By continuing to browse the site, you are agreeing to our use of cookies. (More details)

PhD (H/F) - EM field reconstruction in 5G with machine learning.

This offer is available in the following languages:
Français - Anglais

Ensure that your candidate profile is correct before applying. Your profile information will be added to the details for each application. In order to increase your visibility on our Careers Portal and allow employers to see your candidate profile, you can upload your CV to our CV library in one click!

Faites connaître cette offre !

General information

Reference : UMR8520-LAUCLA-001
Workplace : VILLENEUVE D ASCQ
Date of publication : Monday, April 27, 2020
Scientific Responsible name : Laurent Clavier, Joe Wiart, Davy Gaillot
Type of Contract : PhD Student contract / Thesis offer
Contract Period : 36 months
Start date of the thesis : 1 September 2020
Proportion of work : Full time
Remuneration : 2 135,00 € gross monthly

Description of the thesis topic

I. SUMMARY OF THE PHD
With the increasing use of wireless communication facilities [1], we live in a permanent electromagnetic field (EMF) that has induced concern and risk perception despite existing regulation. The people in charge of networks deployment, the ones who check the compliance of EMF exposure to safety limits are facing such questions [2]. The assessment of exposure requires specific equipment or simulation tools that can be complex to operate, contributing therefore to increase the gap between lay people and experts. The 5G deployment, that is just starting, reinforce these issues [3, 4] bringing new uncertainties resulting from the ever increasing band used and the beamforming induced by massive MIMO.

This PhD will address these questions, several locks remaining to be overcome. The objective is to be able to build an EMF exposure map. Today the EMF monitoring is carried out using measurement campaigns. These in situ measurements are of great interest but they cannot be performed everywhere. Taking advantage of progress in connected devices technologies, the exposure assessment has been investigated using connected sensors. Recently, wireless sensors, autonomous in energy, have been designed, and proposed. Associated in networks, these sensors are of great interest since they can grab the exposure's temporal variations; nevertheless, since they are still localized they cannot, alone, provide an exposure mapping because of unsampled locations. Reconstruction using measurements localized in some specific places has been investigated using methods based on geo-statistic and Gaussian processes [5, 6, 7]. Recently works have been carried out using Artificial Intelligence and machine learning [8, 9]. These works have been performed using simulations taking into account information that are available through databases (like the position of the base stations) as well as drive testing. Since these works dealt with simulations, confrontation with measures will require an adaptation of the method to increase the accuracy and take into account specific properties of the EM field. Many improvements can be studied. The first one is to include some data from the field, like the city density, the position of the base station, or any other significant parameters, which are publicly available. It is also possible to manage sensors of different qualities and to include mobile sensors to improve the reliability of the reconstruction. Taking into account the density of the sensors, the imprecision of the location and the measurement and the temporal evolution of the EM field also constitute research challenges.

II. ORGANISATION OF THE WORK
The work is based on data that will be measured in different locations in Lille. A large number of sensors (≈50) will allow to cover an area in the city center and another set of sensors will cover another place. The objective of the work is to develop a machine learning process to analyze the data and reconstruct in space and time the electro-magnetic field. This will allow to evaluate the exposure of a given person in the city.

III. REQUIRED SKILLS
Telecommunication, 5G. Part of the PhD is devoted to data processing and, possibly, measurement campaigns. Knowledge of Telecommunications is important and experimental abilities would be appreciated.
Machine learning. Python, Anaconda, Tensorflow

IV. FURTHER INFORMATION
For further information, contact Joe Wiart, Davy Gaillot and Laurent Clavier.

REFERENCES
[1] L. Clavier, T. Pedersen, I. Rodriguez, M. Lauridsen, and M. Egan. Experimental Evidence for Heavy Tailed Interferencein the IoT. Working paper or preprint; https://hal.archives-ouvertes.fr/hal-02521928/file/ExpWCL4p.pdf, March 2020.
[2] Joe Wiart.Radio-Frequency Human Exposure Assessment: From Deterministic to Stochastic Methods. 01 2016.
[3] Miroslava Karaboytcheva. Effects of 5g wireless communication on human health. Briefing PE 646.172, EPRS – European Parliamentary Research Service, March 2020.
[4] M. Egan, L. Clavier, M. de Freitas, L. Dorville, J. Gorce, and A. Savard. Wireless communication in dynamic interference. In GLOBECOM 2017 - 2017 IEEE Global Communications Conference, pages 1–6, 2017.
[5] Sam Aerts, D. Deschrijver, L. Verloock, T. Dhaene, Luc Martens, and Wout Joseph. Assessment of outdoor RF-EMF exposure through hotspot localization using kriging-based sequential sampling. Environmental research, 2013.
[6] A. Solin, M. Kok, N. Wahlstrom, T. B. Schon, and S. Sarkka. Modeling and interpolation of the ambient magnetic field by gaussian processes. IEEE Transactions on Robotics, 34(4):1112–1127, 2018.
[7] Thomas Lemaire, Joe Wiart, and Philippe Doncker. Variographic analysis of public exposure to electromagnetic radiation due to cellular base stations: Variographic analysis of BTS EMF exposure. Bioelectromagnetics, 37, 10 2016.
[8] Sam Aerts, Joe Wiart, Luc Martens, and Wout Joseph. Assessment of long-term spatio-temporal radio frequency electromagnetic field exposure. Environmental research, 161:136–143, 11 2017.
[9] S. Wang and J. Wiart. Sensor aided emf exposure assessments in urban environment using artificial neural networks. Int. Journal Environmental research and public hereconstructionalth, 2020

Work Context

IEMN brings together in a single structure the essentials of regional research in a vast scientific field ranging from nanosciences to instrumentation.

Bringing together researchers with different cultures, approaches and motivations, building a continuity of knowledge ranging from fundamental problems to applications is what makes IEMN unique today. Nearly 500 people, including around 100 international researchers, work together.

The core of our activities is centered on micros and nanotechnologies and their applications in the fields of information, communication, transport and health. Our researchers have at their disposal exceptional experimental means, in particular technology and characterization centers whose possibilities and performances are at the best European level.

The date of recruitment may be postponed until the end of the confinement period.

We talk about it on Twitter!