Pith. sign in

REVIEW 4 major objections 6 minor 46 references

Enhanced Gas Source Localization Using Distributed IoT Sensors and Bayesian Inference

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Nine IoT sensors locate a gas source within 20% of their spacing

desk verdict A solid engineering integration of SMC with IoT gas sensors, but the real-data validation only tests sources on the sensor array's symmetry axis, so the headline claim is broader than the evidence. read the letter →

arxiv 2411.13268 v1 pith:EUFBEJ4H submitted 2024-11-20 physics.flu-dyn

Leonardo Balocchi , Lorenzo Piro , Luca Biferale , Stefania Bonafoni , Massimo Cencini , Iacopo Nannipieri , Andrea Ria , Luca Roselli This is my paper · ORCID
classification physics.flu-dyn
keywords gassourcelocalizationBayesianinferencesequentialMonteCarloGaussianplumemodelIoTsensornetworkwatervaporsensingPoissondetectionindoorairquality
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 tries to establish that a network of nine cheap, low-power IoT gas sensors, running a Bayesian inference algorithm on their discrete 'hit' counts, can localize a gas source in an indoor room to better than the spacing between sensors. The authors test this with water vapor released from a kettle on the floor, using a nine-sensor array hanging 110 cm below the ceiling, and report that the average localization error falls to 10-20% of the 60 cm sensor spacing within about 15 minutes. This matters because gas leak detection in homes and workplaces could then be done with inexpensive distributed hardware instead of costly analytical instruments, and the same probabilistic machinery might extend to turbulent environments where gas signals are intermittent. The paper's central burden is to show that a deliberately simplified Gaussian plume model, one that ignores convection and buoyancy, still contains enough information for the Bayesian update to find the source.

What carries the argument

The central object is the Gaussian plume model of Eq. (2), $c(\mathbf{r}-\mathbf{r}_s) = \frac{Q}{\pi u \lambda^2} e^{-\|\mathbf{r}-\mathbf{r}_s\|^2/\lambda^2}$, paired with the Poisson detection model of Eqs. (3)-(4), where each sensor records hits $h_i$ with mean hit rate $\mu(d_i) = \frac{\tilde{Q} a \Delta t}{\lambda^2} e^{-d_i^2/\lambda^2}$. The algorithm is a Sequential Monte Carlo particle filter with a Metropolis-Hastings perturbation step that carries a belief distribution over the source position $\mathbf{r}_s$ and the two unknown parameters $\tilde{Q}$ and $\lambda$, updating it with Bayes' rule at every measurement round.

What would settle it

Run the identical nine-sensor array and SMC algorithm in a room with a steady cross-draft, for example a household fan producing a 0.5 m/s wind across the plume, and measure the average localization error against the known source position; if the error systematically exceeds the 60 cm sensor spacing or is biased downwind, the Gaussian plume assumption fails. A second check would be to compare the measured time-averaged concentration profile along the sensor line with the exponential $e^{-d^2/\lambda^2}$ shape of Eq. (2); a mismatch predicts exactly where the method will break.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that a sequential Monte Carlo algorithm, which jointly infers the source position and two environmental parameters (effective emission rate $\tilde{Q}$ and diffusion length scale $\lambda$) from the Poisson hit counts of nine static sensors, localizes a water vapor source accurately in both simulations and a real room. In numerical experiments with known ground truth, the localization error averaged over one thousand runs settles at about 10% of the sensor spacing. In the real experiment, repeated ten times at each of two source positions, the average error is 10-20% of the 60 cm spacing by the end of the run, with convergence typically reached before the full 20 minutes. The authors conclude that, despite the oversimplified environment model, the combination of the Gaussian plume likelihood and Monte Carlo sampling is sufficient for practical indoor source localization in the tested configurations.

Load-bearing premise

The load-bearing premise is that the time-averaged concentration field at the sensor plane is a symmetric Gaussian plume described by a single diffusion length scale, with Poisson-distributed sensor hits; if real room convection or plume buoyancy distorts this shape, the inferred source position could be biased despite the success in the two tested configurations.

