pith. sign in

arxiv: 1906.09769 · v1 · pith:HPABVAOQnew · submitted 2019-06-24 · 💻 cs.LG · cs.AI· stat.ML

Fault Matters: Sensor Data Fusion for Detection of Faults using Dempster-Shafer Theory of Evidence in IoT-Based Applications

Nimisha Ghosh , Rourab Paul , Satyabrata Maity , Krishanu Maity , Sayantan Saha This is my paper

Pith reviewed 2026-05-25 17:39 UTC · model grok-4.3

classification 💻 cs.LG cs.AIstat.ML
keywords fault detectionDempster-Shafer Theorysensor data fusionInternet of Thingsevidence theorymachine learning comparison
0
0 comments X

The pith

Dempster-Shafer Theory fuses uncertain sensor readings to identify faults in IoT nodes at 99.8 percent benchmark and 99.9 percent lab accuracy.

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

The paper applies Dempster-Shafer Theory of Evidence to combine information from multiple sensors and decide whether an IoT sensor node is faulty. This matters because IoT systems generate large volumes of data that drive critical decisions, so inaccurate readings can lead to errors. The authors test the approach on both a benchmark dataset and data from a laboratory testbed, reporting high accuracy across fault types and better results than other machine learning methods. The work focuses on precise detection to limit false alarms.

Core claim

Dempster-Shafer Theory of Evidence collates information from sensors to reach a decision on the faulty status of a sensor node; simulations on benchmark data achieve 99.8 percent accuracy and laboratory data achieve 99.9 percent accuracy for different fault types, outperforming state-of-the-art machine learning techniques.

What carries the argument

Dempster-Shafer Theory of Evidence, used to combine uncertain sensor measurements into a single fault decision.

If this is right

  • Accurate fusion supports reliable data for critical IoT decisions that depend on sensor readings.
  • The approach handles multiple fault types while keeping false alerts low.
  • Performance holds on both standardized benchmark sets and real laboratory collections.
  • The method exceeds the accuracy of several standard machine learning classifiers on the tested cases.

Where Pith is reading between the lines

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

  • The same evidence-combination step could be tested on sensor streams from non-IoT domains such as industrial monitoring.
  • Real-time implementation would allow immediate node replacement once a fault decision is reached.
  • Scaling the number of sensors per node might require adjustments to the combination rules to maintain the reported accuracy levels.

Load-bearing premise

Dempster-Shafer Theory can be applied directly to sensor data to produce accurate fault decisions without domain-specific adjustments that would change the reported performance.

What would settle it

Running the same fusion procedure on a new IoT sensor dataset and obtaining accuracy below 90 percent for common fault types would indicate the method does not generalize as claimed.

Figures

Figures reproduced from arXiv: 1906.09769 by Krishanu Maity, Nimisha Ghosh, Rourab Paul, Satyabrata Maity, Sayantan Saha.

Figure 1
Figure 1. Figure 1: Probability plot for the labelled data set [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Probability plot for the laboratory data set [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: ROC for different types of faults (Fault rate =10%, [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Accuracy of DSTE for different types of faults [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
read the original abstract

Fault detection in sensor nodes is a pertinent issue that has been an important area of research for a very long time. But it is not explored much as yet in the context of Internet of Things. Internet of Things work with a massive amount of data so the responsibility for guaranteeing the accuracy of the data also lies with it. Moreover, a lot of important and critical decisions are made based on these data, so ensuring its correctness and accuracy is also very important. Also, the detection needs to be as precise as possible to avoid negative alerts. For this purpose, this work has adopted Dempster-Shafer Theory of Evidence which is a popular learning method to collate the information from sensors to come up with a decision regarding the faulty status of a sensor node. To verify the validity of the proposed method, simulations have been performed on a benchmark data set and data collected through a test bed in a laboratory set-up. For the different types of faults, the proposed method shows very competent accuracy for both the benchmark (99.8%) and laboratory data sets (99.9%) when compared to the other state-of-the-art machine learning techniques.

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

0 major / 1 minor

Summary. The paper proposes using Dempster-Shafer Theory of Evidence to fuse uncertain measurements from multiple sensors for detecting faults in IoT sensor nodes. It evaluates the approach via simulations on a benchmark dataset (99.8% accuracy) and laboratory testbed data (99.9% accuracy) across different fault types, reporting superior performance relative to listed state-of-the-art machine learning baselines.

