?Mathematical formulae have been encoded as MathML and are displayed in this HTML version using MathJax in order to improve their display. Uncheck the box to turn MathJax off. This feature requires Javascript. Click on a formula to zoom.Abstract
Limiting human exposure to radio frequency electromagnetic fields in crowded indoor public spaces such as shopping malls, is identified as one of the key metrics in the process of green and digital environment transformation. To assess the exposure levels in shopping malls, we conducted a human centric extensive measurement campaign using three-axial frequency selective measurement equipment. Our study involved various indoor locations within the mall, capturing electrical field levels at different times and days of the week. A total of 529,340 samples were collected during working days, and 430,020 samples were collected during weekends. We measured E-field strength, power density, and calculated total exposures for frequency bands ranging from 88 MHz to 5850 MHz. The presented comparative analysis revealed that mobile communications technologies operating at 900 MHz and 1800 MHz were the main contributors to personal exposure levels in different mall environments.
Reviewing Editor:
Introduction
Diversification of wireless technologies and densification of radio frequency electromagnetic field (RF-EMF) sources aiming to support coverage and quality of experience in indoor public areas such as shopping malls, is accompanied by debates regarding possible effects for low-level exposures that are within safety limits and standards (Lin, Citation2023).
Limiting human exposure to EMF is identified as one of the key actions (Cuiñas et al., Citation2023) in the process of design thinking methodology for network planning and exploitation of EMF-constrained networks. Different approaches are being considered to find optimal balance and trade-off between EMF exposure and the performance of cellular systems and wireless local area networks (WLANs) in various operating scenario (Ibrani et al., Citation2018; Malandrino et al., Citation2022). Since in these kinds of networks the traffic is generated from both user equipment (uplink) and network infrastructure stations (downlink), the EMF constrained networks should consider both streams. One solution, based on stochastic geometry model that considers the above-mentioned constraints, is presented in Chen et al. (Citation2023).
The RF-EMF exposure problem is pronounced especially in closed indoor environments with high density of emitting nodes such as shopping malls. Therefore, the assessment of electric field (E-field) exposure levels in shopping mall-like environments through measurements is crucial, both to understand the severity and nature of the problem and verify compliance with standards, but also to leverage collected data for multiparametric network planning and optimization.
While it is possible to derive the RF-EMF characteristics in dynamic environments such as shopping malls via mathematical models or software simulations, measurement campaigns are more preferable. Such measurements can be conducted with broadband or spectrum analyser measurements in selected spots and locations. To capture real-life human exposure scenario, personal exposure meters (PEMs) may also be used. The comparison between different measurement instrumentations in terms of their technical specification such as sensitivity and standard uncertainty due to axial isotropy in the vertical plane in each band has been presented in Bhatt et al. (Citation2016) and Paniagua-Sánchez et al. (Citation2023). Summarized RF-EMF measurement data in shopping malls can be fed to Machine Learning models or Artificial Neural Networks to extrapolate or predict environment-oriented and/or human-centric RF-EMF, as elaborated for similar scenarios (Chikha et al., Citation2023; Tognola et al., Citation2021).
While there are works that show that the overall downlink exposure levels may exhibit a Gaussian distribution in certain indoor environments (Mulugeta et al., Citation2023), results of measurements conducted with broadband equipment and a spectrum analyser in a shopping mall presented in (Engiz & Kurnaz, Citation2017) show that E-field levels in these environments do not exhibit a normal distribution. However the authors in (Engiz & Kurnaz, Citation2017) show that the statistical properties vary depending on the day of the week and can be modelled with the non-parametric approach, while the mean measured total E- field (100 kHz–3 GHz) is 0.59 V/m. A comparative study in two different shopping malls, conducted with PEMs, found that Uplink mobile component and wide-band internet access may be the most significant contributors in RF-EMF exposure (Karpowicz et al., Citation2018). The results of the case study of E-field levels in an underground shopping mall are given in (Onishi et al., Citation2021), while a recently published study (da LA Silva et al., Citation2023), re-confirms that there are a limited number of publications on this topic. Their measurements show that the E-field highest average is 1.96 V/m for shopping mall and scenarios under the study. A comprehensive study from an empirical and simulation perspective of the RF-EMF exposure in shopping malls (Celaya-Echarri et al., Citation2021) and points future work directions. Another study focused on estimating the distribution of the specific absorption rate (SAR) in humans using surrogate modelling and stochastic dosimetry caused RF-EMF due to wireless communications (Jiang et al., Citation2023).
