was present—transferring to other rooms and other people without losing accuracy.

Editorial extensions

If this is right

  • A contactless home monitor for sleep breathing and heart rate becomes feasible with equipment already present in many homes, at a fraction of the cost of polysomnography.
  • The reported per-posture performance suggests supine monitoring is easiest, while prone breathing and left-recumbent heart-rate measurements are the hardest; a practical system would need to handle those cases explicitly.
  • Because the system runs on commodity WiFi CSI, it could be integrated into existing routers or smart-home access points without new hardware.
  • The 40-second accumulation window for the FFT means the system can report vital signs at interactive time scales, suitable for overnight logging.

Reading between the lines

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

  • The system's dependence on manually rediscovering the best antenna after every restart suggests that automating stream selection is the key engineering step for deployment outside the test room; the paper leaves this as future work.
  • The incidental plastic-plate result indicates that deliberate reflectors could be used as a design parameter, potentially replacing the unexplained lead-sheet trick with a reproducible rule.
  • A natural testable extension is apnea detection: a prolonged absence of energy in the 0.25-0.5 Hz band, using the same pipeline with a shorter FFT window, would flag respiratory pauses without new hardware.
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

3 major / 8 minor

Summary. The paper proposes a WiFi-CSI-based system for contactless, real-time monitoring of breathing rate and heart rate during sleep, using one transmitting and one receiving commodity WiFi device with omnidirectional antennas. The authors motivate antenna placement with Fresnel-zone theory, implement a real-time Matlab processing pipeline (subcarrier selection, Hampel filtering, bandpass segmentation, FFT-based rate extraction), and report average errors of 0.575 bpm for breathing and 3.9 bpm for heart rate across five participants in four sleep postures, claiming overall accuracies of 96.636% and 94.215%. They further claim to be the first to achieve real-time individual breathing and heart-rate monitoring in different sleeping postures with a single pair of WiFi devices and omnidirectional antennas.

Significance. If the central claim held, the system would be a meaningful low-cost, contactless sleep vital-sign monitor, and the real-time implementation on commodity hardware is a practical strength. The authors use external ground-truth sensors (accelerometer and pulse oximeter), report comparisons across postures, and are candid about the limitations of Fresnel theory in their own setting. However, the current evidence does not establish transferability: the evaluation is small and homogeneous, the accuracy metric is undefined, and the system relies on manually discovered, environment-specific antenna configurations that the paper itself shows are not predicted by the theory it invokes.

major comments (3)
  1. [Section IV-C, Table I] The overall-accuracy figures (96.636% for breathing and 94.215% for heart rate) are never defined; no formula links them to the reported mean errors, and there are no per-participant or per-posture distributions, confidence intervals, or statistical tests. Please provide the exact definition of accuracy, the underlying per-measurement/per-subject data, and quantitative variance measures.
  2. [Section III-A, Section II-B] The system's operation depends on a manually identified "best data stream" that must be re-found after every restart (Section III-A states "Every time we restart the system, we need to find the best data stream"), and the final setup includes a lead sheet under T1 and an incidental plastic plate that are not part of any reproducible design rule. Because Section II-B reports that Fresnel theory failed to predict that R3 would be the best stream in Setting 1, the reported accuracies cannot be separated from the particular lab configuration. Please specify an automatic, principled criterion for stream selection and validate the configuration transfer across rooms, antenna placements, and users.
  3. [Section IV-B, Section IV-C] The evaluation is under-powered for the generality claimed: five university students aged 21–26 in a single office-like lab, with no cross-environment testing, no repeated trials, no inter-subject variability analysis, and no statistical significance testing. The statement in Section IV-C that "in general, our system can accurately monitor vital signs with different sleeping postures" goes beyond what this dataset can support. Please add a larger and more diverse participant pool, multiple environments, repeated sessions, and appropriate statistical reporting.
minor comments (8)
  1. [Section IV-B, Section IV-C] The word "sout" appears twice ("We soutconduct experiment" and "directlysout") and should be corrected.
  2. [Figure 9] The label "pron" should be "prone."
  3. [Figure 3] The x-axis label "Packages" should be "Packets" for consistency with the rest of the paper.
  4. [Table I] The Performance column for reference [12] contains "Na"; either provide the performance figure or explain the abbreviation.
  5. [Reference [9]] Reference [9] lacks venue and publication details and should be completed.
  6. [Section III-B] The "FFT time threshold" is a free parameter; please state its chosen value and report the sensitivity of the results to it.
  7. [Section II-A] The notation in Equation (1) should be made consistent (|TxRx| vs. |TxRx|) and the "effective displacement" along the normal line should be defined formally.
  8. [Figure 7] The schematic in Figure 7 does not show the lead sheet or the plastic plate that the text says are important; please include these elements in the figure or a separate setup diagram.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the reported accuracy is benchmarked against external accelerometer and pulse-oximeter ground truth, and the Fresnel model is used only as a heuristic setup guide, not as the source of the vital-sign estimates.

