Good practice guide for Power Density exposure measurements of 5G new radio base stations, based on the D5 report, and a new validated measurement methodology for measuring the PD exposure levels of 5G NR base stations, with evidence of their submission to standards bodies, for example, CENELEC CLC/TC 106X, IEC TC106 MT3 and JWG12, ITU-T SG5, IEEE ICES TC95, IEC 62232, and IEC 62669 for their consideration as an input to standards or technical specifications

Confidentiality Status: PU - Public, fully open (remember to deposit public deliverables in a trusted repository) Deliverable Cover Sheet Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or EURAMET. Neither the European Union nor the granting authority can be held responsible for them. The project has received funding from the European Partnership on Metrology, co-financed from the European Union’s Horizon Europe Research and Innovation Programme and by the Participating States. 1 of 78 21NRM03 MEWS D6 Good practice guide for Power Density exposure measurements of 5G new radio base stations, based on the D5 report, and a new validated measurement methodology for measuring the PD exposure levels of 5G NR base stations, with evidence of their submission to standards bodies, for example, CENELEC CLC/TC 106X, IEC TC106 MT3 and JWG12, ITU-T SG5, IEEE ICES TC95, IEC 62232, and IEC 62669 for their consideration as an input to standards or technical specifications Organisation name of the lead participant for the deliverable: METAS Due date of the deliverable: 31/07/2025 Actual submission date of the deliverable: 31/07/2025 2 of 71 Glossary CORESET: Control Resource Set DM-RS: Demodulation Reference Signal OFDM: Orthogonal Frequency-Division Multiplexing NR: New Radio PBCH: Physical Broadcast Channel PDCCH: Physical Downlink Control Channel PDSCH: Physical Data Shared Channel PSS: Primary Synchronization Signal RB: Resource Blocks RE: Resource Element SCS: Subcarrier Spacing SNR: Signal-to-noise Ratio SSB: Synchronization Signal Block SSS: Secondary Synchronization Signal SS/PBCH: Synchronization Signal and Physical Broadcast Channel UE: User Equipment 21NRM03 MEWS 3 of 78 Table of Contents 1 Summary ................................................................................................................................. 4 2 Traceability of 5G Exposure Measurements ............................................................................. 5 3 Improved Measurement Method for Exposure from Traffic Beams ........................................... 6 4 Definitions of Measurement Protocol and Scenarios for RF Exposure from 5G NR Base Stations .................................................................................................................................... 8 4.1 Methodology ...................................................................................................................... 8 4.2 Measurement Parameters .................................................................................................. 8 4.3 Generation of Additional Traffic ........................................................................................ 13 4.4 Measurement Protocol ..................................................................................................... 13 5 Analysis Methods for Measurements with 5G NR RE Statistics ............................................. 16 5.1 Histogram of the Resource Element Power in a Frame .................................................... 16 5.2 Fading Problem ................................................................................................................ 17 6 Conversion of Measurement Results, Calculation of Appreciation Values and Measurement Uncertainties .......................................................................................................................... 20 6.1 Conversion from Power (dBm) to Power Density (W/m2) .................................................. 20 6.2 Measurement Uncertainty of Appreciation Values ............................................................ 23 7 Comparison of Different Traffic Protocols and Evaluation Methods ........................................ 24 7.1 Measurement Procedure .................................................................................................. 24 7.2 Measurement Results ...................................................................................................... 25 8 Measurements of RF Exposure for Different Times of Day ..................................................... 28 8.1 Measurement Procedure .................................................................................................. 28 8.2 Measurement Results of Scenario 1 (Open Air Exposure) ............................................... 29 8.3 Measurement Results of Scenario 2 (LOS Exposure) ...................................................... 32 8.4 Measurement Results of Scenario 3 (nLOS Exposure) .................................................... 35 9 Consistency Analysis of Measurements ................................................................................. 38 9.1 Frequency Selective vs Code Selective Measurements and Comparison of Appreciation Values ....................................................................................................................................... 38 9.2 Comparison of Traceable Code Selective SSS measurements and Newly Proposed Methods with 5G NR RE Statistics Option ................................................................................. 40 10 Summary of Important Findings ............................................................................................. 45 11 References ............................................................................................................................ 47 Annex A – Traceable Calibration of 5G NR Measuring Receivers ................................................. 48 Annex B – An Improved Measurement Method for Radio Frequency Exposure from 5G NR Base Stations .................................................................................................................................. 58 Annex C – Analysis of Different Threshold Values for Method 5 and Different Measurement Bandwidths ............................................................................................................................ 65 Annex D – Emails sent as Evidence of Submission to Standards Bodies ...................................... 