Editorial extensions

If this is right

  • Source localization becomes feasible with hardware costing under 10 USD per sensing chip and drawing less than 1.4 mW, since the algorithm needs only discrete hit counts from a static sensor array.
  • The method converges online, typically within 1000 seconds, fast enough for real-time leak response and continuous indoor air quality monitoring.
  • Because the likelihood is a Poisson model of sparse detections, the same algorithm should carry over to turbulent environments where gas signals are intermittent, not just the near-stationary plume tested here.
  • The algorithm simultaneously estimates the environmental parameters $\tilde{Q}$ and $\lambda$, so it does not need prior knowledge of emission rate or room diffusivity, removing a common practical obstacle.
  • The Sensibus single-wire protocol lets nine sensors share one cable and take synchronized readings, making the hardware simple to deploy on ceilings, walls, or mobile platforms.

Reading between the lines

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

  • A likely failure mode is a ventilated room: because the model has no mean wind, a steady cross-draft would break the circular symmetry of the Gaussian plume, biasing the estimate downwind; this can be tested directly with a desk fan.
  • The two tested source positions both lie roughly along the horizontal symmetry axis of the sensor cross; the paper does not test sources near the arena boundary, where fewer sensors would see the plume and the posterior may become multimodal.
  • The Poisson hit discretization suggests a direct recipe for other sparse chemical sensors: calibrate the per-sensor background, threshold to hits, and reuse the same likelihood, which would let an e-nose array adopt the method without new algorithm development.
  • The claim 'consistently lower than the sensor distance' rests on 20 real trials total (two positions, ten repetitions); whether it holds for arbitrary source positions, room sizes, and source temperatures remains open.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes a gas source localization method that combines nine low-cost IoT sensors with a sequential Monte Carlo (SMC) Bayesian inference scheme. The environment is modeled by a stationary Gaussian plume (Eq. 2) and the sensor readings are converted into Poissonian hit counts (Eqs. 3-4). The method is tested first on synthetic data generated from the same model and then on real water-vapor experiments with two source locations. The authors report localization errors of 10-20% of the sensor spacing in both settings and claim that the error is consistently lower than the distance between adjacent sensors.

Significance. If the general two-dimensional localization claim is supported, the paper would provide a useful demonstration that a simple Bayesian Monte Carlo approach, combined with inexpensive distributed IoT hardware, can localize an indoor gas source with sub-sensor-spacing accuracy. The strengths of the manuscript are the concrete end-to-end system description, the repeated real experiments (10 repetitions per location), and the algorithmic detail in the appendix that makes the SMC implementation reproducible. However, the current experimental validation is too narrow to establish the headline claim: the two real source locations lie on the symmetry axis of the sensor array, so the reported accuracy is essentially a one-dimensional along-row result.

major comments (4)
  1. [Sec. IV.B, Fig. 8] The real-world validation tests only two source positions, (98,128) cm and (38,128) cm, both lying on the horizontal symmetry line y=128 of the nine-sensor cross. Because the likelihood (Eqs. 3-5) depends only on distances to the sensors, and the prior is uniform, the posterior is exactly symmetric about y=128 for a true source on that line. Consequently, the SMC center-of-mass y-coordinate is unbiased by construction, and the reported 10-20% of ds error is essentially an along-row interpolation error. To support the claim of general 2D source localization, off-axis source positions must be tested, or at minimum the separate x and y errors should be reported; as it stands, the experiment does not challenge the model in the perpendicular direction.
  2. [Sec. IV.A] The synthetic experiment generates sensor data from the same Gaussian plume and Poisson hit model that is later used for inference (Eqs. 2-4). This is a valid self-consistency check of the SMC implementation, but it cannot validate the environmental model. The only model-adequacy evidence is the real-data experiment, which is limited to two collinear source locations and only 10 repetitions each. The paper should either add more real source configurations, including off-axis ones, or explicitly characterize how the reported accuracy degrades as the model mismatch increases.
  3. [Sec. IV.B] No baseline or comparison method is reported for the real-data experiment. The claim that the Bayesian SMC algorithm localizes the source to 10-20% of the sensor spacing would be much stronger if compared to simple alternatives such as the nearest-sensor estimate, the weighted centroid of the hit counts, or a direct least-squares fit of the plume model. Without such a baseline, it is unclear how much of the reported accuracy is due to the Bayesian machinery and how much is already contained in the sensor geometry and the hit-rate pattern.
  4. [Sec. IV.B and Fig. 7] The preprocessing pipeline relies on hand-tuned choices: the moving-average window of 20 s and the hit threshold of 500 Ω. These choices directly determine the discrete hit counts h that enter the likelihood, so the reported localization performance may be sensitive to them. The paper should include a sensitivity analysis over these parameters, or at least justify the chosen values with a principled criterion, before the 10-20% accuracy can be considered a robust property of the method.
minor comments (6)
  1. [Eq. (2)] The exponent in Eq. (2) appears to be missing the square on the distance: it should be exp(-||r - r_s||^2 / lambda^2), consistent with Eq. (4). The current notation exp(-||r - r_s|| / lambda^2) is dimensionally inconsistent with the rest of the paper.
  2. [Acknowledgments] The acknowledgments paragraph is duplicated verbatim before the Abbreviations section; one copy should be removed.
  3. [Abbreviations] The abbreviation 'IAQ Indoor Air Quality' is listed twice in the Abbreviations section.
  4. [Fig. 8] The shaded region in the bottom panels is described only as a confidence interval; please state explicitly whether it is one standard deviation, a 95% interval, or another quantile, and over how many experiments it is computed.
  5. [Sec. IV.B] The statement that the algorithm 'could converge to the correct source location typically within such a shorter time frame' would benefit from a quantitative criterion defining convergence, rather than a visual inspection of the figures.
  6. [Sec. IV.A and Table A1] The priors for lambda and Q~ are described only as 'uniform' in Sec. IV.A, while Table A1 lists ranges; please state explicitly whether the uniform priors are in the parameters themselves or in some transformed variables, since this affects the SMC implementation.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation chain: the source position is jointly inferred from sensor data, model parameters are not set by the target, and the synthetic test is an acknowledged self-consistency check rather than a prediction.