full rationale

The central performance claims (average error 0.575 bpm for breathing and 3.9 bpm for heart rate) are obtained by comparing CSI-derived rates with readings from an accelerometer placed on the abdomen and a fingertip pulse oximeter (Section IV-B), i.e., external ground truth, not with quantities derived from the model's own assumptions. The Fresnel-zone model in Section II-A is used only to guide antenna placement, and the paper explicitly reports that the theory is limited: 'The guidance provided by Fresnel theory is limited' (Section II-B), noting that the best stream in Setting 1 contradicted the prediction of prior work [12]. The subcarrier-selection rule cites the authors' prior work [16], but it is a practical selection heuristic, not a derivation of the final vital-sign output. Self-citations [14]-[16] are motivational or experiential and are not load-bearing: removing them does not change any equation or measured accuracy. The accuracy percentages are not produced by fitting a parameter and then renaming the fit a prediction; the processing chain uses standard filters (Hampel filter, Butterworth bandpass filters) and FFT, and the reported errors are direct differences measured against external sensors. Therefore no step reduces by construction to its own inputs, and no significant circularity is present.

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

The central claim is an empirical system result, not a theoretical derivation. It rests on standard Fresnel-zone physics and physiology, plus several hand-tuned hardware choices (antenna placement, lead sheet, best-stream selection, FFT window) that are specific to the test environment. No invented entities are introduced. The non-Fresnel empirical findings (lead sheet, blocking plate) actually undercut the paper's framing that Fresnel theory drives the performance.

free parameters (4)
  • FFT time threshold = 40 s (manually adjustable)
    The system accumulates 40 seconds of data before extracting rates; this hand-chosen window trades latency against accuracy and is not derived from theory.
  • Best antenna data stream = Selected per restart (one of R1-R3)
    The authors state that after each system restart the best-performing stream must be found manually, making the choice an environment-specific calibration.
  • Antenna placement and lead sheet position = T1-R3 distance 80 cm; lead sheet under T1
    The hardware configuration is hand-tuned in the test room to make R3 sensitive; the authors note the sensitivity depends on an unexplained blocking plate and lead sheet.
  • Subcarrier selection criterion = None (heuristic: maximum variance)
    The chosen subcarrier is selected by highest variance, a heuristic from prior work, not a fitted value; it affects which CSI stream is used.
assumptions (4)
  • domain assumption The Fresnel zone model as defined in Eq. (1), with the phase shift relation e^{-j2πd(t)/λ}, describes how a point reflector's motion affects CSI amplitude.
    Inherited from prior work [12], [13]; it assumes a point reflector and line-of-sight propagation, which the paper's preliminary experiments show is not always valid in real environments.
  • domain assumption Breathing and heartbeat produce measurable chest/abdomen displacements that modulate WiFi signals.
    Supported by preliminary experiments in Section II-B, but the magnitude and reliability vary with posture and body position.
  • domain assumption Breathing and heart rate occupy separable frequency bands (0.25-0.5 Hz and 1-2 Hz), so bandpass filtering isolates them.
    Standard physiology; used in Section III-B for the Butterworth filters.
  • ad hoc to paper A 'least sensitive' CSI stream with SNR around 20 always exists when three antennas are used, and it is the most stable for vital sign extraction.
    This is an empirical observation from the authors' hardware (Section III-A), not derived from theory or prior work; it justifies keeping the other two antennas.