72 21NRM03 MEWS 4 of 78 1 Summary This document describes the full study of a measurement campaign of power density exposure from a selected base station. In this study, the current established exposure measurement, which is based on the measurement of synchronization symbols such as secondary synchronization symbol (SSS), is brought one step beyond to include the quantification of the exposure from data traffic. The document describes first the traceability chain of such measurements based on D5 report (calibration of code selective 5G NR measuring receivers). Then, the considerations for the exposure measurements from 5G NR base stations are explained. The current methodology is dependent on the measurement of periodic, traffic-independent cell-specific synchronization signals. However, with the beam forming characteristics of base stations employing smart antennas would make the exposure from traffic beams crucial parameters to evaluate. For this purpose, an improved measurement method for the power density exposure from traffic signals is designed. The basics of this method, its implementation on a real product and the proof of the applicability of the method are presented in detail. The study is extended to the include the measurements of an operational base station where the proposed measurement method is applied. All the details on the selected exposure cases and the measurement protocol including additional traffic generation are explained in detail. For the calculation of the decisive values of the exposure, the required calculations and the conversions are given step by step. Since the new method for the measurement of exposure from traffic beams requires an adapted analysis, some possible methods are provided and their effectiveness are studied. As the next step, the repeatable traffic generation is studied in detail. The measurements of six different traffic protocols in different exposure cases and their adapted analysis are presented. Based on this study, the traffic protocol for further measurements is selected. As part of the study, the ground exposure levels for different cases (line-of-sight, non-line-of-sight and open air) and for different times of a day are measured. For the measurements, both code selective and frequency selective methods are applied. The provided traffic measurement method is applied to verify the applicability and the consistency. Last, the summary of the findings from this study is then provided to conclude the study. With the proposed methods, provided measurement details, analysis approaches, suggestions and observations, this document serves as a good practice guide for exposure measurements of the base stations. 21NRM03 MEWS 5 of 78 2 Traceability of 5G Exposure Measurements The exposure measurements from base stations are typically performed with the following equipment set: 1) Antenna 2) Cable 3) Measuring receiver A representative measurement case with such equipment in a room having a direct line-of-sight (LOS) is given in Fig. 1. Fig. 1. Typical Exposure Measurements from Base Stations For the exact determination of the electric field (E Field) or the power density, each of these equipment must be calibrated and the associated correction or conversion factors must be obtained. For antennas and cables, the determination of these factors is performed using established calibration methods, such as S parameter measurements or comparison with a reference antenna. For measuring receivers used to measure 5G signals, the ideal calibration method would be a codeselective method, where the cell-specific signals, such as SSS, are measured. Such 5G measuring receivers are capable of measuring the exposure for one resource element (RE) of SSS by decoding the synchronization symbol physical broadcast channel (SS/PBCH). For the traceable calibration of this functionality, a setup with proper decoding capabilities with traceable power measurements must be presented. For this purpose, a calibration method providing the traceability over a calibrated measurement setup and power meter was presented in D5 report together with the detailed measurement uncertainty budget. The most important aspects for this traceable calibration method are summarized in Annex A. 21NRM03 MEWS 6 of 78 3 Improved Measurement Method for Exposure from Traffic Beams The current methodology for exposure measurements is based on the measurement of the periodic, traffic-independent signals such as secondary synchronization signal (SSS). However, the beamforming capabilities of 5G NR antennas creates different exposure conditions for traffic and broadcast transmissions. The difference of traffic and broadcast beams lays in the periodicity, antenna gain and aperture. For this reason, the exposure can change depending on the signal type. Fig. 2. Antenna Diagram of a typical 5G Antenna (Blue curve for maximum SSS diagram, black curve for worst case traffic diagram) [1]. In the given direction (denoted with a black arrow), SSS gain is around 3 dB less than that of traffic pattern. Fig. 2 gives an example for the antenna diagrams from a 5G antenna. The blue curve gives the antenna pattern for maximum SSS transmissions where the black curve gives the worst-case traffic diagram. In the given direction (denoted with a black arrow), SSS antenna gain is almost 3 dB less than that of traffic transmission. Such a case would mean more exposure from traffic beams than broadcast beams. In the previous technologies such as 4G LTE or earlier, there was no difference between the transmission gains and the broadcast and the traffic beams had the same antenna patterns. Therefore, measuring the synchronization signals such as cell-specific reference signal (CRS) in LTE was enough to evaluate also for the exposure from the data traffic [2]. However, for 5G NR, this approach needs adaptation to compensate for the difference. Code-selective 5G measuring receivers can directly measure the exposure created by REs within SSS. On the other hand, the exposure created by REs within a traffic transmission requires an adapted method. For this purpose, a new statistical method to measure the exposure from traffic with a practical application is proposed. In the proposed method, the whole resource grid is decoded by the receiver and the exposure from resource elements are obtained. However, since the total number of REs would reach a value of 1 million, the receiver cannot list and record the individual contribution of REs in a practical measurement duration. In order to simplify the processing duty, in the proposed method, REs are grouped together according to their power and the measurement results is displayed in a power histogram. Power levels (denoted as power bins) can have step of 0.1 dB, 0.2 dB or 0.5 depending on the application type. By this way, the measurement and recording duration would be fast enough so that even in the worst case (i.e., highest bandwidth of 100 MHz), one measurement would take only up to 4 seconds. 21NRM03 MEWS 7 of 78 A sample output of such measurement is display in Fig. 3. Fig. 3. Sample histogram output. System noise, traffic and SSS peaks are shown. As it can be seen from Fig. 3, different peaks in the histogram depict different signal types. The largest peak belongs to the system noise, which can be discarded for the final evaluation. On the other hand, two other peaks belong to traffic and synchronization signals. They form almost a normal distribution, having different variances and peak values. The location of these peaks on the x-axis (i.e., power axis) would point the power levels of these signals. By this way, the exposure from SSS and traffic signals can be determined. The validation of this new methodology and its application for different case of 5G transmissions are presented in the publication given in [3]. In Annex B, some important aspects of this publication are summarized. It is worth to emphasize that this publication also resulted from the research performed in the scope of MEWS project. This improved measurement method was used in the measurement campaign for RF exposure from a base station provided in this document. The exposure from traffic beams for different cases of exposure including the daily change and the effects of different traffic protocols are investigated using this new methodology. The details are given in the following chapters. 