The objective of this study is to derive the measurement-based human RF-EMF exposure levels in shopping malls for main technologies operating in sub 6 GHz. Our measurement campaign collected in total 959,360 E-field samples. Human-centric frequency selective measurements were conducted with PEMs. The cumulative distribution functions (CDF) of exposure levels caused by uplink (UL) and downlink (DL) mobile communications, WLANs 2 G and 5 G, broadcast transmissions are presented and compared and main contributors in total RF-EMF levels in shopping malls, for different exposure scenarios, are identified. To summarize, in we have presented the maximum E-field value recorded in different measurement campaigns for different frequency bands and technology using similar measurement equipment’s with this study.
Table 1. Maximum E-field value recorded in different environments and locations.
To our knowledge there is a very limited number of works addressing the RF-EMF exposure patterns and trends in shopping centres and similar locations, therefore one of the main contributions of this work is to provide insight into what are the main contributors to the total RF-EMF exposure levels in such environments. In particular, shopping malls are categorized as vulnerable RF-EMF public environments, together with hospitals and schools. In addition, they exhibit unique properties in terms of propagation of RF waves since they are usually built from construction materials with specific electro-magnetic properties. They are also crowded environments, frequented by different age groups, with a presence of heterogeneous RF sources and a varying pattern of wireless UE usage (an asymmetry between UL and DL traffic compared to other sensitive environments). Another contribution of this work is the collection of a significant amount of measurement data which can be used to train machine learning models to obtain realistic approximation and characterizations of RF-EMF exposure patterns in heterogeneous indoor scenario. The data can be further used for a comparative analysis of RF-EMF exposure between different environments and for optimizing the trade-off between exposure and network performance.
The 5 G mobile services have been recently launched and will be part of future assessments as we look towards the optimization of exposure aware 5 G topologies for high density indoor environments such, as highlighted for some scenarios in (Castellanos et al., Citation2022).
Materials and methods
The human-centric frequency selective measurement in sub 6 GHz range are conducted in most popular, frequented by all ages, shopping mall in country, that has 100,000 sqm and 4 floors. Measurements are conducted by qualified junior researchers, in 3 months period, in different time slots of the days. In total our measurement campaign collected: 529,340 RF-EMF samples for working days and 430,020 RF-EMF samples for weekends. Measurements are taken every 5 s with EME SPY 200 equipment, three-axial E-field probe, that measures E-field V/m, Power density (S) in W/m2, and calculates total exposures based on recorded samples in 20 predefined frequency bands in the range 88 MHz–5850 MHz. The measurement protocol differentiates UL and DL for mobile communications. The upper detection limit of measurement instrumentation is 6 V/m while the sensitivity varies for different bands and is 0.005 V/m for most of frequency bands under investigation, excluding WiFi5G with the sensitivity 0.01 V/m. The measurement instruments is portable and during the measurement period is handheld at 1.5 m above the ground level. The measurements were collected in many different spots (>50) all over the shopping mall premises, so that measurements could reflect values for various shopping mall scenarios.
For shopping mall under the study, the E-field samples are collected for: FM (87–107 MHz); TV3 (174–223 MHz); TETRA I (380–400 MHz); TETRA II (410–430 MHz); TETRA III (450–470 MHz); TV4&5 (470–770 MHz); LTE 800 DL (791–821 MHz); LTE 800 UL (832–862 MHz); GSM + UMTS 900 UL (880–915 MHz); GSM + UMTS 900 UL (925–960 MHz); GSM 1800 UL (1710–1785 MHz); GSM 1800 DL (1805–1880 MHz); DECT (1880–1900 MHz); UMTS 2100 UL (1920–1980 MHz); UMTS 2100 DL (2110–2170 MHz); Wi-Fi 2 G (2400–2483.5 MHz); LTE 2600 UL (2500–2570 MHz); LTE 2600 DL (2620–2690 MHz); WiMAX (3300–3900 MHz) and Wi-Fi 5 G (5150–5850 MHz).