how reviews work

0 comments
Cite this review

Pith. "Pith review of WiFi-based Real-time Breathing and Heart Rate Monitoring during Sleep." pith.science (2026). https://pith.science/paper/JOAFEHW7

@misc{pith2026190805108,
  author       = {Pith},
  title        = {Pith review of: WiFi-based Real-time Breathing and Heart Rate Monitoring during Sleep},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JOAFEHW7}},
  note         = {Machine review of arXiv:1908.05108}
}
read the original abstract

Good quality sleep is essential for good health and sleep monitoring becomes a vital research topic. This paper provides a low cost, continuous and contactless WiFi-based vital signs (breathing and heart rate) monitoring method. In particular, we set up the antennas based on Fresnel diffraction model and signal propagation theory, which enhances the detection of weak breathing/heartbeat motion. We implement a prototype system using the off-shelf devices and a real-time processing system to monitor vital signs in real time. The experimental results indicate the accurate breathing rate and heart rate detection performance. To the best of our knowledge, this is the first work to use a pair of WiFi devices and omnidirectional antennas to achieve real-time individual breathing rate and heart rate monitoring in different sleeping postures.

Figures

Figures reproduced from arXiv: 1908.05108 by the authors.

Figure 1
Figure 1. Fresnel Zone of dynamic paths. A subject can reflect a WiFi signal, and if the subject moves a small distance, it leads to changes in the phase of the WiFi signal on the corresponding path. If the subject moves d(t), since wireless signals travel at the speed of light, denoted as c, τk(t) can be represented as d(t)/c. Let λ represent the wavelength, where λ = f /c. Thus, the phase shift can be written as e −j2πd(t)/… view at source ↗
Figure 2
Figure 2. (a) Setting1; T1 is transmit antenna and R1,R2,R3 are [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. (a)Detection results of setting1; (b)Detection results of setting2; (c)Detection results of setting3. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: System architecture we must place the other two antennas. This is due to that the CSI tool must receive the data stream of three receiving antennas at the same time to collect the CSI data. Moreover, we find that no matter how the signal is blocked, there is always a d…
Figure 5
Figure 5. Figure 5: Comparison of processed CSI and ACC sensor read [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Comparison of processed CSI and ACC sensor read [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 8
Figure 8. Figure 8: Illustration of the vital sign (breathing and heart rate) [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 9
Figure 9. Figure 9: Performances of different sleep postures. [PITH_FULL_IMAGE:figures/full_fig_p006_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

  1. [1]

    Who technical meeting on sleep and health: Bonn, germany 22-24 january 2004,

    W. H. Organization et al., “Who technical meeting on sleep and health: Bonn, germany 22-24 january 2004,” in WHO technical meeting on sleep and health: Bonn, Germany 22-24 January 2004, 2004

    work page 2004

  2. [2]

    Breathing adapted radiotherapy of breast cancer: reduction of cardiac and pul- monary doses using voluntary inspiration breath-hold,

    A. N. Pedersen, S. Korreman, H. Nystr ¨om, and L. Specht, “Breathing adapted radiotherapy of breast cancer: reduction of cardiac and pul- monary doses using voluntary inspiration breath-hold,” Radiotherapy and oncology, vol. 72, no. 1, pp. 53–60, 2004

    work page 2004

  3. [3]

    Noncontact respiration rate measurement system using an ultrasonic proximity sensor,

    S. D. Min, J. K. Kim, H. S. Shin, Y . H. Yun, C. K. Lee, and M. Lee, “Noncontact respiration rate measurement system using an ultrasonic proximity sensor,” IEEE Sensors Journal, vol. 10, no. 11, pp. 1732– 1739, 2010

    work page 2010

  4. [4]

    Bridging the collaborative gap: Realizing the clinical potential of breath analysis for disease diagnosis and monitoring–tutorial,

    P. X. Braun, C. F. Gmachl, and R. A. Dweik, “Bridging the collaborative gap: Realizing the clinical potential of breath analysis for disease diagnosis and monitoring–tutorial,” IEEE Sensors Journal, vol. 12, no. 11, pp. 3258–3270, 2012

    work page 2012

  5. [5]

    Sleep disordered breathing in a high-risk cohort prevalence and severity across pregnancy,