0 200 400 600 800 1000 1200 1400 1600 -150-140-130-120-110-100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 Number of Resource Elements () Power (dBm) Sample Histogram Noise Traffic SSS 21NRM03 MEWS 8 of 78 4 Definitions of Measurement Protocol and Scenarios for RF Exposure from 5G NR Base Stations 4.1 Methodology The methodology to quantify the RF exposure including the traffic beams is based on the measurement of an operational 5G NR base station in different conditions using adapted equipment and the proposed method given in Section 3. For this purpose, one commercial base station was selected. The measurement locations were selected in order to depict different cases of RF exposure. Different type of exposures, namely, the direct exposure or indirect exposure causing more diffused radiation were taken into account. The base station was measured in a normal operating mode, i.e., no special changes or conditions were applied on the base station. Different traffic conditions (low, medium and high) were also measured by repeating the measurements at different times of a day. In addition to the traffic generated by other users, our own traffic was generated using a cell phone in order to quantify the additional RF exposure (See Section 4.3 for more details on traffic generation). Five different quantities (i.e., measurands) were considered. Three measurands are based on code selective parameters and two measurands are based on frequency selective parameters, which enables a comparison of different measurement and quantification methods, providing additional insight for 5G NR RF exposure from base stations. For the code selective measurements with traffic, a special option of the 5G NR measuring receiver was used (i.e., "5G NR RE Statistics" implemented on the Deviser EM 860). This option enables a quantification of the RF exposure from traffic beams in terms of power histograms. Since the analysis of these histograms are not straightforward, different mathematical methods are proposed and the results of the methods are compared for the applicability. No specific polarization was selected during the indoor measurements. In both exposure cases (direct or indirect), the polarization, for which the maximum signal level was detected during spatial search, was measured (See Section 4.4 for details on spatial maximum search). For outdoor measurements, horizontal polarization was selected. The established RF exposure methods based on SSS and its measurement uncertainty are also integrated in the analysis. Considering the measurement uncertainties, the consistency between the code selective and frequency selective measurements are compared and the applicability of the newly selected methods to measure the RF exposure from traffic beams is also evaluated. 4.2 Measurement Parameters Measurement Locations The selected base station is located in Bern / Switzerland. The overview of the measurement locations with respect to the base station is given in Fig. 4. The list of three selected exposure measurement locations is given below: 1. Indoor measurements having Line of Sight (LoS) to the selected base station, denoted as "LOS" (See Fig. 5). With LoS locations, we mean locations with direct view on the base station antenna. The LoS room is located on the highest floor of a commercial building, having mainly offices, laboratories, and meeting rooms. In the selected LoS room, the 21NRM03 MEWS 9 of 78 windows could not be opened, and therefore, measurements have been performed with closed windows. 2. Indoor measurements without LoS to the selected base station, denoted as "nLOS" (See Fig. 6). According to the definition, nLoS locations are the locations from where the base station cannot be observed visually, due to walls, corridors or other obstructions. For these locations, a diffuse exposure is expected from the base station, where the radiation caused by reflections or secondary radiators is coming from different directions. The selected nLoS room is also located in the same building on the highest floor, having no direct sight to the base station. The measurements have been performed with closed windows. 3. Outdoor measurements in location with LoS (Line of Sight) to the selected base station, denoted as "Open air" (See Fig. 7). The selected place is located on the parking lot near the building. Fig. 4. Measurement locations shown on the map (a) (b) Base Station 21NRM03 MEWS 16 of 78 5 Analysis Methods for Measurements with 5G NR RE Statistics 5.1 Histogram of the Resource Element Power in a Frame The main goal of the analysis is to obtain the effective maximum RF exposure from the base stations (meaning the traffic beams) for different cases regardless of traffic beams or synchronisation signals. Using the software option "5G NR RE Statistics" on the 5G NR measuring receiver, the measurement results are obtained in power histogram representation (See Annex B for detailed explanation of this software option, the histogram characteristics and some measurement examples). The measuring mode "5G NR RE Statistics" provides a histogram (bin data) of the power of each resource elements (RE) in one NR frame in regularly spaced intervals (1 to 4 seconds depending on the bandwidth). Due to the computing time, not all consecutive frames can be evaluated, but in the worst case, 74 frames can be obtained within 5 minutes for 100 MHz bandwidth. The number of obtained frames increases with smaller bandwidths. In the histogram representation of captured signal as shown in Fig. 10, two components can be identified: 1. System noise (practically zero RE power, or power near to the noise level of the measuring system) 2. Traffic and Synchronization, represented as peaks laying on higher power levels. With the chosen settings, these components lay around 50 dB apart from each other (See Fig. 10). The evaluation for RF exposure analysis is performed on the traffic and synchronization part and the system noise is disregarded. Fig. 10. Sample Output Showing Different Components of the Captured Signal from Base Station Before conducting experimental measurements, we expected to see different clear peaks of traffic, each representing a traffic beam of the NR base station. The peaks with the highest power (not with the highest amplitude) would represent the traffic beam in our direction. However, as it can be seen in Fig. 10, the distribution of the traffic and synchronization signals on the histogram plot is not straightforward, and this creates a challenging case to analysis and to obtain the effective exposure from traffic beams. Therefore, we developed five different analysis methods for this evaluation. The details on the analysis of these measurements are given in next sections. System Noise Traffic + Sync 21NRM03 MEWS 17 of 78 5.2 Fading Problem The first encountered challenge was that the peaks of the traffic beams were not really sharp as expected, but rather broadened, thus overlapping in the histogram representation. This was not the case in conducted experiments performed directly with a NR signal generator connected to the receiver, where the SSS RE could clearly be identified from the traffic RE. Our explanation for this observation is the fading effect, in other words, the superposition of reflection from different sources with different time delays. On a signal with only one frequency, or limited bandwidth, the fading would have limited effect on the width of the RE power as shown by Eq. (1) representing the addition of two signals with different time delays. 