Tasneem Sofri, Hasliza A Rahim, Allan Melvin Andrew, Ping Jack Soh, Latifah Munirah Kamarudin, & Nishizaki Hiromitsu. (2023). Data Normalization Methods of Hybridized Multi-Stage Feature Selection Classification for 5G Base Station Antenna Health Effect Detection. Journal of Advanced Research in Applied Sciences and Engineering Technology, 30(2), 133–140. https://doi.org/10.37934/araset.30.2.133140
Data normalization methods of hybridized multi-stage feature selection classification for 5G base station antenna health effect detection
|Author:||Sofri, Tasneem1; Rahim, Hasliza A1; Andrew, Allan Melvin2;|
1Faculty of Electronic Engineering & Technology, Universiti Malaysia Perlis, Arau 02600, Malaysia
2Advanced Communication Engineering, Centre of Excellence (ACE), Universiti Malaysia Perlis, Kangar 01000, Malaysia
3Centre for Wireless Communications (CWC), University of Oulu, 90570 Oulu, Finland
4Centre of Excellence for Advanced Sensor Technology (CEASTech), Universiti Malaysia Perlis, Kangar 01000, Perlis, Malaysia
5Integrated Graduate School of Medicine, Engineering, and Agricultural Sciences, University of Yamanashi, Kofu, 400-8510, Yamanashi, Japan
|Online Access:||PDF Full Text (PDF, 0.4 MB)|
|Persistent link:|| http://urn.fi/urn:nbn:fi-fe2023061354486
Semarak Ilmu Publishing,
|Publish Date:|| 2023-06-13
It is essential to assess human exposure to Fifth Generation (5G) Radiofrequency Electromagnetic Field (RF-EMF) signal from Base Station (BS) sources operating at Low Band 5G at 700 MHz, Sub-6 Band 5G at 3.5 GHz, and Millimeter Wave (mmWave) 5G at 28 GHz. This assessment will help determine whether 5G technology is safe for people. Inconsistent results were found in previous epidemiological studies on the health effects of radiation exposure from Mobile Phones (MP) and BS when normalization methods were used to prepare the data for Machine Learning (ML), which could lead to misclassification because of the dataset’s quality. The effects on adult health are assessed in terms of physiological parameters (body temperature, heart rates, and blood pressure), cognitive performance (brain memory, motor control, and attention), Visual Analogue Scales (VAS) for well-being parameter, and Electromagnetic Field (EMF) perception parameter. The purpose of the research was to identify changes in the physiological parameters of adult individuals before, during, or after exposure to 5G signals, including Sham (No Exposure). 12 normalization methods are selected which are Z-score normalization (z score), Linear scaling (LS), Binary normalization (BNN), Bipolar normalization (BPN), Min-Max scaling (MMS), t-score normalization (t score), Decimal Inverse Logarithmic Scaled Normalization (DILSN), Relative Mean Normalization (RMN), Relative Standard Deviation Normalization (RSDN), Variation Normalization (VN), Robust Normalization (RN) and Relative Interquartile Normalization (RIN) and based on their F-value and p-value analyses, the three best normalization techniques are selected in this research. The original distribution of each parameter data variable is different, these techniques are beneficial for converting data so that it is dimensionless and has equivalent distributions. Based on the selection criteria for the hybridized Multi-Stage Feature Selection (MSFS) classification for 5G base station antenna health effect detection, BNN, MMS, and RSDN were named as the top three normalization techniques.
Journal of advanced research in applied sciences and engineering technology
|Pages:||133 - 140|
|Type of Publication:||
A1 Journal article – refereed
|Field of Science:||
213 Electronic, automation and communications engineering, electronics
This research is funded under University-Private Matching Fund (UniPRIMA), Ref. No. UniMAP/RMC/9001-00702 (UniMAP Fund) / 9002-00142 (FILPAL (M) Sdn Bhd Fund), and Tasneem, scholarship from JASSO Student Exchange Support Program with University of Yamanashi, Japan. The Malaysian Communications and Multimedia Commission (MCMC) approval to use the data from the UniMAP research grant funded by MCMC (grant number: 2020/01/001) is obtained. MCMC allows the authors to use the data in this publication.
Beginning 2021, Journal of Advanced Research in Applied Sciences and Engineering Technology is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. All articles published in this journal are open access and freely available online, free to download, share and re-use.