Assessment of varied mobile network topologies on human exposure, mobile communication quality and sustainability

Abstract

Assessment of varied mobile network topologies on human exposure, mobile communication quality and sustainability Margot Deruyck, German Castellanos, Wout Joseph, Luc Martens, Sven Kuehn, Niels Kuster. Assessment of varied mobile network topologies on human exposure, mobile communication quality and sustainability. Final Report of Project CRR-954. Zurich, Switzerland, IT'IS Foundation. Sep 21, 2021. Executive Summary In October 2020, the Swiss Federal Office of Communications (OFCOM) mandated the IT’IS Foundation to evaluate various 5G network topologies regarding human exposure, mobile communication quality, and sustainability to address the questions posed by the political Motion Häberli-Koller (19.4043) dated July 30, 2020. The study was conducted jointly with the IMEC WAVES group of the Department of Information Technology of Ghent University, Belgium, and it supplements an earlier project performed by IT’IS for the Swiss Federal Office for the environment in 2019 [1]. Statement of work. Prior to the study, OFCOM specified a number of endpoints in the statement of work (SoW), each of which is addressed below. The study was conducted using a mobile network planning tool developed by the IMEC WAVES group that was adapted to the specifics of the mobile network and regulatory situation in Switzerland. Using this tool, we simulated a variety of mobile networks to address the study endpoints specified by OFCOM. The simulations included 4G and 5G networks in rural, suburban, and urban environments with usage requirements extrapolated to the year 2030. We analyzed the effects of separate operators compared to a unified network, the separation of indoor and outdoor networks, and different data rates and networks optimized for low downlink exposure. Human exposure to the electromagnetic field from the mobile communication system is expressed as the exposure ratio, the induced 6-minute time-averaged electromagnetic field level divided by the safety limits in percent, separately for the uplink (exposure to a user’s own mobile device) and downlink (exposure to the base station network). All mobile networks were realized to comply with an approximation to the current precautionary limits imposed by the Swiss Ordinance for Protection against Non-Ionizing Radiation, i.e., the effect of an increase of the precautionary limits was not studied. Note that for all of the following statements, in general, the user’s own mobile device (uplink) exposure ratio contributed to the user’s total exposure on average with a minimum tenfold higher level than the downlink (base station) exposure. Therefore, for any active mobile device user, a reduction of the downlink exposure will always remain insignificant in terms of the overall exposure. Which network structures are possible based on the technologies available today (4G, 5G, Wireless Local Area Network WLAN, etc.), and what influence do they have on the spatial distribution of radiation exposure of the population? In the present study, network infrastructures based on the 4G and 5G communication technologies were simulated. The use of WLAN as a supplementary link for indoor reception was not considered in the simulation models as 4G and 5G provide better spectral efficiency technology and output power control than WLAN. Therefore, the use of WLAN in terms of network quality and exposure reduction is not considered beneficial. Other reasons such as network separation and costs may favor the use of WLAN. Our results show that the transition from 4G to 5G will reduce human exposure in most simulated scenarios while offering a tenfold capacity. A unified mobile network results in downlink exposure ratios similar to that of the largest user base in Switzerland; however, the unified network can serve twice as many users. Network unification would improve the uplink exposure compared to the uplink exposure in the networks of the second and third largest user base in Switzerland. A unified network would also reduce the number of required base station locations between - 13% and -50% (depending on technology and environment). In particular, the frequently propagated approaches of the "St. Gallen model", the separation of indoor and outdoor coverage are to be analyzed in depth. We analyzed the coverage of indoor and outdoor locations as well as indoor and outdoor exposure. Our results show that a complete separation of indoor and outdoor coverage will lead to lower (factor o 4) downlink exposure outdoors and uncovered indoor locations on average. Indoor downlink exposure is not affected by the separation of indoor and outdoor coverage. Uplink exposure remained in the same range for separate indoor and outdoor networks despite additional building attenuation. We found a trend towards lower uplink exposure (factor of 4) of the 5G technology compared to 4G. This effect is likely related to the use of the MaMIMO (Massive Multiple Input Multiple Output) capabilities in 5G. In addition, the advantages and disadvantages of using adaptive antennas, also with regard to the data rate and spatial distribution of radiation exposure, should be shown. The 5G network employing adaptive antenna systems is well suited to reduce human exposure while increasing the network capacity by a factor of 10. Especially, in less densely populated environments, adaptive antennas improve the exposure ratio. In the rural environment, the exposure ratio is reduced by a factor of two, while in the urban and suburban environments the average exposure is not affected by the use of adaptive antennas. We found a clear trend towards lower uplink and downlink exposure ratios for the 5G networks using adaptive antennas compared to 4G networks not using adaptive antenna technology. What influence do the various network structures have on the quality of mobile communication coverage in Switzerland? All networks in the study were planned with a user coverage of at least 95% such that a lack of coverage was compensated by additional base station sites. Our results show that coverage with 5G speeds compared to 4G speeds requires on average three times as many base stations. The coverage of only outdoor locations would primarily reduce the need for base stations in urban environments and only for 4G. 5G will result in a base station densification for outdoor coverage that is already suited for indoor coverage in many locations. Where insufficient, e.g., in large buildings, the full coverage can be obtained by supplementary 5G indoor base stations. What influence do the various network structures have on the expansion of mobile communications networks in Switzerland (number of antenna systems)? The transition from a 4G to a 5G network with a tenfold data bandwidth requires, on average, three times more base stations in our network optimization simulations. Base station count is increased by 60% (4G) and 14% (5G) if indoor locations are also covered by the mobile network in the urban environment. The reduction decreases to 20% (4G) and 6% (5G) for the suburban environment and vanishes in the rural environment. Extending the data rate capacity by an additional factor of ten for 5% of the users resulted in only a few additional base station locations but in higher human exposure. The study also demonstrated that future 5G networks can be realized without an increase of