1∙𝑒𝑖 2 𝜋 𝑓 𝑡+ 𝜆∙𝑒𝑖 2 𝜋 𝑓 (𝑡+𝑑𝑡)=(1+𝜆∙𝑒𝑖 2 𝜋 𝑓 𝑑𝑡)∙𝑒𝑖 2 𝜋 𝑓 𝑡 (1) where • 𝑓 is the frequency of the signal • 𝑑𝑡 is the additional time delay • 𝑡 is the time The amplitude of the superposed signal is therefore: |1+𝜆∙𝑒𝑖 2 𝜋 𝑓 𝑑𝑡| This means that for a reference frequency 𝑓0 the signal amplitude is: |1+𝜆∙𝑒𝑖 2 𝜋 𝑓0 𝑑𝑡| The more the frequency 𝑓 differs from the frequency 𝑓0, the more the difference between the amplitude of both signals. As order of magnitude for the significance of the frequency deviation, the time delay expressed in distance is a good criterium (See Table 7): Time Delay (expressed in distance (m)) Corresponding frequency (MHz) 𝑐∙𝑑𝑡 1/𝑑𝑡 3 m 100 MHz 10 m 30 MHz Table 7. Time Delays and Corresponding Frequencies In other words, already with delays of a few meters, the effect can be observed with a few MHz frequency deviations. In our measurement campaign, measurements were performed with a full bandwidth of 100 MHz. The effect of fading on the histograms is illustrated in Fig. 11. Fig. 11. Effect of Fading on Powers of SSB and Traffic REs in Histogram Representation In order to be able to estimate the power of the highest peak, we propose five different methods, whose efficiencies and validities are to be evaluated. Fading Sum REs vs Power # of REs SSB Power Traffic Power # of REs # of REs REs vs Power REs vs Power Power Power Power 21NRM03 MEWS 18 of 78 Method 1 The first method is based on the averaging of the power of the resource elements in a defined power window defined as follows: 1. Find 𝑃𝑑𝐵 so that 2. Find 𝑃𝑑𝐵−30 3. Integrate from (𝑃𝑑𝐵−30) to 𝑃𝑚𝑎𝑥𝑑𝐵 4. Average Fig. 12. Method 1 Method 2 The second method is based on the averaging of the power of the resource elements in a reduced power window defined as follows, in order to increase the robustness against outliers with high power. 1. Find 𝑃𝑑𝐵 so that 2. Find 𝑃𝑑𝐵−30 3. Integrate from 𝑃𝑑𝐵−30 to 𝑃𝑑𝐵 4. Average Fig. 13. Method 2 Method 3 The third method is based on the thresholding with respect to the given number of REs. 1. Find highest 𝑃𝑑𝐵 so that at least a given number of REs belongs to the same bin 2. Find 𝑃𝑑𝐵−30 3. Integrate from 𝑃𝑑𝐵−30 to 𝑃𝑑𝐵 4. Average The parameter "Given number of REs" are selected as 1000 in the analysis. Fig. 14. Method 3 # of REs Power Bins PdB PdB -30 PmaxdB # of REs Power Bins PdB PdB -30 PmaxdB # of REs Power Bins PdB PdB -30 PmaxdB Given # of REs ∑𝑅𝐸𝑖 𝑃𝑚𝑎𝑥𝑑𝐵 𝑖=𝑃𝑑𝐵 =2.5 % 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝑅𝐸𝑠 ∑𝑅𝐸𝑖 𝑃𝑚𝑎𝑥𝑑𝐵 𝑖=𝑃𝑑𝐵 =2.5 % 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝑅𝐸𝑠 21NRM03 MEWS 19 of 78 Method 4 The fourth method tries to identify the maximum value simply based on a threshold. 1. Find highest 𝑃𝑑𝐵 so that at least given number of REs belongs to the same bin 2. 𝑃𝑑𝐵 is the appreciation value The parameter "Given number of REs" are selected as 1000 in the analysis. Fig. 15. Method 4 Method 5 The fifth method considers the average power of the REs, the total number of which corresponds to the 10% of all REs and having the highest power. It assumes that the traffic generation of the app produces at least 10% traffic on the base station. 1. Get the recordings where 2. Integrate from 𝑃𝑚𝑎𝑥𝑑𝐵−30 to 𝑃𝑚𝑎𝑥𝑑𝐵 3. Average Fig. 16. Method 5 # of REs Power Bins PdB Given # of REs PmaxdB # of REs Power Bins PmaxdB -30 PmaxdB ∑𝑅𝐸𝑖 𝑃𝑚𝑎𝑥𝑑𝐵 𝑖=𝑃𝑚𝑎𝑥𝑑𝐵−30 ≥10% 𝑜𝑓 𝑎𝑙𝑙 𝑅𝐸𝑠 21NRM03 MEWS 20 of 78 6 Conversion of Measurement Results, Calculation of Appreciation Values and Measurement Uncertainties 6.1 Conversion from Power (dBm) to Power Density (W/m2) Correction Factors from Measurement Equipment Components Parameter Value @3649.8 MHz Measuring Receiver Correction Factor (𝐾MR) 0.06 (dB) Antenna Antenna Factor (𝐴𝐹) 32.46 (1/m) Cable Cable Loss (𝐶𝐿) 0.792 (dB) Table 8. Correction factor of measurement equipment These values are obtained with calibrations in the laboratory environment, which are in the service catalogue of METAS EMC Lab. It is worth to note that the correction factor of the measuring receiver is obtained with the work performed in the scope of MEWS project. Please refer to Annex A for more details. Base Station Parameters Based on the measuring recommendation of METAS [5], the following factors are used for the extrapolation of the SSS and traffic measurements. Factor Explanation Value 𝐾(φ,θ) (Obtained from the operator) Global Extrapolation Factor for field intensity per RE based on SSS identification, to the maximum field intensity over the full bandwidth 30.386 for LOS 35.741 for nLOS 115.657 for Open Air 𝐾Traffic Extrapolation Factor for Traffic field intensity per RE to the maximum field intensity over the full bandwidth =√Bandwidth Subcarrier Spacing =√100 MHz 30 kHz = 57.74 Table 9. Correction factors for base station parameters The global extrapolation factor 𝐾(φ,θ) is calculated by the operator as follows: 𝐾(φ,θ)=𝐾SSS∙𝐾Antenna (φ,θ) (2) where • 𝐾Antenna (φ,θ) Antenna correction factor taking into account the difference between the antenna diagram of the SS/PBCH signal and the antenna diagram of the total signal in the maximum permitted operating condition. The antenna correction factor depends on the azimuth φ and on the elevation θ. • 𝐾SSS SSS extrapolation factor. 𝐾SSS is calculated as 21NRM03 MEWS 21 of 78 𝐾SSS=√𝑃permitted 𝑃SSS(RE) (3) where • 𝑃permitted Maximum permitted effective radiated power (ERP) in W, taking into account the signal of all antenna ports: PSDCH, PBCH, and PDCCH • 𝑃SSS(RE) Actual effective radiated power (ERP) per resource element (RE) of the SSS of the SS/PBCH block in W For the code selective measurements of SSS as well as for traffic, the measuring receiver provides measurements for each resource element (RE). For the appreciation values, the extrapolation from one RE to whole bandwidth should be performed to obtain the total RF exposure. Conversion from dBm to dBµV 𝑈measured (dBµV)=𝑃 measured (dBm)+106.99 (4) where • 𝑃 measured (dBm) is the power expressed in dBm as measured by the receiver at a reference impedance of 50 Ω. • 𝑈measured (dBµV) is the voltage expressed in dB(μV) as measured by the receiver at the reference impedance of 50 Ω. Conversion from Voltage in dBµV to Field Intensity E in (V/m) 𝐸𝑥 (V/m)=10(𝑈𝑥(dBµV)+𝐾MR+𝐴𝐹+𝐶𝐿) 20 106 (5) where 𝑥 can be one of the following: • Frequency selective • SSSCode selective per RE • TrafficCode selective per RE Obtaining Extrapolated Field Intensity E (V/m) in case of Code Selective Measurements For code selective measurements, in order to obtain the total exposure from different signal types (SSS or Traffic), the corresponding extrapolation must be applied to the value obtained in Eq. 5 as follows: 𝐸SSSCode selective,Total (V/m)=𝐸SSSCode selective per RE (V/m)∙ 𝐾(φ,θ) (6) 𝐸TrafficCode selective,Total (V/m)=𝐸TrafficCode selective per RE (V/m)∙ 𝐾Traffic (7) Please consult [5] for more details and explanations. 