ML Algorithms Analysis and Prediction of Broadband Electric Field Levels in Telecommunication Systems Environment

This study investigates artificial intelligence neural networks to analyze and predict sub-3 GHz electric field exposure levels. The research encompasses a diverse set of environments including urban areas, residential zones, public transportation settings, and office spaces, which collectively represent realistic exposure environments for radio frequency electromagnetic field emissions. Empirical data, acquired from multiple urban and indoor settings, underpin the development of predictive models. The predictive models are developed using artificial neural network methodologies, specifically the Generalized Regression Neural Network and Radial Basis Function Neural Network. The study presents a detailed assessment of the simulation results, highlighting the effectiveness of these artificial neural network-based approaches in predicting electric field levels across varied environmental conditions. While optimized for broadband exposure, the models may not generalize to millimeter-wave frequencies, which behave differently in terms of propagation and penetration. The findings emphasize the potential of Generalized Regression and Radial Basis Function neural networks for accurate and reliable prediction of radio frequency electromagnetic field levels.