precautionary limits. The number of base stations is mostly driven by the data requirements and not by the exposure limits. What influence does the number of mobile communication networks have on the radiation exposure of the population or would a single network lead to less radiation exposure than three separate networks? Our results show that a unification of the mobile network infrastructure does not change human exposure considerably compared to multiple operators. However, a unification of the network infrastructure could lead to a smaller number of required base station sites (-13% to -50% depending on the environment and communication system). Due to the user limitation per base station for 5G adaptive antennas, the possible site reduction is greater for 4G (-30% to -50%) than for 5G (-13% to -30%). What does an ideal network structure look like in order to minimize radiation exposure for the population and at the same time ensure a good quality of cell phone coverage? To minimize human exposure to electromagnetic fields, network planning should always take into account both uplink and downlink exposure. For active users, the uplink exposure is tenfold higher than the downlink exposure. Our results also show that consideration of downlink-only exposure in the planning stage will not reduce the exposure for non-users (members of the society without their own mobile device). A network for minimizing the combined uplink and downlink exposure employs 5G technology and has a dense base station infrastructure, supplemented locally, such as in large buildings, by indoor base stations. Another important outcome is that the 5G base station density is mainly driven by the coverage requirements, i.e., lower base station count under higher limits is not expected. The same rationale leads to the conclusion that relaxed precautionary limits likely increase the uplink exposure, i.e., the overall exposure. Although the frequencies above 6 GHz are not yet available in Switzerland, the study should also include the future use of these frequencies (millimeter waves, in Switzerland probably in the 24.25-27.5 GHz range). At present, the use of millimeter-wave technologies in Switzerland cannot be predicted. Therefore, we did not include millimeter waves in our mobile network simulation model. Based on initial roll-outs internationally, the current main application of 5G millimeter links is for the last mile. The last mile application (fixed wireless access) is not strictly limited to mobile communication. Here, wireless point-to-point links would replace copper or fiber links. Based on the point-to-point nature, human exposure to last-mile links is unlikely. Recently, the first mobile devices (US models of Apple iPhone 12, Samsung S21) with millimeter-wave communication capabilities were placed on the market. This development is driven by the unavailability of the 3.5 GHz band in the United States (US) and the attempt to offload the majority of the data volume over millimeter waves. There is still very little use of millimeter-wave communications in the US [2]. Due to shadowing effects of the human body [3] and the highly directive beam-forming in this frequency range, it can be assumed that the usage of millimeter waves for mobile applications will also mostly be limited to line-of-sight situations. This may lower human exposure due to the highly directive and adaptive nature of the communication links required for signal quality reasons. Study limitations. Even though the study includes the most currently advanced simulations on user exposure as a function of network topology, several assumptions were made due to the limitations of the tools and to missing information, which are described in detail in the Methods Section (Section 4). The impact of these assumptions on the results is discussed in detail in Section 6 of this report. Remaining Knowledge Gaps. In the course of this study, we identified future work and research needs to fill the remaining knowledge gaps. As soon as a possible application of millimeter waves emerges in Switzerland, its influence on human exposure should be analyzed. In the present study, we applied harmonized, yet simplified models to analyze indoor exposure to mobile networks. To lower the uncertainties on exposure in indoor scenarios, additional indoor modeling would help to substantiate our results. They could further be strengthened by validation measurements in the up- and downlinks of real 4G and 5G networks in Switzerland. Other areas of future research include the extension of the networks with distributed MaMIMO, mixed technology networks as well as more realistic assignment and weighting of user and base station locations. Excerpt To the knowledge of the authors, this study is currently the most advanced study on user exposure as a function of network topology. In order to perform the study using the developed tools and available information, several assumptions were made and described in detail in the Methods Section (Section 4). The impact of these assumptions on the results are as follows: • The use of simplified models for the propagation and environments (buildings, etc.) may not exactly represent the propagation environment present in the real mobile network environments. • The chosen coverage goal of 95% of the users in all the study environments at any user location might be higher in some environments than in real networks. • The selected configurations for the modeling of mobile networks may be different in the actual network implementation. • A global 5 V/m downlink exposure limit and a 6 minute average of the exposure were applied to map the regulatory boundaries. This means that not all details of the current regulatory requirement with respect to mobile communication exposure could be modeled 1:1 in the simulation tool. • Even though the precautionary limits were only approximated, the results imply that the networks can be realized under the current regulation. • The results were based on average usage, yet due to the dominating effect of the uplink exposure, exposure ratios are strongly dependent on personal mobile device usage which was not studied in detail here. • Base station locations in the extended set were assigned by the optimization algorithm irrespective of the actual building feasibility. • Currently, there is relatively limited knowledge about the actual deployment and development of 5G in Switzerland. • Only frequency ranges currently licensed for mobile communication use in Switzerland were included, i.e., millimeter-wave communication links (5G NR FR2) were neglected in the network planning. • The statistical-based modeling of exposure and limited environmental details and resolution statements on the absolute levels of exposure have a higher level of uncertainty than relative comparisons between scenarios. • Other sources of exposure than from the simulated mobile communication networks were not included. • It was assumed that other mobile communication services, e.g., 2G, 3G, were not present in the analyzed scenarios. • The currently discussed relaxation of the precautionary limits was not studied in detail; however, the simulations indicate that the 5G base station density is mainly driven by the coverage requirements, i.e., lower base station count under higher limits is not expected. Furthermore, [1] relaxed limits likely increase the uplink exposure, i.e., the overall exposure. Open access report: bit.ly