Currently, all broadcast and mobile communication providers, in the area under the study, operate in sub 3 GHz range except for recently launched 5 G mobile services which operate also in 3400 MHz band. WLAN networks operate in 2.4 and 5 GHz range, therefore all identified RF sources in shopping mall operate in sub 6 GHz band.
Not all technologies were present in shopping mall, therefore the samples were frequency matched. Technologies for which > 95% of samples were not detected were removed from further analysis. Even though WiMax does not operate in region under the study, 5 G mobile communications trial network is recently launched in this band. For each band when the signal is not detected the PEM assigns the value of below threshold, that could produce overestimation of total exposure, therefore different methods are used to post process the measurement data.
The measurements were conducted in a shopping center consisting of four floors where various type of commercial activities such as shops, cafeterias, and playgrounds for children are spread out. Most of the shops are positioned in a circular fashion along the main halls, characterized by a gallery opening in the centre from the ground floor to the uppermost floor. The food corner and the children’s playgrounds are mainly located in the top floor, while cafeterias can be found in each floor. Measurements were collected in different spots at different floors, at different times of the day, and different days of the week. During the measurement period there were no special promotion events or other activities organized that would urge for instantaneous enhanced need for data traffic. Measured values were screened to avoid potential error samples influencing measurement dataset.
The measurement data were transferred to computers and EME Spy software was used to initially analyse the results. The in-house build MATLAB script was used to derive CDFs (Cumulative Distribution Function) for various technologies operating in shopping mall and to compare and contrast results.
Since this study includes a large dataset of measurement samples, the statistical analysis of the collected data is performed for each sub 6 GHz technology separately. The comparative results are presented in next section.
Results
In this section, we present the comparative analysis based on human centric frequency selective measurement of E-field in the frequency range of 88 MHz to 6 GHz at shopping malls on different days, including weekdays and weekends. To have a better understanding on the impact of each technology and their contribution on the total exposure during different days of the week, the Results section is divided into four sub sections.
Comparison of personal exposure in shopping malls during weekdays and the weekend
To more accurately portray the personal exposure levels that individuals encounter at the shopping mall, the technologies where detection was not observed in over 95% of the samples, were omitted from the analysis.
Personal exposure to RF-EMF during weekdays and weekends for different technologies is presented in , which showcases the measured levels of RF-EMF in relation to the specific technologies.
Figure 1. Distribution of E-field exposure for different technologies during the week days (top) and weekend (bottom).
From the figure, it is clear that the dominant technologies observed during weekdays are mobile communications operating at 900 MHz and 1800 MHz. However, notable spikes in the graph indicate that the highest E-field values, including those at 6 V/m (limited by the dosimeter’s upper detection threshold, potentially concealing even higher values), were recorded for Uplink 1800 and Wi-Fi at 5 GHz technologies. Similarly, during the weekend the highest E-field values are recorded for Wi-Fi 5 G technology (5 V/m) and Uplink 900 technology (4.3 V/m).
presents the CDF of power density for all technologies during week days and weekend measurements,
Figure 2. CDF of power density during the week and weekend.
From the figure we can see that mobile communications (2G–4G), specifically Downlink 900 2G–3G, and Downlink 1800 2G–4G, dominate in terms of EMF exposure. Even though the all technologies have the same mean values, the highest measured values are for Downlink 2G–3G during the weekdays.
For better clarification a comparative statistical analysis of personal exposure for weekdays and weekend is shown in and .
Figure 3. CDF of the total power density during the weekdays and weekend.
Figure 4. Box and whiskers plot for the total E-field during weekdays and weekend.
From , we can see that the largest values are taken during the weekdays even though the mean values are the same in both cases. From the CDF we can conclude that 80% of the measurements during the weekday are recorded with power density less than 1 mW/m2, while 90% of the measurements during the weekend are recorded with the power density less than 1 mW/m2.
