Machine Learning Modeling for Radiofrequency Electromagnetic Fields (RF-EMF) Signals From mmWave 5G Signals

Al-Jumaily, Abdulmajeed; Sali, Aduwati; Riyadh, Mohammed; Wali, Sangin Qahtan; Li, Lu; Osman, Anwar Faizd

doi:10.1109/access.2023.3265723

Cited by 5 publications

(1 citation statement)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Machine learning and its subclass, deep learning, have been successfully used. Related to EMF exposure, a series of papers have been published in the last few years [45][46][47][48][49].…”

Section: Introductionmentioning

confidence: 99%

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

Miclaus,

Deaconescu,

Vatamanu

et al. 2023

Technologies

View full text Add to dashboard Cite

To gain a deeper understanding of the hotly contested topic of the non-thermal biological effects of microwaves, new metrics and methodologies need to be adopted. The direction proposed in the current work, which includes peak exposure analysis and not just time-averaged analysis, aligns well with this objective. The proposed methodology is not intended to facilitate a comparison of the general characteristics between 4G and 5G mobile communication signals. Instead, its purpose is to provide a means for analyzing specific real-life exposure conditions that may vary based on multiple parameters. A differentiation based on amplitude-time features of the 4G versus 5G signals is followed, with the aim of describing the peculiarities of a user’s exposure when he runs four types of mobile applications on his mobile phone on either of the two mobile networks. To achieve the goals, we used signal and spectrum analyzers with adequate real-time analysis bandwidths and statistical descriptions provided by the amplitude probability density (APD) function, the complementary cumulative distribution function (CCDF), channel power measurements, and recorded spectrogram databases. We compared the exposimetric descriptors of emissions specific to file download, file upload, Internet video streaming, and video call usage in both 4G and 5G networks based on the specific modulation and coding schemes. The highest and lowest electric field strengths measured in the air at a 10 cm distance from the phone during emissions are indicated. The power distribution functions with the highest prevalence are highlighted and commented on. Afterwards, the capability of a convolutional neural network that belongs to the family of single-shot detectors is proven to recognize and classify the emissions with a very high degree of accuracy, enabling traceability of the dynamics of human exposure.

show abstract

“…Machine learning and its subclass, deep learning, have been successfully used. Related to EMF exposure, a series of papers have been published in the last few years [45][46][47][48][49].…”

Section: Introductionmentioning

confidence: 99%