21NRM03 MEWS 22 of 78 Obtaining Extrapolated Field Intensity E (V/m) for Frequency Selective Measurements According to [5], the appreciation value of the frequency selective measurements is calculated by extrapolating the measurements with RMS detector in the case without any additional traffic 𝐸𝐹𝑟𝑒𝑞−𝑠𝑒𝑙,𝑅𝑀𝑆,𝑁𝑜 𝑡𝑟𝑎𝑓𝑓𝑖𝑐 (𝑉/𝑚) (i.e. the result obtained in Eq. 5 for the case without any additional traffic) (corresponding to Parameter 4 in Table 6 according to the following formula: 𝐸freq−sel (𝑉/𝑚)=𝐸Freq−sel,RMS,No traffic (𝑉/𝑚)∙√1 127∙𝐾(φ,θ) (8) where 𝐾(φ,θ) is given in Table 9. This is in fact a simplification of the general formula where all 3 sectors from the same base station should be taken into account. As the selected measurement locations have each a small azimuthal angle towards the selected sector, the contribution from other sectors can therefore be omitted. Calculation using Spatial Averaging (V/m) For LOS and nLOS measurements, the values for different heights are obtained for code selective and frequency selective measurements and the spatial averaging of these measurements is obtained using the following formula (as given in Chapter B.3.3.3 in IEC 62232:2022 [3]): 𝐸=√∑𝐸𝑖2 3 𝑖=1 𝑁𝑝 (9) where • 𝐸 Spatially averaged electric field intensity at the evaluation location • 𝐸𝑖 Electric field intensity at the 𝑖𝑡ℎ measurement point Conversion from field intensity E in (V/m) to power density in (W/m2) 𝑃𝐷 (W/𝑚2)=𝐸2 377 (10) when the far field region condition is satisfied. This condition is fulfilled for all measurement points given in this document as the shortest distance to the base station was 100 m. 21NRM03 MEWS 23 of 78 6.2 Measurement Uncertainty of Appreciation Values Measurement uncertainty calculated for appreciation values in field intensity V/m given in Section 6.1 at the carrier frequency is given Table 10. Contributing Factor Source of Data Distribution Uncertainty (%) Deviser EM860 Uncertainty of Device Manufacturer Datasheet Rectangular 16.1 Uncertainty of Code / RMS Detection Normal 4.7 Antenna Uncertainty of antenna factor Calibration Certificate Normal 10.3 Cable Uncertainty from cable calibration Calibration Certificate Normal 0.8 Mismatches Deviser – Cable U-form 3.4 Cable – Antenna U-form 1.8 Total (k=1) 20.1 Uncertainty (k=2) 40.2 Table 10. Measurement uncertainty calculation for appreciation values 21NRM03 MEWS 24 of 78 7 Comparison of Different Traffic Protocols and Evaluation Methods All the selected data traffic protocols given in Section 4.3 are run on the corresponding apps on a 5G cell phone, placed in the direction of the traffic beam behind the measuring antenna (See Section 4.3). For these selected traffic cases, sample outputs are given in Fig. 17 for a measurement duration of 5 minutes. Fig. 17. Sample Outputs from Measurements of Different Apps for Traffic Generation and No Traffic Case 7.1 Measurement Procedure The following steps are followed to obtain the RF exposure appreciation values for different traffic protocols: 1) For each protocol, a measurement is performed at the spatial maximum location for 5 minutes, yielding in 74 histogram frames (Youtube measurements could not be performed in nLOS case as the signal level was too low). 2) Each histogram frame is evaluated using five different methods (see section 5.2) to obtain corresponding the RF exposure value in dBm. 3) The maximum of 74 values for each method is obtained in dBm. 4) The conversion from dBm to W/m2 given in Section 6.1 is performed for values of step 3 to obtain the appreciation value of the power density 𝑃𝐷 corresponding to the maximum RF exposure. 5) Steps 1 – 4 are repeated for each type of exposure (LOS, nLOS and Open Air) Note: The measurements are performed just at spatial maximum and for high traffic condition (See Table 3), therefore no spatial averaging is applied. 21NRM03 MEWS 25 of 78 7.2 Measurement Results The resulting extrapolated field strength values are given in tables Table 11 to Table 13 and depicted in Fig. 18 to Fig. 20. Protocol Method 1 Method 2 Method 3 Method 4 Method 5 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) Facetime (Both Cameras) 18.79 9.88 28.49 22.22 10.28 Facetime (Only Download Camera) 34.94 26.05 31.67 82.55 30.49 iPerf 16.94 9.05 17.60 50.90 21.45 No Traffic 3.67 2.83 9.95 7.88 4.55 RTS App Streaming 16.95 6.58 11.51 31.39 4.79 RTS App Video 41.71 18.98 26.87 70.26 16.77 Youtube 8.70 3.03 8.48 3.00 5.06 Table 11. Results of Measurements for Different Traffic Protocols in LOS Exposure Fig. 18. Results of Measurements for Different Traffic Protocols in LOS Exposure 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 Facetime (Both Cameras) Facetime (Only Download Camera) iPerf No Traffic RTS App Streaming RTS App Video Youtube Power Density (µW/m2) Traffic Protocols LOS Exposure Method 1 Method 2 Method 3 Method 4 Method 5 21NRM03 MEWS 32 of 78 8.3 Measurement Results of Scenario 2 (LOS Exposure) Measurand 1: SSS Time of Day SSS (Code Selective Appreciation Value) 𝑃𝐷 (µW/m2) 10:00 - 18:00 12.32 18:00 - 22:00 8.31 05:00 - 09:00 9.93 Table 18. Measurand 1 measurements for LOS exposure Measurand 2: No Traffic Time of Day Method 1 Method 2 Method 3 Method 4 Method 5 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 10:00 - 18:00 1.30 0.53 2.23 7.18 4.60 18:00 - 22:00 2.77 1.58 11.85 5.65 2.90 05:00 - 09:00 2.24 1.69 7.15 12.67 4.21 Table 19. Measurand 2 measurements for LOS exposure Fig. 25. Measurements for LOS exposure for Measurands 1 and 2 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Power Density (µW/m2) Time of Day LOS - No Traffic and SSS Method 1 Method 2 Method 3 Method 4 Method 5 SSS 21NRM03 MEWS 33 of 78 Measurand 3: Additional traffic with iPerf Time of Day Method 1 Method 2 Method 3 Method 4 Method 5 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 10:00 - 18:00 13.72 7.22 22.28 64.00 43.78 18:00 - 22:00 18.14 10.56 18.69 33.34 20.04 05:00 - 09:00 25.31 7.61 64.63 30.06 26.77 Table 20. Measurand 3 measurements for LOS exposure Fig. 26. Measurements for LOS exposure for Measurands 2 and 3 When Table 19 and Table 20 are compared, the increase of exposure with additional traffic can be calculated. The ratio 𝑃𝐷 𝑇𝑟𝑎𝑓𝑓𝑖𝑐/ 𝑃𝐷 𝑆𝑆𝑆 is depicted in Fig. 27. Fig. 27. Increase in Exposure with Additional Traffic for Different times of Day for LOS Exposure 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Power Density (µW/m2) Time of Day LOS - iPerf and SSS Method 1 Method 2 Method 3 Method 4 Method 5 SSS 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Increase in Exposure (µW/m2/ µW/m2) Time of Day Increase of Exposure