In the other hand, a Box and whiskers plot displaying the total E-field values categorized as weekdays and weekends is shown in . The mean value for both weekdays and weekends are depicted as 0.33 V/m. This indicates that, on average, the data points from both datasets tend to cluster around this value. During the weekdays, 50% of the measured electric fields values center from 0.09 V/m to 0.37 V/m, while on the weekends, 50% of the measured values range from 0.11 V/m to 0.48 V/m. Higher E field exposure values were measured during weekdays, with a maximum value of 6.02 V/m, compared to 4.98 V/m on weekends. Additionally, measurements during the weekdays had a higher outlier percentage of 6.7% (depicted in red), whereas during the weekends had only 1.1%.
Comparison of personal exposure from mobile communication technologies
A comparative analysis of the CDF for both UL and DL mobile communications (2 G–4 G) as the dominating technologies during the measurements is presented in the following figure. presents the comparison of 900 2G–3G bands. It can be seen that the higher values are captured for the DL where 80% of the values are less than 0.2 mW/m2, while 90% of the Uplink values are less than 0.05 mW/m2.
Figure 5. CDF of the Power Density of Uplink and Downlink Mobile Communications a. 2G–3G Technology, b. 2G–4G technology.
In , the CDF of 1800 2G–4G bands comparison is presented. It is clear that 90% of the measured values are below 0.3 mW/m2 for the Downlink, while 95% of the Uplink measurements are below 0.06 mW/m2. Interesting point is that the mean values for the 1800 2G–4G bands are the same, compared to 900 2G–3G where there is a significant difference.
We continue with graphical representation of total E-field values for the same above presented technologies.
From , it can be concluded that 50% of the data for both Uplink and Downlink mobile communications is clustered around the same average values smaller than 0.2 V/m during the Weekdays and smaller than 0.3 V/m during the Weekends.
Figure 6. Box and whiskers plot for the E-field of Uplink and Downlink mobile communications a. During the weekday b. During the weekend.
During the weekdays, a higher percentage of 4.9% of outliers is measured for the frequency Uplink 900 2G–3G, Downlink 1800 2G–4G with 3.3% and lastly Downlink 2G–3G and Uplink 2G–4G with 3% and 2% respectively. The same is observed during the weekends, where the highest percentage of outliers is recorded for the Uplink 900 2G–3G band with 5%, followed by Uplink 1800 2G–4G band with 1.8%, while the other two bands were consistent with smaller than 1% of outliers.
From these findings we can conclude that uplink and downlink communications exhibit different patterns of EMF exposure. In downlink the E-field values exhibit a higher mean and consistent values around the mean value, while in uplink the EMF exposure exhibits sporadic spikes but overall significantly lower mean values.
Comparison of personal exposure from WI-FI technologies
Other technologies that impact the total exposure and are crucial to be investigated are Wi-Fi at 2 and 5 GHz, denoted as WiFi2G and WiFi5G, respectively. To compare the two Wi-Fi technologies, we present the CDF of the Power Density and the statistical analysis of the data via a box and whiskers plot.
From the we can see that both Wi-Fi 2G and 5G have very similar distribution, same mean values, and 85% of the data measured is less than 0.01 mW/m2.
Figure 7. CDF of the power density for WI-FI technologies.
In , The E-field values of WI-FI technologies, 2G and 5G are presented. From the ) we can conclude that during the weekdays, higher exposure values are measured for the Wi-Fi 2G band, reaching a maximum value exceeding the upper detection limit of 6 V/m and 50% of the data clustering around 0.07 V/m, while for the WI-FI 5G band, the highest value was 2.87 V/m, and 50% of the data clustered around 0.03 V/m.
Figure 8. Box and whiskers plot of WI-FI technologies. (a) During the weekdays (b) During the weekend.
During the weekend, from , we can see that 50% of the data are recorded between 0.2 V/m and 0.7 V/m, but the outlier percentage is still higher for WI-FI 2G with a percentage of 6.2% compared to 3.9% for WI-FI 5G. Contrary to the measurements during the weekdays, the highest value was measured for WI-FI 5G with a maximum value of 4.97 V/m during the weekend.
Contribution of technologies to the total personal exposure
In addition to comparing the power density values of nine different technologies that had the highest impact on total exposure levels, we also calculated the overall exposure based on the recorded samples of the selected frequency bands.
From , we can see that the dominant technology to personal exposure is Downlink 900 2G–3G. 90% of the measured exposure values are below 0.6 mW/m2. The second biggest contributing factor to personal exposure is the Downlink 1800 2G–4G band, where 90% of the data is less than 0.4 mW/m2.