Significance. If the reported accuracies hold under the stated protocol, the work provides an empirical demonstration that DST-based fusion can yield high-accuracy fault decisions from noisy sensor streams in IoT settings. The dual evaluation on benchmark and real laboratory data, together with explicit (if simple) definitions for basic probability assignments, combination rule, and decision threshold, strengthens the practical claim. The comparison against ML techniques supplies a useful performance baseline.

minor comments (1)
  1. [Abstract] Abstract: the headline accuracy figures are given without any mention of dataset sizes, class balance, number of independent runs, or statistical significance testing; this is a presentation issue that reduces immediate assessability of the empirical claims.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary, significance assessment, and recommendation of minor revision. The review accurately captures the core contribution of applying Dempster-Shafer Theory for sensor fusion in IoT fault detection, along with the dual evaluation on benchmark and laboratory data.

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper applies standard Dempster-Shafer Theory combination rules to basic probability assignments derived from sensor measurements for fault detection, then reports empirical accuracies (99.8% benchmark, 99.9% lab) against external datasets and ML baselines. No equations reduce a claimed prediction to a fitted input by construction, no self-citation chain bears the central result, and the method is presented as a direct application without renaming known results or smuggling ansatzes. The reported performance is therefore falsifiable on the stated data and protocol.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Review is based solely on the abstract; no free parameters, axioms, or invented entities are described. The central claim rests on the applicability of standard Dempster-Shafer combination rules to sensor fault data.

pith-pipeline@v0.9.0 · 5759 in / 1077 out tokens · 27396 ms · 2026-05-25T17:39:31.161003+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

26 extracted references · 26 canonical work pages

  1. [1]

    A. Abid, A. Kachouri, and A. Mahfoudhi. Outlier detection for wire less sensor networks using density-based clustering approach. IET Wireless Sensor Systems , 7(4):83–90, 2017

    work page 2017

  2. [2]

    Aggarwal

    Charu C. Aggarwal. Outlier Analysis , pages 237–263. Springer Interna- tional Publishing, 2015

    work page 2015

  3. [3]

    J. An, M. Hu, L. Fu, and J. Zhan. A novel fuzzy approach for co mbining uncertain conflict evidences in the dempster-shafer theory. IEEE Access, 7:7481–7501, 2019

    work page 2019

  4. [4]

    Dempster

    Arthur P. Dempster. A Generalization of Bayesian Inference , pages 73–104. Springer Berlin Heidelberg, Berlin, Heidelberg, 2008

    work page 2008

  5. [5]

    Djenouri, A

    Y. Djenouri, A. Belhadi, J. C. Lin, and A. Cano. Adapted k-neare st neigh- bors for detecting anomalies on spatiotemporal traffic flow. IEEE Access, 7:10015–10027, 2019

    work page 2019

  6. [6]

    Mokhtar, and Osman Hegazy

    Asmaa Fawzy, Hoda M.O. Mokhtar, and Osman Hegazy. Outliers de tec- tion and classification in wireless sensor networks. Egyptian Informatics Journal, 14(2):157 – 164, 2013

    work page 2013

  7. [7]

    Ghorbel, W

    O. Ghorbel, W. Ayedi, H. Snoussi, and M. Abid. Fast and efficient ou t- lier detection method in wireless sensor networks. IEEE Sensors Journal , 15(6):3403–3411, June 2015

    work page 2015

  8. [8]

    Intrusion detection systems in the Internet of things: A c ompre- hensive investigation

    Somayye Hajiheidari, Karzan Wakil, Maryam Badri, and Nima Jafari Nav- imipour. Intrusion detection systems in the Internet of things: A c ompre- hensive investigation. Computer Networks , 2019

    work page 2019

  9. [9]

    Identification of outliers , volume 11

    Douglas M Hawkins. Identification of outliers , volume 11. Springer, 1980

    work page 1980

  10. [10]

    L. Jin, J. Chen, and X. Zhang. An outlier fuzzy detection metho d using fuzzy set theory. IEEE Access, 7:59321–59332, 2019

    work page 2019

  11. [11]