with Additional Traffic for Different Times of Day Method 1 Method 2 Method 3 Method 4 Method 5 21NRM03 MEWS 34 of 78 Measurand 4 and 5: Frequency Selective (No Traffic and Additional Traffic with iPerf) Time of Day Peak with no Traffic RMS with no Traffic Peak with Additional Traffic RMS with Additional Traffic Frequency Selective Appreciation Value (from RMS no Traffic) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 10:00 - 18:00 62.76 1.42 1003.00 17.05 10.29 18:00 - 22:00 25.75 0.93 462.74 9.05 6.74 05:00 - 09:00 11.53 0.76 438.34 8.24 5.54 Table 21. Measurands 4 and 5 measurements for LOS exposure Fig. 28. Measurements for LOS exposure for Measurands 4 and 5 (Peak and RMS) 0 200 400 600 800 1000 1200 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Power Density (µW/m2) Time of Day LOS - Frequency Selective Peak without Traffic RMS without Traffic Peak with Traffic RMS with Traffic 21NRM03 MEWS 35 of 78 8.4 Measurement Results of Scenario 3 (nLOS Exposure) Measurand 1: SSS Time of Day SSS (Code Selective Appreciation Value) 𝑃𝐷 (µW/m2) 10:00 - 18:00 2.18 18:00 - 22:00 1.16 05:00 - 09:00 1.22 Table 22. Measurand 1 measurements for nLOS exposure Measurand 2: No Traffic Time of Day Method 1 Method 2 Method 3 Method 4 Method 5 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 10:00 - 18:00 0.62 0.42 1.58 2.34 0.83 18:00 - 22:00 1.36 1.12 1.55 5.58 1.80 05:00 - 09:00 0.11 0.05 0.66 0.27 0.48 Table 23. Measurand 2 measurements for nLOS exposure Fig. 29. Measurements for nLOS exposure for Measurands 1 and 2 0.0 1.0 2.0 3.0 4.0 5.0 6.0 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Power Density (µW/m2) Time of Day nLOS - No Traffic and SSS Method 1 Method 2 Method 3 Method 4 Method 5 SSS 21NRM03 MEWS 36 of 78 Measurand 3: Additional traffic with iPerf Time of Day Method 1 Method 2 Method 3 Method 4 Method 5 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 10:00 - 18:00 38.42 18.43 37.26 57.25 48.68 18:00 - 22:00 12.42 4.42 12.04 9.85 14.70 05:00 - 09:00 19.24 6.05 17.12 9.16 10.70 Table 24. Measurand 3 measurements for nLOS exposure Fig. 30. Measurements for nLOS exposure for Measurands 2 and 3 When Table 23 and Table 24 are compared, the increase of exposure with additional traffic can be calculated. The ratio 𝑃𝐷 𝑇𝑟𝑎𝑓𝑓𝑖𝑐/ 𝑃𝐷 𝑆𝑆𝑆 is depicted in Fig. 31. Fig. 31. Increase in Exposure with Additional Traffic for Different times of Day for nLOS Exposure 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Power Density (µW/m2) Time of Day nLOS - iPerf and SSS Method 1 Method 2 Method 3 Method 4 Method 5 SSS 0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 200.0 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Increase in Exposure (µW/m2/ µW/m2) Time of Day Increase of Exposure with Additional Traffic for Different Times of Day Method 1 Method 2 Method 3 Method 4 Method 5 21NRM03 MEWS 37 of 78 Measurand 4 and 5: Frequency Selective (No Traffic and Additional Traffic with iPerf) Time of Day Peak with no Traffic RMS with no Traffic Peak with Additional Traffic RMS with Additional Traffic Frequency Selective Appreciation Value (from RMS no Traffic) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 𝑃𝐷 (µW/m2) 10:00 - 18:00 12.54 1.01 795.27 16.15 10.15 18:00 - 22:00 4.06 0.57 425.32 6.99 5.76 05:00 - 09:00 4.92 0.70 681.17 14.71 7.03 Table 25. Measurands 4 and 5 measurements for nLOS exposure Fig. 32. Measurements for nLOS exposure for Measurands 4 and 5 (Peak and RMS) 0 100 200 300 400 500 600 700 800 900 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Power Density (µW/m2) Time of Day nLOS - Frequency Selective Peak without Traffic RMS without Traffic Peak with Traffic RMS with Traffic 21NRM03 MEWS 38 of 78 9 Consistency Analysis of Measurements The consistency of the measurements is investigated in two different aspects: 1) Traceable frequency selective exposure appreciation value vs traceable code selective SSS appreciation value 2) Traceable code selective SSS measurements vs the measurements obtained using 5G NR RE statistics option and analysed with newly proposed analysis methods 9.1 Frequency Selective vs Code Selective Measurements and Comparison of Appreciation Values The frequency selective and code selective appreciation values for SSS are also calculated in field intensity (E) unit V/m. In this way, the measurement uncertainty calculation given in Section 6.2 can be applied to compare these values using the confidence intervals. The results are given from Table 26 to Table 28 and depicted in figures from Fig. 33 to Fig. 35 with the corresponding intervals of confidence obtained from the extended uncertainty (k=2) of 40.2 % as given in Table 10. Time of Day SSS (Code Selective Appreciation Value) SSS (Frequency Selective Appreciation Value from RMS no Traffic) 𝐸 (V/m) 𝐸 (V/m) 10:00 - 18:00 0.459 0.752 18:00 - 22:00 0.443 0.475 05:00 - 09:00 0.564 0.626 Table 26. Code selective and frequency selective appreciation value for SSS for Open Air Exposure Fig. 33. Comparison of frequency and code selective appreciation values for open air exposure 0.000 0.200 0.400 0.600 0.800 1.000 1.200 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Field Intensity (V/m) Time of Day Open Air - Frequency Selective vs Code Selective Code Selective Frequency Selective 21NRM03 MEWS 39 of 78 Time of Day SSS (Code Selective Appreciation Value) SSS (Frequency Selective Appreciation Value from RMS no Traffic) 𝐸 (V/m) 𝐸 (V/m) 10:00 - 18:00 0.068 0.062 18:00 - 22:00 0.056 0.050 05:00 - 09:00 0.061 0.046 Table 27. Code selective and frequency selective appreciation value for SSS for LOS Exposure Fig. 34. Comparison of frequency and code selective appreciation values for LOS exposure Time of Day SSS (Code Selective Appreciation Value) SSS (Frequency Selective Appreciation Value from RMS no Traffic) 𝐸 (V/m) 𝐸 (V/m) 10:00 - 18:00 0.029 0.062 18:00 - 22:00 0.021 0.047 05:00 - 09:00 0.021 0.051 Table 28. Code selective and frequency selective appreciation value for SSS for nLOS Exposure 0.000 0.020 0.040 0.060 0.080 0.100 0.120 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Field Intensity (V/m) Time of Day LOS - Frequency Selective vs Code Selective Code Selective Frequency Selective 21NRM03 MEWS 40 of 78 Fig. 35. Comparison of frequency and code selective appreciation values for nLOS exposure These results show that the measurements using code selective and frequency selective SSS appreciation values are consistent for all cases of exposure and almost for all selected times of day. It is worth to note that the consistency is worst in nLOS case (i.e., the error bars barely crossing each other, or even just touching for low traffic case (05:00 – 09:00)). This is an expected result as the signal strength is the lowest in this case and the frequency selective measurement is much affected by the ambient noise. 