full rationale

The paper's derivation chain is not circular. The target quantity, source position r_s, is estimated from sensor detections through a likelihood built on the Gaussian plume model (Eq. 2) and Poisson hit model (Eqs. 3-4). The unknown environmental parameters (Qtilde, lambda) are inferred online by the SMC algorithm from the same sensor measurements, but they are not derived from, nor fitted to, the source position: r_s is the inferential target, not an input to the parameter estimates. The numerical experiment generates sensor data from the exact model used for inference, which the paper explicitly frames as a preliminary self-consistency test ('synthetic data generated from the same model of the environment used for the inference'), not as independent empirical validation. The real experiment uses measured sensor data and reports localization errors of 10-20% of the sensor spacing, which is an empirical result, not a consequence of the model assumptions. Self-citations (e.g., Refs. [22], [31], [33-38]) support hardware and protocol descriptions and are not load-bearing in the localization derivation. No uniqueness theorem, ansatz, or known result is invoked via self-citation to force the conclusion. The geometry concern that both real source locations lie on the symmetry axis affects external validity, but it is not a circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a simple Gaussian plume model with two inferred parameters (λ, \tilde{Q}) and hand-chosen preprocessing constants. The calibration coefficients and threshold/window are fitted or selected by hand, but they are not fitted to the source position, so the circularity burden is low. No new physical entities are introduced.

free parameters (5)
  • Diffusion length scale λ = Estimated online by SMC; ground truth 80 cm in synthetic experiments
    Model parameter for the Gaussian plume (Eq. 2, 4). In synthetic tests it is set to 80 cm and recovered by the filter; in real experiments it is a free parameter inferred from the sensor data.
  • Effective source emission rate \tilde{Q} = Estimated online by SMC; ground truth 2·10^3 mol·cm/s in synthetic experiments
    Lumped rate in the Poisson hit model (Eq. 4). Absorbs source strength, mean wind speed and detection radius; inferred from the data.
  • Hit threshold = 500 Ω
    Impedance change above which a sensor reading is counted as a hit (Sec. IV.B, Fig. 7d). Chosen by hand; sensitivity not analyzed.
  • Moving-average window = 20 s
    Smoothing window applied to calibrated impedance before thresholding (Sec. IV.B, Fig. 7c). Chosen by hand.
  • Per-sensor calibration coefficients α_i, β_i = Nine pairs, see Table I (at 26°C)
    Linear calibration to equalize sensor responses (Eq. 1). Fitted to calibration data with 5 repetitions; values in Table I.
assumptions (5)
  • domain assumption Gaussian plume model (Eq. 2) describes the stationary mean concentration field of the buoyant water vapor plume in the sensor plane.
    Invoked in Sec. III.A; neglects convective motions, room boundaries and intermittency.
  • domain assumption Poisson detection statistics (Eqs. 3-4) with Smoluchowski relation relate mean concentration to hit counts.
    Used to build likelihood in Eq. 5; approximate for real sensors.
  • domain assumption The source height z_s (110 cm below the sensor plane) is known, reducing the search to 2D.
    Stated in Sec. III.A and Figure 1; if wrong, the model is misspecified.
  • domain assumption Calibrated and thresholded sensor readings are conditionally independent given the source position.
    Used in Eq. (5) for the joint likelihood; real sensors share the same environment and may be correlated.
  • standard math State-of-the-art SMC with MCMC perturbation correctly approximates the Bayesian posterior for this 4D problem.
    Relies on standard particle filter convergence results, cited to refs 26 and 27.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enhanced Gas Source Localization Using Distributed IoT Sensors and Bayesian Inference." pith.science (2026). https://pith.science/paper/EUFBEJ4H

