Machine Learning-Based Power Density Prediction for Mobile Handset Antenna

Compliance with electromagnetic field (EMF) exposure limits is critical in the development of 5G millimeter-wave (mm-wave) terminals equipped with array antennas, requiring peak incident power density (PD). Simulating array antennas within the frame of a mobile handset is time-consuming due to their small size. In this work, an efficient machine learning (ML) method is proposed to predict the maximum PD of array antennas within a mobile handset, significantly reducing computational time. Validation on a 4×1 cavity-backed slot array antenna demonstrated high accuracy in PD prediction, confirming its effectiveness.