Assessment of EMF Exposure from Urban Sensor Measurements by Using Artificial Neural Network

Abstract

This paper studies the electromagnetic field (EMF) exposure emitted by base stations (BSs) from cellular networks in the urban city environment. We reconstruct the EMF exposure by using artificial neural network (ANN) based on data measured by sensors. We take consideration of spatial locations of real BSs in 14th district of Paris, time variation and antenna orientation. And most importantly, we propose a new path loss model to capture the Light-of-Sight (LoS) and None-Light-of-Sight (NLoS) effects caused by complicated and varying blockages in urban cities. By applying the ANN, we are able to reconstruct EMF exposure for the locations of interest with R2 up to 0.767. And we provide results under different splitting percentage of training and testing data.