9.2 Comparison of Traceable Code Selective SSS measurements and Newly Proposed Methods with 5G NR RE Statistics Option Regarding the RF exposure from traffic beams, the selection of the proper analysis method for the measurements can be based on the comparison with SSS measurements. The measurements of SSS (i.e., Measurand 1) in field intensity unit V/m is performed with a well-established method having a validated measurement uncertainty (as given in Section 6.2). In "No Traffic" case, the appreciation values obtained by 5 different analysis methods can be used to compare with the same RF exposure from SSS. In reality, not only SSS, but also PSS, PBCHDMRS, other synchronization signals such as CORESET, PDSCH DM-RS, PDCCH DM-RS are transmitted using the same RF power from the base stations. Even though no additional traffic is generated, the RF exposure from these signals can also be captured using Deviser "5G NR RE Statistics" option. Therefore, all selected methods are applicable also for "No Traffic" case. In the below figures, for the consistency verification, the measured values are depicted together with the error bars corresponding to the measurement uncertainty. 0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Field Intensity (V/m) Time of Day nLOS - Frequency Selective vs Code Selective Code Selective Frequency Selective 21NRM03 MEWS 41 of 78 Measurand 1: SSS Time of Day SSS (Code Selective Appreciation Value) 𝐸 (V/m) 10:00 - 18:00 0.459 18:00 - 22:00 0.443 05:00 - 09:00 0.564 Table 29. Measurand 1 measurements for open air exposure Measurand 2: No Traffic Time of Day Method 1 Method 2 Method 3 Method 4 Method 5 𝐸 (V/m) 𝐸 (V/m) 𝐸 (V/m) 𝐸 (V/m) 𝐸 (V/m) 10:00 - 18:00 0.254 0.189 0.745 0.503 0.278 18:00 - 22:00 0.421 0.393 0.445 0.788 0.453 05:00 - 09:00 0.237 0.180 0.290 0.598 0.362 Table 30. Measurand 2 measurements for open air exposure Fig. 36. Measurements for open air exposure for Measurands 1 and 2 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 0.900 10:00 - 18:00 18:00 - 22:00 05:00 - 09:00 Field Intensity (V/m) Time of Day Open Air - No Traffic and SSS Method 1 Method 2 Method 3 Method 4 Method 5 SSS 21NRM03 MEWS 48 of 78 Annex A – Traceable Calibration of 5G NR Measuring Receivers Using the synchronization signals, such as PSS, SSS or PBCH-DMRS, the traceability of 5G measurements can be achieved and the precision of 5G measuring receivers can be evaluated. PSS is present in the downlink resource grid with a periodicity given by the network and emitted by the base station at a predefined power when no power locking is applied [6]. In order to establish the fundamental capability for calibrating 5G measuring receivers, which deliver 5G EMF values for SSS signals, the setup is adjusted to calculate the power of SSS symbols. The 5G measurement setup consists of a hardware and a software module. In the following subsections, these modules are explained in detail. A.1 Hardware To retrieve 5G resource grid in a traceable way, it is crucial to obtain a physical setup, which can be calibrated using standards derived from SI units. To extract and the measure the power of synchronization signals present in the resource grid, a laboratory hardware system has been assembled in order to calibrated the signal of an Arbitrary Waveform Generator (AWG) capable to simulate 5G signals. The 5G signal from the generator is typically modulated at a selected carrier frequency either in FR1 or FR2-1, with a bandwidth ranging from 10 MHz to 50 MHz. The signal is sent to a calibrated 6-dB power splitter, one output of which is connected to a power meter to measure the instantaneous RMS power of the incoming signal. The other output of the power splitter is input to the RF mixer. This signal is down-converted to obtain the signal at an Intermediate Frequency (IF) of 45 MHz (see Fig. 39). This down conversion is necessary to obtain a signal at a lower carrier frequency, which allows for more accurate digitization (in terms of the number of digitization bits) and less noise compared to digitizing the original signal. The down conversion is achieved by mixing the 5G signal with a sine wave from a local oscillator, and then filtering out the higher frequencies (see Fig. 39). The down-converted signal is then sampled and recorded by a digital oscilloscope. To ensure the setup is traceable and calibrated, the frequency response of the experimental setup is determined using a continuous wave (CW) signal, as the metrology of CW signals is well established. Correction factors can be applied to the digitized data to compensate for the non-linearity and the broadband characteristics of the mixer and of the calibration chain, including the oscilloscope. The samples collected by the oscilloscope is then stored as a digital recording of the signal. This digital recording is then analyzed in terms of a post processing software. Fig. 39. Experimental setup for 5G power measurements using oscilloscope-based measurement method Oscilloscope Local Oscillator RF Mixer Low-pass Filter Hardware Module 5G NR Signal Power Splitter -6 dB Calibrated Power Meter 21NRM03 MEWS 49 of 78 A.2 Post-processing Software The extraction of SSS signals is accomplished purely by mathematical signal analysis. For this purpose, the recorded signal is treated as follows: I and Q signals (in-phase and quadrature components from the digitized 5G signal referred as I and Q respectively) are extracted by multiplying the input 45 MHz IF signal with cos(2𝜋𝑓𝐼𝐹∙𝑡) and sin(2𝜋𝑓𝐼𝐹∙𝑡), where 𝑓𝐼𝐹 is the intermediate frequency of 45 MHz, which is also known as digital-down conversion [7]. This complex baseband signal is digitally filtered in order to avoid the interaction of higher harmonics generated during the extraction of I and Q signals. Next, the following steps listed below are conducted to process the 5G signal: • Resampling the signal • Timing synchronization of the signal (i.e., defining timing reference of the 5G grid structure) • Extraction of the 5G symbols in time-domain (i.e., removing the cyclic prefix) • Calculating the 5G symbols in the frequency domain (also called resource elements) • Applying phase and frequency correction to the resource elements • Determining the position of the SSS resource elements within the 5G grid • Evaluating the mean power of the SSS resource elements within the 5G grid and computing the standard deviation of these values to get information on the quality of the power measurement. 5G employs numerous different configurations and the applications of the above-mentioned steps are not all trivial. Moreover, it is not required to perform the decoding of 5G signals in real-time. Hence, the above-mentioned post-processing algorithm has been chosen to preserve the full integrity of the resource elements. The most important steps of the above list are explained precisely in the following subsections. A.3 Resampling Input Signal As 5G NR is rather flexible for the employed subcarrier spacing defined by the numerology µ, where µ = 0, 1, …, 6, the input signal must be resampled according to the basic time unit for the applied numerology. The basic time unit for NR is 𝑇C=1 (480 𝑘𝐻𝑧)∙4096≅0.509 ns (1) where 𝑇C corresponds to the sampling time for a subcarrier spacing of 480 kHz and a Fourier length of 4096. The applied NR sampling time 𝑇𝑠 is 𝑇𝑠=𝜅∙𝑇𝐶 sec (2) where 𝜅=32 for numerology 0 and 𝜅=16 for numerologies 1 and 2 as defined in the ETSI standard for NR systems [8]. 