    W. Lu, Y. Cheng, C. Xiao, S. Chang, S. Huang, B. Liang, and T. H uang. Unsupervised sequential outlier detection with deep architecture s. IEEE Transactions on Image Processing , 26(9):4321–4330, Sep. 2017

    work page 2017

  12. [12]

    An analysis of fault dete ction strategies in wireless sensor networks

    Thaha Muhammed and Riaz Ahmed Shaikh. An analysis of fault dete ction strategies in wireless sensor networks. Journal of Network and Computer Applications, 78:267 – 287, 2017

    work page 2017

  13. [13]

    Qin and F

    B. Qin and F. Xiao. A non-parametric method to determine basic p robabil- ity assignment based on kernel density estimation. IEEE Access, 6:73509– 73519, 2018

    work page 2018

  14. [14]

    Reppa, P

    V. Reppa, P. Papadopoulos, M. M. Polycarpou, and C. G. Panay iotou. A distributed architecture for hvac sensor fault detection and isola tion. IEEE Transactions on Control Systems Technology , 23(4):1323–1337, July 2015

    work page 2015

  15. [15]

    A mathematical theory of evidence , volume 42

    Glenn Shafer. A mathematical theory of evidence , volume 42. Princeton university press, 1976

    work page 1976

  16. [16]

    Detection of urban- induced rainfall anomalies in a major coastal city

    J Marshall Shepherd and Steven J Burian. Detection of urban- induced rainfall anomalies in a major coastal city. Earth Interactions , 7(4):1–17, 2003

    work page 2003

  17. [17]

    Suthaharan, M

    S. Suthaharan, M. Alzahrani, S. Rajasegarar, C. Leckie, and M. Palaniswami. Labelled data collection for anomaly detection in wire- less sensor networks. In 2010 Sixth International Conference on Intelligent Sensors, Sensor Networks and Information Processing , pages 269–274, Dec 2010

    work page 2010

  18. [18]

    E. U. Warriach and K. Tei. Fault detection in wireless sensor netw orks: A machine learning approach. In 2013 IEEE 16th International Conference on Computational Science and Engineering , pages 758–765, Dec 2013

    work page 2013

  19. [19]

    J. Xia, Y. Feng, L. Liu, D. Liu, and L. Fei. An evidential reliability indicator-based fusion rule for dempster-shafer theory and its a pplications in classification. IEEE Access, 6:24912–24924, 2018

    work page 2018

  20. [20]

    K. Xie, X. Li, X. Wang, J. Cao, G. Xie, J. Wen, D. Zhang, and Z. Qin . On-line anomaly detection with high accuracy. IEEE/ACM Transactions on Networking , 26(3):1222–1235, June 2018

    work page 2018

  21. [21]

    Yessembayev, D

    A. Yessembayev, D. Sarkar, and F. Sikder. Detection of good and bad sensor nodes in the presence of malicious attacks and its application to data aggregation. IEEE Transactions on Signal and Information Processing ove r Networks, 4(3):549–563, Sep. 2018

    work page 2018

  22. [22]

    T. Yu, X. Wang, and A. Shami. Recursive principal component an alysis- based data outlier detection and sensor data aggregation in iot sys tems. IEEE Internet of Things Journal , 4(6):2207–2216, Dec 2017

    work page 2017

  23. [23]

    Zervas, A

    E. Zervas, A. Mpimpoudis, C. Anagnostopoulos, O. Sekkas, an d S. Had- jiefthymiades. Multisensor data fusion for fire detection. Information Fu- sion, 12(3):150 – 159, 2011

    work page 2011

  24. [24]

    Zhang, Q

    H. Zhang, Q. Zhang, J. Liu, and H. Guo. Fault detection and rep airing for intelligent connected vehicles based on dynamic bayesian network mo del. IEEE Internet of Things Journal , 5(4):2431–2440, Aug 2018

    work page 2018

  25. [25]

    H. Zhu, S. Feng, and F. Yu. Parking detection method based on finite- state machine and collaborative decision-making. IEEE Sensors Journal , 18(23):9829–9839, Dec 2018

    work page 2018

  26. [26]

    Fault detection in wirele ss sensor networks through svm classifier

    Salah Zidi, Tarek Moulahi, and Bechir Alaya. Fault detection in wirele ss sensor networks through svm classifier. IEEE Sensors Journal , 18(1):340– 347, 2017

    work page 2017