An Exposimetric Electromagnetic Comparison of Mobile Phone Emissions: 5G versus 4G Signals Analyses by Means of Statistics and Convolutional Neural Networks Classification

Abstract

Overview

To gain a deeper understanding of the debated non-thermal biological effects of microwaves, this study introduces new metrics and methodologies focusing particularly on peak exposure analysis rather than only time-averaged analysis.

Methodology

The study employs a methodology that analyzes specific real-life exposure conditions of mobile phone users under 4G and 5G networks, accomplished by running various applications. Tools used included signal and spectrum analyzers, measuring amplitude probability density, complementary cumulative distribution functions, and channel power, with data collated into recorded spectrogram databases.

Findings

• The study established differences in amplitude-time features of 4G versus 5G signals when operating four types of mobile applications, showing diverse conditions of user exposure.
• Distinct emission features were observed for file download, file upload, Internet video streaming, and video call applications.
• 5G emissions were generally higher in electric field strength, particularly noticeable during Internet video streaming.
• The capability of a convolutional neural network (YOLO v7) to classify emissions signals was analyzed, showing high degrees of accuracy.

Conclusions

This work contributes to understanding human exposure dynamics to EMF, outside of traditional studies limited to thermal effects. It highlights the importance of understanding emission variability, significantly augmenting current knowledge on non-thermal effects of mobile communication signals. The study emphasized the necessity for ongoing research on realistic exposure scenarios across varied conditions.