AMPERE

Projet Completed
ANSES

The AMPERE project is dedicated to the characterization of residential exposure to radio frequency waves

Project partners

Télécom Paris, France / INERIS, France / ULB, Belgium / IMEC, Belgium / CNR, Italy / UU, the Netherlands

Project duration

40 months – End of the project : March 2020

Project participants

  • Joe Wiart : Télécom Paris, COMELEC/C2M Chair – Paris
  • Philippe De Doncker : Université Libre de Bruxelles, OPERA-WCG CP 165/81 – Brussels, Belgium
  • Wout Joseph : iMinds – Gent, Belgium
  • Paolo Ravazzani : Consiglio Nazionale delle Ricerche – Milan, Italy
  • Julien Caudeville : INERIS, Direction des Risques Chroniques, Pôle RISK, Unité ISAE
  • Roel Vermeulen : IRAS, Universiteit Utrecht, The Netherlands
  • Project objectives

    The AMPERE projects aims to develop an advanced approach to producing a space-time map of exposure to electromagnetic fields (EMFs), using advanced statistical tools.

    Computing the evaluation of EMF space-time distribution can take a long time because of extended measurements and data that is not readily accessible to the general public. The combination of exposure to EMFs, use of ITCs, population density and the environmental characteristics involved in merging the data is therefore significantly curtailed. In order to overcome the latter, AMPERE will develop advanced methods based on statistics and substitution models.

    In addition, in order to be of use to the agencies tasked with assessing these risks – together with the epidemiologists and the general public, the toolkits involved in the process of merging data will be as simple as possible and will not require too much expertise in telecommunications, propagation or electromagnetism.

  • Search axes

    Modeling exposure to external EMFs:

    External exposure will be characterized by using measurements (wide-band and portable exposure meters and trackers) and simulations carried out by our partners. Exposure induced by access points and devices will be estimated in various environments, frequency bands and services (for example, cell networks and radio broadcasting). With several teams involved in AMPERE, exposure will be produced in different countries and cities, in a range of socioeconomic groups and communication network architectures. In addition to the project’s specifically acquired data, existing data will be collected from partners’ databases and previous projects, for an effective use of resources and the study of changes in exposure over time.

    Taking into account the work carried out as part of previous projects (ANR Samper and B WARE for example), measurements and simulations, together with advanced statistical tools such as stochastic geometry (Yu 14) and geostatistics (Kriging for instance), will be combined to interpolate measurements and simulations. Sensitivity analysis will be carried out using several methods (Sobol’, Kucherenko 12), to identify the parameters with most influence on exposure.

    Modeling exposure to internal EMFs:

    Internal exposure assessment is affected by uncertainty and a lack of information on the internal environment and the location of internal sources. It is therefore difficult to use a deterministic approach, such as ray-tracing methods. A new statistical model of internal exposure to EMFs is necessary to overcome these limitations. Internal exposure will first be characterized by measurements (wide-band and portable exposure meters and trackers) and simulations with antennas located inside or outside buildings.

    Given that the environment and the location of internal antennas and users are not properly known, the main goal of this working paper (WP) is to build a new statistical model of internal exposure to EMFs. To be usable, a stochastic model will need to use a certain number of integral parameters, such as the kind of building, its height and the number of its windows, to produce typical maps of an internal residential environment. For sources located inside, statistical distribution is defined by parameters such as averages and quantiles that depend on the environment (for example, a shopping area, a detached house, an apartment block and so on). Based on sensitivity analysis, a simplified model of internal exposure using advanced stochastics will be built and the resulting data will be compared to the existing data collected for internal exposure.

    Advanced substitution modeling of the temporal variation of residential exposure

    Exposure changes over time (over the course of a day for example) because of signal variation and uncertainty caused by environmental changes. In fact, modeling the same measurement in the same conditions does not necessarily produce the same result, because of the flow of people, traffic, etc. As a result, the previous algorithms need to be able to deal with the stochastic variability inherent to a dynamic environment. This work group will stochastically model and factor in the temporal characteristics of exposure such as fading, (cell) traffic, non-regular (WiFi) and rare emissions (IoT). The objective of this working paper is to study the temporal variation of exposure in the area examined and develop procedures that enable models to include time variables, thanks to stochastic modeling. We will examine how to use exposure measurements in the temporal domain in a carefully selected number of measurement locations in the area (Paris, Amsterdam, Ghent, etc.) to produce an exposure map in the region’s temporal domain.

    Firstly, we will start our research at a specific measurement point and undertake a detailed analysis of various time measurements (for example, variations in a day, and from one day to next). Secondly, in order to filter the temporal elements of noise, we will study and compare a range of smoothing algorithms (such as the distance weighted least squares smoothing, bicubic spline smoothing, exponentially weighted negative smoothing , etc.). Note that there are no entirely automated algorithms that could identify elements of trends in exposure time data. Once the chronological series are transformed, they will be non-biased by outliers, making it easier to model and predict time series. The literature shows that several algorithms can be applied to carry out this analysis in the temporal domain (self-correlation and cross-correlation analysis) and in the frequency domain (spectral analysis and wavelet analysis). Thirdly, we will combine (WP5) the temporal modeling solutions with the spatial modeling algorithms, which will be developed in WP2 and WP3. This will result in new procedures for designing exposure maps over time, able to process time measurements in multiple locations in the zone of interest. By applying this methodology to various fields, new strategies will be designed to assess exposure. The new models that allow space and time variations in the exposure map will be compared to the models used previously. They will also be disseminated as part of ongoing epidemiological studies. The application will provide information on the degree of accuracy gained in exposure assessment in epidemiological studies. It will also provide more precise analysis over time (exposure when people are at home for example), which could lead to more precise risk prediction.