AI evidence extraction

Main findings

Using a Swiss-adapted mobile network planning tool, simulations of 4G and 5G topologies (rural/suburban/urban; 2030 usage) found that uplink (user device) exposure ratios were on average at least tenfold higher than downlink (base station) exposure ratios. The transition from 4G to 5G was reported to reduce human exposure in most simulated scenarios while providing ~10× capacity; 5G with adaptive antennas (MaMIMO) showed trends toward lower uplink and downlink exposure ratios versus 4G, with rural downlink exposure ratio reduced ~2× by adaptive antennas (average exposure not affected in urban/suburban). Separating indoor/outdoor coverage reduced outdoor downlink exposure (~4×) but led to uncovered indoor locations on average; indoor downlink exposure was not affected and uplink exposure stayed in the same range. Unifying operators’ networks did not change exposure considerably but could reduce required base station sites (~13% to 50%, depending on environment/technology) and improve uplink exposure versus smaller-operator networks; achieving 5G speeds vs 4G speeds required ~3× as many base stations on average, and additional extreme capacity for 5% of users increased exposure. Millimeter-wave (24.25–27.5 GHz) was not modeled.

Outcomes measured

• Human EMF exposure ratio (uplink and downlink; % of safety limits)
• Spatial distribution of exposure
• Mobile communication quality/coverage (>=95% user coverage target)
• Network capacity/data rate requirements
• Number of required base station sites/antenna systems (densification)
• Sustainability (mentioned as endpoint; specific metrics not detailed in abstract)

Limitations

• Results are based on simulations with multiple assumptions; simplified propagation/environment (buildings) models may not match real networks
• Coverage goal fixed at 95% of users at any location may exceed real-network targets
• Modeled network configurations may differ from actual implementations
• Regulatory boundaries approximated using a global 5 V/m downlink limit and 6-minute averaging; not all regulatory details modeled 1:1
• Exposure results based on average usage; personal device usage patterns (dominant for total exposure) not studied in detail
• Optimized base station locations may ignore real-world feasibility constraints
• Limited knowledge about actual 5G deployment/development in Switzerland at time of study
• Only currently licensed Swiss mobile bands modeled; 5G NR FR2/mmWave neglected
• Absolute exposure levels more uncertain than relative comparisons between scenarios due to statistical modeling and limited environmental detail/resolution
• Other exposure sources beyond simulated mobile networks not included
• Assumed absence of other services (e.g., 2G/3G)
• Relaxation of precautionary limits not studied in detail (only indicated by simulations)

Suggested hubs

• 5g-policy (0.9)
  Commissioned by Swiss regulator/political motion to evaluate 5G topologies, exposure, and implications under precautionary limits.
• who-icnirp (0.25)
  Uses safety limits/exposure ratios and discusses regulatory limits, though no explicit WHO/ICNIRP evaluation is stated.