Figure 9. CDF of the power density for different technologies at the shopping mall.
From the measured data and . Downlink 900 2G–3G contributed >50% to the total exposure in shopping malls during the weekday. The same band is also the highest contributor during the weekend with 37.4%. While during the weekday Downlink 1800 2G–4G and Uplink 1800 2G–4G contributed both with 15.7% to the total exposure, in the weekend we saw a rise of Downlink 1800 2G–4G to 34.4% of the total exposure. WI-FI 2G contribution percentages were consistent during the weekdays and weekend with 10% and 8.5% respectively. Interesting point was that WI-FI 5G exposure during the weekend was 2.5 times higher than during the weekdays with 7.1%. It should be noted however that the EMF measured values for all scenarios under study, are well below the EMF safety exposure standards (Lin, Citation2023) for both general population and workers.
Figure 10. Wireless technology contribution on power density per day of the week.
In order to gain a better understanding of the data, make inferences, and forecast exposure levels we analyzed the data distribution. Our analysis included both the total exposure value and exposure values associated with each different technology. However, none of the data conformed to any distribution model. The lack of fit to any distribution model can be attributed to several factors, including the heterogeneity of the data, the presence of outliers, and other contributing factors. As the data is measured, it demonstrates non-linear relationships that cannot be adequately captured by standard distribution models.
A summary of statistical properties for the recorded measured values for the different technologies, and for different days of the week (weekday vs. weekend) are shown in .
Table 2. Statistical properties for all measurements.
Table 3. Statistical properties for weekday measurements.
Table 4. Statistical properties for weekend measurements.
Conclusions
This paper presents the exposure levels in shopping malls during different days of the week, obtained through a rigorous 3-month measurement campaign utilizing frequency selective measurement equipment. The collected data underwent careful processing, including screening of each measured E-field exposure level to ensure accurate sampling and minimize errors.
The average total E-field exposure value during the weekday was 0.39 V/m, while during the weekend 0.33 V/m. The highest contributors to the total exposure levels in various shopping mall environments are mobile communication technologies operating at 900 MHz and 1800 MHz (2G–3G–4G). The Downlink 2G–3G operating at 900 MHz, has an average electric field exposure value of 0.2 V/m during the weekdays and during the weekend, while the Downlink 2G–4G operating at 1800 MHz has an average E-field value of 0.1 V/m during the weekdays and 0.17 V/m during the weekend.
On weekdays, the main factor contributing to total exposure levels was 2G–3G mobile communications operating at 900 MHz (Downlink), accounting for 52.4% of the exposure, followed by mobile communications at 1800 MHz (Uplink and Downlink. Furthermore, the highest E-field values during weekdays were observed in the Downlink 2G–3G band and Wi-Fi 2 G technology, exceeding the upper limit of 6 V/m.
During the weekend, the primary contributor to total exposure levels remained the 2G–3G communications at 900 MHz (Downlink), accounting for 37% of the exposure. However, there was an increase in e-field values for 2G–4G communications at 1800 MHz (Downlink), with its contribution rising to 34% compared to 15% during weekdays. The highest observed value was for Wi-Fi 5G technology, reaching 4.98 V/m, although its contribution to the total exposure levels was only 0.6%.
Based on the cumulative distribution analysis, it can be inferred that during the weekdays, 90% of the measured data have an E-field exposure value of less than 1 V/m. On the other hand, during the weekend, 90% of the measured data have a value smaller than 0.64 V/m. These values are in line with the E-field values recorded in similar studies as referenced in .
In conclusion, based on our extensive measurement campaign, personal exposure levels are higher in shopping malls during the weekdays compared to the weekend. The highest contributors to the total personal exposure at shopping malls are the Mobile communication technologies (Uplink/Downlink) and WI-FI technologies.
Contrary to our initial expectations regarding the indoor environment of the shopping mall, our measurement data reveals that Wi-Fi technologies are not the primary contributors to the total exposure. Instead, mobile communication technologies operating at 900 MHz and 1800 MHz have emerged as the dominant sources due to their high data rates and extensive coverage.
