Tutorials

Deterministic-Stochastic Methods in Electromagnetic Thermal Dosimetry for the Assessment of Human Exposure to Electromagnetic Fields in GHz Range

Dragan Poljak, PhD

University of Split, FESB, Split, Croatia

Although deterministic numerical techniques provide a satisfactory insight into bioelectromagnetics
phenomena in GHz range, there are still some issues that cannot be addressed efficiently by using
deterministic modeling only. The biological tissues are represented by electrical properties;
permittivity, conductivity and permeability. However, the exact value of these parameters is not
known, as their values may vary due to frequency, gender, age, or health of a person, respectively.
Most of these parameters are obtained under different measurement on ex vivo animal and human
tissues, and exhibit large variations from their averages. When used in computational models, these
average values lead to rough approximation of the realistic scenarios.
The uncertainty from the input is, hence, inevitably propagated to the output of interest such as;
induced field, specific absorption rate (SAR) or absorbed power density (Sab).
These problems could be somewhat overcome by a combination of deterministic models with
stochastic methods. Therefore, the uncertainty quantification (UQ) deals with the uncertainties in
the model response, due to the input parameters being random variables, while sensitivity analysis
(SA) provides the information on the impact of the mutual interactions between the input variables
to the output value of interest. Usual UQ methods rely upon statistical approaches, such as Monte
Carlo (MC) simulations, being easy to implement but the sample size needs to be very high
(>100.000), resulting in a slow convergence rate. This drawback could be overcome by using
sophisticated stochastic modeling.
Tutorial first covers deterministic approaches in electromagnetic-thermal dosimetry with particular
emphasis to exposures in GHz frequency range mostly pertaining to 5G systems. The use of integral
equation approaches featuring Method of Moments(MoM) and Boundary Element Methods (BEM),
and differential equation approaches such as Finite Element Methods (FEM) is addressed. Some
strengths and weaknesses of both approaches are discussed throughout Tutorial as well.
Tutorial then deals with stochastic procedures in electromagnetic-thermal dosimetry in lower
portion of GHz frequency range featuring the use of Stochastic Collocation Method (SCM).
Several examples pertain to the analysis of planar and non-planar multilayer tissue models and to
the analysis of anatomically based realistic multi-layered models of the head exposed to radiation
from 5G communication systems.
The presentation is based on:

• D. Poljak, Deterministic-Stochastic Modeling in Electromagnetic-Thermal Dosimetry, IEEE
  ICES Workshop on Computational Bioelectromagnetics (virtual), Feb. 2024.
• D. Poljak, A Susnjara, Deterministic-Stochastic Modeling in Computational electromagnetics,
  IEEE Press/Wiley, New Jersey, 2024.