Sequential detection of a partially known transient signal for air quality monitoring
ABG-130090 | Thesis topic | |
2025-03-26 | Public funding alone (i.e. government, region, European, international organization research grant) |

- Data science (storage, security, measurement, analysis)
Topic description
Air quality monitoring is a major societal challenge. The World Health Organization (WHO) estimates that air pollution is responsible for approximately 7 million deaths per year. To address this issue, numerous pollutant measurement stations (monitoring fine particles, nitrogen dioxide, ozone, sulfur dioxide, carbon monoxide, and volatile organic compounds) have been deployed. While these stations provide precise measurements and allow for macroscopic pollution detection, their limited number prevents localized detection within urban areas. The high cost and bulkiness of these stations hinder their widespread deployment.
To enable local pollution detection, low-cost micro-sensors have emerged. These sensors are significantly less accurate than fixed monitoring stations but are inexpensive and easy to install. Combining data from multiple micro-sensors and integrating them with fixed stations could provide more reliable and extensive pollution monitoring in cities. However, determining the optimal method for combining these different sources to produce meaningful results remains a challenge.
Many approaches using black-box algorithms (such as Machine Learning and Deep Learning) have been developed for this problem ([1], [2]). However, the interpretability and explainability of these models remain open questions.
In collaboration with our partner AtMO Grand Est, we have developed exploratory research on sequential detection algorithms capable of combining multiple fixed stations to achieve more accurate results than a single station alone ([3], [4], [5]). These algorithms can detect anomalies with high probability while maintaining a false alarm rate predefined by the operator. However, applying these algorithms to micro-sensors requires further research, particularly in modeling the pollutant measurements from these devices.
Additionally, it is essential to study how to sequentially detect pollution events when the pollution profile is uncertain. This study will account for variations in micro-sensor performance across different environmental conditions.
Ultimately, the proposed methods for air pollution detection and localization could contribute to a broader decision-making framework, helping to identify and target pollution hotspots, whether from specific zones or industrial sources. This could enable real-time local interventions to improve air quality.
References
[1] N. H. Motlagh et al., "Toward Massive Scale Air Quality Monitoring," in IEEE Communications Magazine, vol. 58, no. 2, pp. 54-59, February 2020, doi: 10.1109/MCOM.001.1900515.
[2] D. Zhang and S. S. Woo, "Real Time Localized Air Quality Monitoring and Prediction Through Mobile and Fixed IoT Sensing Network," in IEEE Access, vol. 8, pp. 89584-89594, 2020, doi: 10.1109/ACCESS.2020.2993547.
[3] F. E. Mana, B. K. Guépié, I. Nikiforov, "Centralized and Decentralized Strategies for Sequential Detection of Transient Changes," IFAC-PapersOnLine, vol. 55, issue 6, 2022, pp. 360-365.
[4] F. E. Mana, B. K. Guépié, R. Deprost, E. Herber, I. Nikiforov, "Air Pollution Monitoring by Sequential Detection of Transient Changes," IFAC-PapersOnLine, vol. 55, issue 5, 2022, pp. 60-65.
[5] F. E. Mana, B. K. Guépié & I. Nikiforov (2023) "Sequential Detection of an Arbitrary Transient Change Profile by the FMA Test," Sequential Analysis, DOI: 10.1080/07474946.2023.2171056.
Starting date
Funding category
Funding further details
Presentation of host institution and host laboratory
Research Unit: Laboratoire Informatique et Société Numérique (LIST3N)
Website LIST3N : http://recherche.utt.fr/list3n
Website :
PhD title
Country where you obtained your PhD
Institution awarding doctoral degree
Graduate school
Candidate's profile
"Applicants for this PhD position must meet the following requirements: Education: Master’s degree or engineering diploma in signal processing, applied mathematics, mathematical statistics, or data science. Skills in modeling and data analysis: Signal processing and time series analysis Applied statistics and probabilistic model estimation Hypothesis testing theory Programming and tools: Languages: Python and MATLAB Research capabilities: Scientific curiosity and rigor Ability to work in a team and collaborate with industrial partners Ability to read and leverage scientific articles Scientific writing and communication of results in English and French"
Vous avez déjà un compte ?
Nouvel utilisateur ?
Get ABG’s monthly newsletters including news, job offers, grants & fellowships and a selection of relevant events…
Discover our members
MabDesign
Nokia Bell Labs France
ONERA - The French Aerospace Lab
Ifremer
Généthon
ASNR - Autorité de sûreté nucléaire et de radioprotection - Siège
CESI
Aérocentre, Pôle d'excellence régional
Tecknowmetrix
Institut Sup'biotech de Paris
Laboratoire National de Métrologie et d'Essais - LNE
ADEME
ANRT
SUEZ
Groupe AFNOR - Association française de normalisation
MabDesign
CASDEN
TotalEnergies
PhDOOC