{
    "study_type": "engineering",
    "exposure": {
        "band": "RF",
        "source": "mobile network (4G/5G base stations and user devices; simulated topologies)",
        "frequency_mhz": null,
        "sar_wkg": null,
        "duration": "6-minute time-averaged field level (exposure ratio metric)"
    },
    "population": "Simulated users/population exposure in Switzerland (rural, suburban, urban environments; extrapolated usage requirements to 2030)",
    "sample_size": null,
    "outcomes": [
        "Human EMF exposure ratio (uplink and downlink; % of safety limits)",
        "Spatial distribution of exposure",
        "Mobile communication quality/coverage (>=95% user coverage target)",
        "Network capacity/data rate requirements",
        "Number of required base station sites/antenna systems (densification)",
        "Sustainability (mentioned as endpoint; specific metrics not detailed in abstract)"
    ],
    "main_findings": "Using a Swiss-adapted mobile network planning tool, simulations of 4G and 5G topologies (rural/suburban/urban; 2030 usage) found that uplink (user device) exposure ratios were on average at least tenfold higher than downlink (base station) exposure ratios. The transition from 4G to 5G was reported to reduce human exposure in most simulated scenarios while providing ~10× capacity; 5G with adaptive antennas (MaMIMO) showed trends toward lower uplink and downlink exposure ratios versus 4G, with rural downlink exposure ratio reduced ~2× by adaptive antennas (average exposure not affected in urban/suburban). Separating indoor/outdoor coverage reduced outdoor downlink exposure (~4×) but led to uncovered indoor locations on average; indoor downlink exposure was not affected and uplink exposure stayed in the same range. Unifying operators’ networks did not change exposure considerably but could reduce required base station sites (~13% to 50%, depending on environment/technology) and improve uplink exposure versus smaller-operator networks; achieving 5G speeds vs 4G speeds required ~3× as many base stations on average, and additional extreme capacity for 5% of users increased exposure. Millimeter-wave (24.25–27.5 GHz) was not modeled.",
    "effect_direction": "mixed",
    "limitations": [
        "Results are based on simulations with multiple assumptions; simplified propagation/environment (buildings) models may not match real networks",
        "Coverage goal fixed at 95% of users at any location may exceed real-network targets",
        "Modeled network configurations may differ from actual implementations",
        "Regulatory boundaries approximated using a global 5 V/m downlink limit and 6-minute averaging; not all regulatory details modeled 1:1",
        "Exposure results based on average usage; personal device usage patterns (dominant for total exposure) not studied in detail",
        "Optimized base station locations may ignore real-world feasibility constraints",
        "Limited knowledge about actual 5G deployment/development in Switzerland at time of study",
        "Only currently licensed Swiss mobile bands modeled; 5G NR FR2/mmWave neglected",
        "Absolute exposure levels more uncertain than relative comparisons between scenarios due to statistical modeling and limited environmental detail/resolution",
        "Other exposure sources beyond simulated mobile networks not included",
        "Assumed absence of other services (e.g., 2G/3G)",
        "Relaxation of precautionary limits not studied in detail (only indicated by simulations)"
    ],
    "evidence_strength": "low",
    "confidence": 0.7399999999999999911182158029987476766109466552734375,
    "peer_reviewed_likely": "no",
    "keywords": [
        "5G",
        "4G",
        "network topology",
        "exposure ratio",
        "uplink exposure",
        "downlink exposure",
        "adaptive antennas",
        "Massive MIMO",
        "base station densification",
        "Switzerland",
        "OFCOM",
        "simulation",
        "precautionary limits",
        "NIS ordinance"
    ],
    "suggested_hubs": [
        {
            "slug": "5g-policy",
            "weight": 0.90000000000000002220446049250313080847263336181640625,
            "reason": "Commissioned by Swiss regulator/political motion to evaluate 5G topologies, exposure, and implications under precautionary limits."
        },
        {
            "slug": "who-icnirp",
            "weight": 0.25,
            "reason": "Uses safety limits/exposure ratios and discusses regulatory limits, though no explicit WHO/ICNIRP evaluation is stated."
        }
    ]
}

Comments

Log in to comment.

No comments yet.