Acknowledgment
This work was supported by the Ministry of Education, Science and Technology of Republic of Kosovo through the annual scientific research project grant scheme.
Data availability statement
The data that support the findings of this study are available from the corresponding author, V.R.H, upon reasonable request.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Additional information
Notes on contributors
Doruntinë Berisha
Doruntinë Berisha is a PhD student at the Faculty of Electrical and Computer Engineering, University of Prishtina, Kosovo. Her research interests include communication networks, EMF aware wireless communications and multi-objective optimization of wireless network performance.
Valdete Rexhëbeqaj-Hamiti
Valdete Rexhëbeqaj Hamiti is an Associate Professor with the Faculty of Electrical and Computer Engineering, University of Prishtina, Kosovo. Her research interests include operator theory and also applied mathematics with a focus on electrical engineering problems.
Jeta Dobruna
Jeta Dobruna is a Teaching Assistant with the Faculty of Electrical and Computer Engineering, University of Prishtina, Kosovo and a PhD student at the Faculty of Electrical Engineering, University of Ljubljana, Slovenia. Her research interests include wireless networks, resource allocation in 5G/6G networks.
Hëna Maloku
Hëna Maloku is an Assistant Professor with the Faculty of Electrical and Computer Engineering, University of Prishtina, Kosovo. Her research interests include heterogeneous wireless networks, mobile communications and cognitive radio.
Zana Limani Fazliu
Zana Limani Fazliu is an Assistant Professor with the Faculty of Electrical and Computer Engineering, University of Prishtina, Kosovo. Her research interests include wireless networks, ML techniques for networking.
Mimoza Ibrani
Mimoza Ibrani is a Professor with the Faculty of Electrical and Computer Engineering, University of Prishtina, Kosovo. Her research interests include applied electromagnetics, EMF exposure, wireless networks and multi-objective optimization of network performance.
References
- Bhatt, C. R., Redmayne, M., Abramson, M. J., & Benke, G. (2016). Instruments to assess and measure personal and environmental radiofrequency-electromagnetic field exposures. Australasian Physical & Engineering Sciences in Medicine, 39(1), 29–42. https://doi.org/10.1007/s13246-015-0412-z
- Breckenkamp, J., Blettner, M., Schüz, J., Bornkessel, C., Schmiedel, S., Schlehofer, B., & Berg-Beckhoff, G. (2012). Residential characteristics and radiofrequency electromagnetic field exposures from bedroom measurements in Germany. Radiation and Environmental Biophysics, 51(1), 85–92. https://doi.org/10.1007/s00411-011-0389-2
- Castellanos, G., De Gheselle, S., Martens, L., Kuster, N., Joseph, W., Deruyck, M., & Kuehn, S. (2022). Multi-objective optimisation of human exposure for various 5G network topologies in Switzerland. Computer Networks, 216, 109255. https://doi.org/10.1016/j.comnet.2022.109255
- Celaya-Echarri, M., Azpilicueta, L., Ramos, V., Lopez-Iturri, P., & Falcone, F. (2021). Empirical and modeling approach for environmental indoor RF-EMF assessment in complex high-node density scenarios: Public shopping malls case study. IEEE Access. 9, 46755–46775. https://doi.org/10.1109/ACCESS.2021.3067852
- Chen, L., Elzanaty, A., Kishk, M. A., Chiaraviglio, L., & Alouini, M. S. (2023). Joint uplink and downlink EMF exposure: Performance analysis and design insights. IEEE Transactions on Wireless Communications, 22(10), 6474–6488. https://doi.org/10.1109/TWC.2023.3244155
- Chikha, W. B., Wang, S., & Wiart, J. (2023). An extrapolation approach for RF-EMF exposure prediction in an urban area using artificial neural network. IEEE Access, 11, 52686–52694. https://doi.org/10.1109/ACCESS.2023.3280125