@misc{pith2026241113268,
  author       = {Pith},
  title        = {Pith review of: Enhanced Gas Source Localization Using Distributed IoT Sensors and Bayesian Inference},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EUFBEJ4H}},
  note         = {Machine review of arXiv:2411.13268}
}
read the original abstract

Identifying a gas source in turbulent environments presents a significant challenge for critical applications such as environmental monitoring and emergency response. This issue is addressed through an approach that combines distributed IoT smart sensors with an algorithm based on Bayesian inference and Monte Carlo sampling techniques. Employing a probabilistic model of the environment, such an algorithm interprets the gas readings obtained from an array of static sensors to estimate the location of the source. The performance of our methodology is evaluated by its ability to estimate the source's location within a given time frame. To test the robustness and practical applications of the methods under real-world conditions, we deployed an advanced distributed sensors network to gather water vapor data from a controlled source. The proposed methodology performs well when using both the synthetic data generated by the model of the environment and those measured in the real experiment, with the source localization error consistently lower than the distance between one sensor and the next in the array.

Figures

Figures reproduced from arXiv: 2411.13268 by the authors.

Figure 1
Figure 1. FIG. 1: Gas measurement setup with an array of sensors [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2: Calibration setup: the 9 sensor platforms (in [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3: Calibration phase: measurement of aluminum [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: FIG. 4: Setup of the gas measurement experiment, with [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5: Measurement of aluminum oxide impedance [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6: Results obtained in the numerical experiment. Starting from the left, the figures represent two spatial plots [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7: Data corresponding to the experiment with the source placed between the middle and left sensors, i.e. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8: Results obtained in the real experiment. [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 42 canonical work pages

  1. [1]

    Gonz´ alez-Mart ´ ın, N

    J. Gonz´ alez-Mart ´ ın, N. J. R. Kraakman, C. P´ erez, R. Le- brero, and R. Mu˜ noz, Chemosphere262, 128376 (2021)

    work page 2021

  2. [2]

    Y. Yan, Y. Hu, L. Wang, X. Qian, W. Zhang, K. Reda, J. Wu, and G. Zheng, Measurement 169, 108506 (2021)

    work page 2021

  3. [3]

    Siregar, A

    B. Siregar, A. N. Nasution, and D. Arisandi, in 2020 International Conference on Data Science, Artificial In- telligence, and Business Analytics (DATABIA) (IEEE,

    work page 2020

  4. [4]

    Y. Chen, D. Chen, T. Song, and K. Song, in 2020 IEEE 4th International Conference on Frontiers of Sensors Technologies (ICFST) (IEEE, 2020) pp. 21–25

    work page 2020

  5. [5]

    Dimitroulopoulou, Building and Environment 47, 109 (2012)

    C. Dimitroulopoulou, Building and Environment 47, 109 (2012)

    work page 2012

  6. [6]

    Marchetti, A

    N. Marchetti, A. Cavazzini, L. Pasti, M. Catani, C. Malag` u, and V. Guidi, in2015 XVIII AISEM Annual Conference (IEEE, 2015) pp. 1–4

    work page 2015

  7. [7]

    Kumar, I

    A. Kumar, I. Singh, and S. Sud, in 2009 International Multimedia, Signal Processing and Communication Tech- nologies (IEEE, 2009) pp. 245–249

    work page 2009

  8. [8]

    Hern´ andez-Gordillo, S

    A. Hern´ andez-Gordillo, S. Ruiz-Correa, V. Robledo- Valero, C. Hern´ andez-Rosales, and S. Arriaga, Air Qual- ity, Atmosphere & Health , 1 (2021). 11

    work page 2021

Show all 46 references
  1. [9]

