HERWICT

Human Exposure to Radiation from new Wireless Comunication Technologies using Advanced Electromagnetic-Thermal Dosimetry Models

A necessity for higher transmission rate and network capacity has led to the use of the mm-wave spectrum for 5G mobile systems which require appreciably higher number of antennas. Public concern on potential adverse health effects due to exposure to 5G systems have become a rather hot issue, hence additional research is required. Recently released guidelines IEEE and ICNIRP have merged towards 6GHz as a unique transition frequency above which absorbed power density (APD) averaged over a specific area is suggested in the near field, instead of specific absorption rate (SAR) being averaged over tissue volume. These international guidelines are soon expected to be implemented in national regulations, such as Croatian and Slovenian.

The proposed project deals with deterministic-stochastic electromagnetic-thermal dosimetry procedures for exposure of humans to 5G systems. We propose a combination of techniques: a conformal method for incident field dosimetry, an accurate model for blood flow and coupled flow – heat transfer model of the human tissue subjected to internal electromagnetic dosimetry.

The principal goal of the project is to develop the realistic models accompanied with highly accurate numerical methods such as Boundary Element Method, Finite Element/Volume Method and hybrid techniques. The obtained numerical results for APD and related temperature increase will be compared to the experimental results obtained using a head phantom. As deterministic computational models are not enough to represent relevant parameters which characterize realistic scenarios they will be combined with stochastic approach, featuring non-intrusive stochastic collocation method.

Finally, as the developed techniques are expected to require substantial computational resources, we will adopt modern approximation techniques, such as physics informed neural networks to develop fast approximate models suitable for use without supercomputers.