- Cuiñas, I., Laska-Leśniewicz, A., Znajdek, K., & Kamińska, D. (2023). Exploring the application of design thinking methodology in cellular communications network planning and deployment. IEEE Transactions on Technology and Society, 4(3), 269–278. https://doi.org/10.1109/TTS.2023.3239261
- da LA Silva, J., de Sousa, V. A., Rodrigues, M. E. C., Pinheiro, F. S. R., da Silva, G. S., Mendonça, H. B., de FH. Silva, R. Q., da Silva, J. V. L., Galdino, F. E. S., de Carvalho, V. F. C., & Medeiros, L. I. C. (2023). Human exposure to non-ionizing radiation from indoor distributed antenna system: Shopping mall measurement analysis. Sensors, 23(10), 4579. https://doi.org/10.3390/s23104579
- Engiz, B. K., & Kurnaz, C. (2017). Long-term electromagnetic field measurement and assessment for a shopping mall. Radiation Protection Dosimetry, 175(3), 321–329. https://doi.org/10.1093/rpd/ncw343
- Ibrani, M., Halili, R., Ahma, L., Hamiti, E., Fejza, N., & Berisha, D. (2018). Experimental assessment of wi-fi signal levels in indoor environments. 2018 18th Mediterranean Microwave Symposium (MMS) (pp. 307–312). IEEE. https://doi.org/10.1109/MMS.2018.8611960
- Jiang, Y., Wang, H., Sun, X., Li, C., & Wu, T. (2023). Evaluation of Chinese populational exposure to environmental electromagnetic field based on stochastic dosimetry and parametric human modelling. Environmental Science and Pollution Research International, 30(14), 40445–40460. https://doi.org/10.1007/s11356-023-25153-y
- Karpowicz, J., de Miguel-Bilbao, S., Zradzńnski, P., Gryz, K., Falcone, F., & Ramos, V. (2018). Comparative study of radiofrequency electromagnetic exposure in the public shopping centers. 2018 International Symposium on Electromagnetic Compatibility (EMC Europe) (pp. 972–975). IEEE. https://doi.org/10.1109/EMCEurope.2018.8485113
- Kim, B. C., & Park, S.-O. (2010). Evaluation of RF electromagnetic field exposure levels from cellular base stations in Korea. Bioelectromagnetics, 31(6), 495–498. https://doi.org/10.1002/bem.20585
- Lin, J. C. (2023). RF health safety limits and recommendations [Health Matters]. IEEE Microwave Magazine, 24(6), 18–77. https://doi.org/10.1109/MMM.2023.3255659
- Malandrino, F., Chiaramello, E., Parazzini, M., & Chiasserini, C. F. (2022). Performance and EMF exposure trade-offs in human-centric cell-free networks. 2022 20th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt) (pp. 377–382). IEEE. https://doi.org/10.23919/WiOpt56218.2022.9930537
- Mulugeta, B. A., Wang, S., Ben Chikha, W., Liu, J., Roblin, C., & Wiart, J. (2023). Statistical characterization and modeling of indoor RF-EMF down-link exposure. Sensors, 23(7), 3583. https://doi.org/10.3390/s23073583
- Onishi, T., Ikuyo, M., Tobita, K., Liu, S., Taki, M., & Watanabe, S. (2021). Radiofrequency exposure levels from mobile phone base stations in outdoor environments and an underground shopping mall in Japan. International Journal of Environmental Research and Public Health, 18(15), 8068. https://doi.org/10.3390/ijerph18158068
- Paniagua-Sánchez, J. M., García-Cobos, F. J., Rufo-Pérez, M., & Jiménez-Barco, A. (2023). Large-area mobile measurement of outdoor exposure to radio frequencies. The Science of the Total Environment, 877, 162852. https://doi.org/10.1016/j.scitotenv.2023.162852
- Tognola, G., Plets, D., Chiaramello, E., Gallucci, S., Bonato, M., Fiocchi, S., Parazzini, M., Martens, L., Joseph, W., & Ravazzani, P. (2021). Machine learning for the estimation of Wi-Fi field exposure in complex indoor multi-source scenario. 2021 XXXIVth General Assembly and Scientific Symposium of the International Union of Radio Science (URSI GASS) (pp. 1–4). IEEE. https://doi.org/10.23919/URSIGASS51995.2021.9560437
- Yang, L., Lu, M., Lin, J., Li, C., Zhang, C., Lai, Z., & Wu, T. Long-term monitoring of extremely low frequency magnetic fields in electric vehicles. International Journal of Environmental Research and Public Health, 2019, 16, 3765. https://doi.org/10.3390/ijerph16193765