    Jinfeng, C

    L. Jinfeng, C. Chen, C. Xiaowei, and W. Wei, in 2021 33rd Chinese Control and Decision Conference (CCDC) (IEEE, 2021) pp. 2776–2780

    2021

  2. [10]

    Bayat, N

    B. Bayat, N. Crasta, A. Crespi, A. M. Pascoal, and A. Ijspeert, Current opinion in biotechnology 45, 76 (2017)

    2017

  3. [11]

    Burgu´ es and S

    J. Burgu´ es and S. Marco, Science of the total environ- ment 748, 141172 (2020)

    2020

  4. [12]

    R. D. Garrick, L. Villasmil, J. Lee, and J. S. Gutterman, in 2011 Future of Instrumentation International Work- shop (FIIW) Proceedings (IEEE, 2011) pp. 164–167

    2011

  5. [13]

    Ernst, A

    H. Ernst, A. Jachimowicz, and G. Urban, IEEE sensors journal 1, 361 (2001)

    2001

  6. [14]

    Dawei, L

    M. Dawei, L. Yufeng, and X. Yongming, in 2011 Fourth International Conference on Intelligent Compu- tation Technology and Automation , Vol. 1 (IEEE, 2011) pp. 59–62

    2011

  7. [15]

    Mishra, M

    D. Mishra, M. Hatto, T. Kuhara, N. Fujita, Y. Osana, and H. Amano, in 2014 Second International Symposium on Computing and Networking (IEEE, 2014) pp. 608– 610

    2014

  8. [16]

    Ferdinandi, M

    M. Ferdinandi, M. Molinara, G. Cerro, L. Ferrigno, C. Marroco, A. Bria, P. Di Meo, C. Bourelly, and R. Sim- marano, in 2019 IEEE international conference on smart computing (SMARTCOMP) (IEEE, 2019) pp. 186–191

    2019

  9. [17]

    Tariq, Z

    S. Tariq, Z. Hu, and T. Zayed, J. Clean. Prod. 289, 125751 (2021)

    2021

  10. [18]

    Esposito, L

    M. Esposito, L. Palma, A. Belli, L. Sabbatini, and P. Pierleoni, Sensors 22, 10.3390/s22062124 (2022)

    2022 doi

  11. [19]

    Celani, E

    A. Celani, E. Villermaux, and M. Vergassola, Phys. Rev. X 4, 041015 (2014)

    2014

  12. [20]

    Piro, Optimal navigation in active matter , Ph.D

    L. Piro, Optimal navigation in active matter , Ph.D. thesis, Georg-August-Universit¨ at G¨ ottingen G¨ ottingen (2023)

    2023

  13. [21]

    Reddy, V

    G. Reddy, V. N. Murthy, and M. Vergassola, Annu. Rev. Condens. Matter Phys. 13, 191 (2022)

    2022

  14. [22]

    Cerro, M

    G. Cerro, M. Ferdinandi, L. Ferrigno, and M. Molinara, in 2017 IEEE International Workshop on Measurement and Networking (M&N) (IEEE, 2017) pp. 1–5

    2017

  15. [23]

    Cerro, M

    G. Cerro, M. Ferdinandi, L. Ferrigno, M. Laracca, and M. Molinara, IEEE Transactions on Instrumentation and Measurement 67, 2504 (2018)

    2018

  16. [24]

    Bruschi, G

    P. Bruschi, G. Cerro, L. Colace, A. De Iacovo, S. Del Cesta, M. Ferdinandi, L. Ferrigno, M. Molinara, A. Ria, R. Simmarano, et al., in 2018 IEEE international conference on smart computing (SMARTCOMP) (IEEE,

    2018

  17. [25]

    Zampolli, I

    S. Zampolli, I. Elmi, P. Bruschi, A. Ria, F. Magliocca, M. Vitelli, and M. Piotto, Sensors and Actuators B: Chemical , 136924 (2024)

    2024

  18. [26]

    Johannesson, B

    G. Johannesson, B. Hanley, and J. Nitao, Dynamic Bayesian Models via Monte Carlo - An Introduction with Examples - , Tech. Rep. (U.S. Department of Energy Of- fice of Scientific and Technical Information, 2004)

    2004

  19. [27]