    Fusion data

    By using AMPERE partners’ current experience in fusion data and previous and ongoing projects, the space and time characteristics of internal and external exposure to EMFs estimated in WPs 2, 3 and 4, as well as geographic and socioeconomic data will be combined in an advanced map of residential RF exposure. This provides information on the temporal variations, frequency and quantiles of exposure. The PLAINE platform developed by INERIS will be adapted to the external and internal space and time models developed as part of this project. The improved platform will be used to collect, incorporate and analyze data on RF sources, EMF measurement, exposure and demographic data. Stochastic computing tools and substitution models will be used to add new levels of complexity in exposure assessment, by incorporating the space-time budget for the target population. Geospatial mobility will be incorporated in the form of a piecewise linear trajectory in a 3D space (dimensions x and y plus the time dimension). Additional exposure covariables, such as socioeconomic status, demographics, and people’s behavior, will be included to complete exposure assessment. We will build scenarios, based on national cases and studies, to estimate the temporal frequency of individuals and populations in a range of micro-environments. Profiles of exposure to the microenvironment will draw on WPs 2, 3 and 4. A model of exposure linked to the space and time model developed as part of a non-spatial database will produce exposure indicators, as well as interesting related statistics. To that end, it will use a Monte-Carlo method to combine the microenvironment’s space and time variability with the individual/population scenario.

References

[bloch 08]I. Bloch (Ed), (2008), Information Fusion in Signal and Image Processing, London, UK, ISTE, Wiley.

[bloch 96]I. Bloch, (1996), “Information Combination Operators for Data Fusion: A Comparative Review with Classification”, IEEE Systems, Man and Cybernetics, vol. 26, n° 1, pp. 52-67.

[Caudeville 12] Caudeville J, et al., 2012 A. Development of a spatial stochastic multimedia model to assess population exposure at a regional scale. Journal of Total Environment 2012,

[Gajšek 15] Gajšek et al.,. Electromagnetic field exposure assessment in Europe radiofrequency fields (10 MHz–6 GHz), JESEE 2015, 25 (1):37-44.

[Isselmou 06] Isselmou YO et al;, Geostatistical interpolation for mapping radio-electric exposure level , First European Conference On Antennas and Propagation, 2006. EuCAP 2006, Nice, France.

[Isselmou 08] Isselmou YO et al. Geostatistical estimation of electromagnetic exposure, in geoENV VI,, Proceedings of the Sixth European Conference on Geostatistics for Environmental Applications, 2008, Part I: 59-70

[Jawad 15] Jawad, O., Dricot, J.-M., Benlarbi-Delaï, A., Lautru, D., De Doncker, P., “Study of human exposure using kriging method”, Progress in Electromagnetics Research B, 61(1), 241-252, 2015

[Joseph 09] W. Joseph et al., In-situ measurement procedures for temporal RF electromagnetic field exposure of the general public, Health Phys 96(5) (2009)

[Joseph 10] W. Joseph et al, Influence of mobile phone traffic on general public base station exposure, Health Physics 99(5) (2010) 631-638.

[Kersaudy 15] P Kersaudy et al., A new surrogate modelling technique combining Kriging and polynomial chaos expansions–Application to uncertainty analysis in computational dosimetry Journal of Computational Physics 286, 103-117 (2015)

[Kucherenko 12] Kucherenko S., et al.. Estimation of global sensitivity indices for models with dependent variables. Computer Physics Communications, (2012). 183 :.

[Liorni 15] I Liorni, et al., Study of the exposure of 1 year-old infant to 3G tablet and femtocell using Polynomial Chaos theory. Electromagnetics in Advanced Applications (ICEAA), 2015

[Mahfouz 13] Z. Mahfouz, et al.,, Comparison of the Temporal Realistic RF Exposure with Worst-Case Estimation in Two Countries, Radiat Prot Dosimetry (2013) doi:10.1093/rpd/nct155.

[Neubauer 07] Neubauer G,et al., . Feasibility of future epidemiological studies on possible health effects of mobile phone base stations Bioelectromagnetics, 2007, 28 (3): 224-230.

[Saib 14] Saib M. S et al. (2014) Spatial relationship quantification between environmental, socioeconomic and health data at different geographic levels. Intel J of Environmental Research and Public Health, 11 (4):

[Sidi 14] Sidi HBA, et al. A. Self-opt,mizing mechanisms for EMF reduction in heterogeneous networks, 12th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks, WiOpt 2014, Hammamet; Tunisia, 12-16 May 2014,

[Sobol 01] Sobol’, I. M. (2001). Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates. Mathematics and Computers in Simulation, 55.

[Tesanovic 14] Tesanovic M et al The LEXNET project: wireless networks and EMF: paving the way for low-EMF networks of the future, IEEE Vehicular Technology Magazine, 2014, 9(2):

[Varsier 15] Varsier N et al., A novel method to assess the human population exposure induced by a wireless cellular network Bioelectromagnetics, 2015

[Vermeeren 13] G. Vermeeren et al., ,, Spatial and temporal RF electromagnetic field exposure of children and adults in indoor micro environments in Belgium and Greece, Progress in Biophysics & Molecular Biology (2013), 2013.

[Yu 14] Yu X, et al., Crack STIT tessellations for city modeling and impact of terrain topology on wireless propagation, 12th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks, WiOpt 2014, Hammamet; Tunisia, 12-16 May 2014.