    F. L. Facco, D. W. Ouyang, P. C. Zee, and W. A. Grobman, “Sleep disordered breathing in a high-risk cohort prevalence and severity across pregnancy,”American journal of perinatology, vol. 31, no. 10, pp. 899– 904, 2014

    work page 2014

  6. [6]

    Practice parameters for the indications for polysomnography and related procedures: an update for 2005,

    C. A. Kushida, M. R. Littner, T. Morgenthaler, C. A. Alessi, D. Bailey, J. Coleman Jr, L. Friedman, M. Hirshkowitz, S. Kapen, M. Kramer et al., “Practice parameters for the indications for polysomnography and related procedures: an update for 2005,” Sleep, vol. 28, no. 4, pp. 499–523, 2005

    work page 2005

  7. [7]

    Litenet: Lightweight neural network for detecting arrhythmias at resource- constrained mobile devices,

    Z. He, X. Zhang, Y . Cao, Z. Liu, B. Zhang, and X. Wang, “Litenet: Lightweight neural network for detecting arrhythmias at resource- constrained mobile devices,” Sensors, vol. 18, no. 4, p. 1229, 2018

    work page 2018

  8. [8]

    Propagation parameter estimation, modeling and measurements for ultrawideband mimo radar,

    J. Salmi and A. F. Molisch, “Propagation parameter estimation, modeling and measurements for ultrawideband mimo radar,” IEEE Transactions on Antennas and Propagation, vol. 59, no. 11, pp. 4257– 4267, 2011

    work page 2011

Show all 16 references
  1. [9]

    Multi-person motion tracking via rf body reflections,

    F. Adib, Z. Kabelac, and D. Katabi, “Multi-person motion tracking via rf body reflections,” 2014

    2014

  2. [10]

    Contactless respiration monitoring via off-the-shelf wifi devices,

    X. Liu, J. Cao, S. Tang, J. Wen, and P. Guo, “Contactless respiration monitoring via off-the-shelf wifi devices,” IEEE Transactions on Mobile Computing, vol. 15, no. 10, pp. 2466–2479, 2016

    2016

  3. [11]

    Monitoring vital signs and postures during sleep using wifi signals,

    J. Liu, Y . Chen, Y . Wang, X. Chen, J. Cheng, and J. Yang, “Monitoring vital signs and postures during sleep using wifi signals,” IEEE Internet of Things Journal, vol. 5, no. 3, pp. 2071–2084, 2018

    2018

  4. [12]

    Human respiration detection with commodity wifi devices: do user location and body orientation matter?

    H. Wang, D. Zhang, J. Ma, Y . Wang, Y . Wang, D. Wu, T. Gu, and B. Xie, “Human respiration detection with commodity wifi devices: do user location and body orientation matter?” in Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ...

    2016

  5. [13]

    From fresnel diffraction model to fine-grained human respiration sensing with commodity wi-fi devices,

    F. Zhang, D. Zhang, J. Xiong, H. Wang, K. Niu, B. Jin, and Y . Wang, “From fresnel diffraction model to fine-grained human respiration sensing with commodity wi-fi devices,” Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 2, no. 1, p. 53, 2018

    2018

  6. [14]

    Emosense: Data-driven emotion sensing via off-the-shelf wifi devices,

    Y . Gu, T. Liu, J. Li, F. Ren, Z. Liu, X. Wang, and P. Li, “Emosense: Data-driven emotion sensing via off-the-shelf wifi devices,” in Proc. of IEEE ICC, May 2018

    2018

  7. [15]

    Mosense: An rf-based motion detection system via off-the-shelf wifi devices,

    Y . Gu, J. Zhan, Y . Ji, J. Li, F. Ren, and S. Gao, “Mosense: An rf-based motion detection system via off-the-shelf wifi devices,” IEEE Internet of Things Journal, vol. 4, no. 6, pp. 2326–2341, Dec 2017

    2017

  8. [16]

    Your wifi knows how you behave: Leveraging wifi channel data for behavior analysis,

    Y . Gu, X. Zhang, C. Li, F. Ren, J. Li, and Z. Liu, “Your wifi knows how you behave: Leveraging wifi channel data for behavior analysis,” in 2018 IEEE Global Communications Conference (GLOBECOM). IEEE, 2018, pp. 1–6

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.