    Elfring, E

    J. Elfring, E. Torta, and R. van de Molengraft, Sensors 21, 10.3390/s21020438 (2021)

    2021 doi

  20. [28]

    L. Piro, R. A. Heinonen, M. Cencini, and L. Biferale, Many wrong models approach to localize an odor source in turbulence: introducing the weighted bayesian update (2024), arXiv:2407.08343 [physics.flu-dyn]

    2024 arXiv

  21. [29]

    O. G. Sutton and G. C. Simpson, Proc. R. Soc. Lond. A 135, 143 (1932)

    1932

  22. [30]

    Hutchinson, H

    M. Hutchinson, H. Oh, and W.-H. Chen, Information Fusion 36, 130 (2017)

    2017

  23. [31]

    Balocchi, M

    L. Balocchi, M. Vitelli, M. Molinara, S. Bonafoni, L. Roselli, S. Contardi, and I. Nannipieri, in 2024 9th In- ternational Conference on Smart and Sustainable Tech- nologies (SpliTech) (IEEE, 2024) pp. 1–5

    2024

  24. [32]

    A. Ria, M. Cicalini, G. Manfredini, A. Catania, M. Pi- otto, and P. Bruschi, in International Conference on Ap- plications in Electronics Pervading Industry, Environ- ment and Society (Springer, 2021) pp. 256–261

    2021

  25. [33]

    A. Ria, A. Motroni, F. Gagliardi, M. Piotto, and P. Br- uschi, in 2023 8th International Conference on Smart and Sustainable Technologies (SpliTech) (IEEE, 2023) pp. 1– 4

    2023

  26. [34]

    Gerevini, G

    L. Gerevini, G. Cerro, A. Bria, C. Marrocco, L. Ferrigno, M. Vitelli, A. Ria, and M. Molinara, Journal of King Saud University-Computer and Information Sciences 35, 499 (2023)

    2023

  27. [35]

    Bourelly, A

    C. Bourelly, A. Bria, L. Ferrigno, L. Gerevini, C. Mar- rocco, M. Molinara, G. Cerro, M. Cicalini, and A. Ria, in 2020 IEEE International Conference on Smart Com- puting (SMARTCOMP) (IEEE, 2020) pp. 410–415

    2020

  28. [36]

    A. Ria, A. Motroni, F. Gagliardi, M. Piotto, and P. Br- uschi, in International Conference on Applications in Electronics Pervading Industry, Environment and Soci- ety (Springer, 2023) pp. 258–263

    2023

  29. [37]

    Molinara, M

    M. Molinara, M. Ferdinandi, G. Cerro, L. Ferrigno, and E. Massera, IEEE Access 8, 72204 (2020)

    2020

  30. [38]

    Bourelly, M

    C. Bourelly, M. Ferdinandi, M. Molinara, L. Ferrigno, and R. Simmarano, in Sensors and Microsystems: Pro- ceedings of the 20th AISEM 2019 National Conference 20 (Springer, 2020) pp. 381–386

    2019

  31. [39]

    Ristic, A

    B. Ristic, A. Gunatilaka, and R. Gailis, Atmos. Environ. 142, 114 (2016)

    2016

  32. [40]

    M. v. Smoluchowski, Z. Phys. Chem. 92, 129 (1918)

    1918

  33. [41]

    Vergassola, E

    M. Vergassola, E. Villermaux, and B. I. Shraiman, Na- ture 445, 406 (2007)

    2007

  34. [42]

    S. M. Stigler, Journal of the Royal Statistical Society. Series A (General) 145, 250 (1982)

    1982

  35. [43]

    G. E. Box and G. C. Tiao, Bayesian inference in statis- tical analysis (John Wiley & Sons, 2011)

    2011

  36. [44]

    J. Wang, R. Zhang, Y. Yan, X. Dong, and J. M. Li, Atmospheric environment 157, 27 (2017)

    2017

  37. [45]

    Asenov, M

    M. Asenov, M. Rutkauskas, D. Reid, K. Subr, and S. Ra- mamoorthy, IEEE Robotics and Automation Letters 4, 1776 (2019)

    2019

  38. [128]

    PON 2022 Ricerca e Innovazione

    and for line (b) (38, 128). The 9 purple circles spaced evenly, represent the 9 sensors (Sensiplus) used for the measurements. By contrast, the triangles of different shades of green, going from the lightest to the darkest, represent the sequential Monte Carlo (SMC) samples at...

    2000

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.