21NRM03 MEWS 50 of 78 A.4 Timing synchronization After resampling the I and Q signals, timing synchronization is performed. This process is commonly referred to as cell search. There are no any precisely defined methods to obtain timing synchronization by the standards. Different synchronization techniques have been suggested in [9, 10, 11] using PSS present in the signal, which is applicable both for LTE and NR [2]. Like SSS, PSS is a frequency domain based BPSK maximum length sequences (known as Msequences). M-sequences were selected because of not having time offset – frequency offset ambiguity which is present in Zadoff-Chu sequences employed in LTE [12]. M-sequence is a pseudorandom binary sequence. It can be created by cycling through every possible state of a shift register of length n. This results in a sequence of length 2𝑛−1. Both PSS and SSS are of length 127 [8]. The sequence 𝑑𝑃𝑆𝑆(𝑛) for the primary synchronization signal is defined by 𝑑𝑃𝑆𝑆(𝑛)=1−2∙𝑥(𝑚) (3) where 𝑚=(𝑛+43∙𝑁𝐼𝐷 (2))𝑚𝑜𝑑 127 0≤𝑛≤127 𝑁𝐼𝐷 (2)∈{0,1,2} (4) and 𝑥(𝑖+7)=(𝑥(𝑖+4)+𝑥(𝑖)) 𝑚𝑜𝑑 2 (5) The initial condition to generate all possible 𝑑𝑃𝑆𝑆(𝑛) is: [𝑥(6) 𝑥(5) 𝑥(4) 𝑥(3) 𝑥(2) 𝑥(1) 𝑥(0)]= [1 1 1 0 1 1 0] (6) The presence and the position of PSS in the resource grid are obtained by correlating the recorded I and Q signals with those a priori known signals [13]. This method is implemented to obtain the timing synchronization. The M-sequence 𝑑𝑃𝑆𝑆(𝑛) as defined by ETSI [8] for PSS can be written as 𝐷𝑓𝑢 for 0≤𝑓≥126. This represents the M-sequence in the frequency domain centered at the carrier frequency. To obtain this sequence in time-domain, its Fourier transform has to be calculated in the following way: 𝐷𝑢(𝑡𝑖)= { ∑𝑑𝑃𝑆𝑆𝑗−1⋅𝑒−2𝜋(𝑖−1)(𝑗−1) 𝑁 63 𝑗=1 + ∑𝑑𝑃𝑆𝑆126−𝑗⋅𝑒−2𝜋(𝑖−1)(𝑁−𝑗) 𝑁 −63 𝑗=−1 (7) for 1≤𝑖≥𝑁 where 𝑁 is the selected Fourier transform size, which is dependent on the numerology. Based on the maximum number of subcarriers, bandwidth and subcarrier spacing, the ideal Fourier transform size can found to be 4096 for numerologies 0 and 1 and 2048 for numerology 2. The upper-part of the Fourier transform is considered as the resource elements below the carrier frequency and the lower-part of the Fourier transform is considered as the resource elements above 21NRM03 MEWS 51 of 78 the carrier frequency. Similar to the synchronization algorithms defined for LTE [2], the well-defined positioning of the PSS in the NR standard helps for the timing synchronization [9]. The structure of the 5G grid can then be found by knowing that PSS is at OFDM symbol number 0 in the SSB and the locations of SSBs are given according to [6]. The index u of the PSS provides 𝑁ID 2 which ranges from 0 to 2. This information would be used later on to decode the cell identity. When the correlation is calculated, the sample with the highest correlation value shows the start of the PSS symbol, which provides the timing synchronization. An example is shown in Fig. 40 for the case of 8 SSBs. 𝑁𝐼𝐷 (2) =0 yields the highest correlation values with 8 peaks, indicating the start of PSSs in every SSB. Fig. 40. Example output of correlation for different 𝑁𝐼𝐷 (2) with 8 SSBs A.5 Extraction of symbols in time domain Once the timing synchronization is successfully completed, the symbols in time domain can be extracted from the 5G NR signals. This is performed after removing the cyclic prefix of the OFDM modulated signal according to the NR modulation definition for the given configuration [8]. The cyclic prefix lengths are given in the following table. 𝝁 𝑵𝒔𝒍𝒐𝒕 𝒔𝒖𝒃𝒇𝒓𝒂𝒎𝒆 𝝁 𝑵𝒔𝒚𝒎𝒃 𝒔𝒍𝒐𝒕 Symbol duration in 𝑻𝐂 Prefix duration in 𝑻𝐂 0 1 14 2048 𝜅 160 𝜅 for 𝑙=0,7 144 𝜅 for other values of 𝑙 1 2 14 1024 𝜅 88 𝜅 for 𝑙=0,14 72 𝜅 for other values of 𝑙 2 4 14 512 𝜅 52 𝜅 for 𝑙=0,28 36 𝜅 for other values of 𝑙 Table 36. Cyclic prefix lengths for different numerologies In this table, 𝑙 is the OFDM symbol number in a subframe taking values 0, 1, …, 𝑁𝑠𝑙𝑜𝑡 𝑠𝑢𝑏𝑓𝑟𝑎𝑚𝑒 𝜇∙𝑁𝑠𝑦𝑚𝑏 𝑠𝑙𝑜𝑡 − 1 and the extended prefix is applied to the first OFDM symbol of the subframe. A.6 Conversion of symbols in the frequency domain After obtaining the symbols in time domain, they are converted into the frequency domain in terms of an FFT of length N, which dependent on the selected numerology as mentioned earlier. The upper Samples (k) Samples (k) Samples (k) Correlation Correlation Correlation 𝑁𝐼𝐷 (2)=0 𝑁𝐼𝐷 (2)=1 𝑁𝐼𝐷 (2)=2 21NRM03 MEWS 52 of 78 part of the FFT is considered as the resource element below center frequencies and the lower part of the FFT is considered as the resource element above the center frequencies. With that, for each slot, the uncorrected resource element 𝑎𝑘,𝑙 uncorr (real and imaginary value) of the lth symbol, and the kth subcarrier is obtained. A.7 Channel estimation When the ideal and the measured PSSs under concern are obtained, the effect of the transmission channel can be estimated. This estimation could be done using linear Minimum Mean Square Error (MMSE) based linear data fitting. The measured PSS is divided by the ideal PSS in the frequency domain complex plane and MMSE-based data fitting is applied to obtain the equalization parameters. Most mathematical modeling and calculation platforms such as MATLAB, Mathematica or Python have the corresponding functions for this purpose. The phase correction of each 5G resource element is obtained in terms of a first order fit of the decoded PSS sequence [2]: 𝜙𝑐𝑜𝑟𝑟(𝑘,𝑛)=𝜙0+𝐶1⋅𝑘+𝐶2⋅𝑛 (8) Herein, k is the index of the sub-carrier on the frequency axis of the 5G resource grid, whose range depends on the bandwidth of the input signal and Fourier size, and n is the index of the OFDM symbol on the time axis of the 5G grid. Moreover, C1 and C2 are the fit coefficients for the subcarrier and OFDM symbol respectively [2]. The phase and frequency corrections are performed by de-rotating the received OFDM symbol, i.e., multiplying them with 𝑒−𝑗2𝜋𝜑𝑐𝑜𝑟𝑟 [2]. After all corrections, the resource element 𝑎𝑘,𝑙 (real and imaginary value) of the lth symbol, and the kth subcarrier is obtained. The frequency correction for the resource element 𝑎𝑘,𝑙 can be implemented as [2] 𝑎𝑘,𝑙=𝑎𝑘,𝑙⋅𝑐𝑘 (9) An example is given in Fig. 41 for the output of the phase correction. Fig. 41. Phase correction based on measurement vs. MMSE-based estimation The application in a laboratory environment would enable a perfect clock synchronization of the measurement equipment, which typically eliminates the need for frequency correction. Phase Correction(rad) Symbols in FrequencyDomain (centeredaround NFFT/2) Phase correction measured MMSE based Estimation 21NRM03 MEWS 53 of 78 The MMSE-based correction is then extrapolated to cover the entire transmission bandwidth. The correction parameters obtained in this way are applied to the whole resource grid [2]. A.8 SSS Decoding SSS is located on OFDM symbol number 2 in SSB [8]. On the other hand, the physical layer cell identity (𝑁𝐼𝐷 𝑐𝑒𝑙𝑙) is determined using the PSS which provides 𝑁ID (2) and SSS which provides 𝑁ID (1) [8]. Therefore, the SSS signal must be decoded to find 𝑁ID 1. After successful detecting 𝑁𝐼𝐷 (2) and achieving timing synchronization, SSS can be detected by applying the following steps to the signal recording, in a similar way to PSS: 1) Generation of all SSS sequences based on detected 𝑁𝐼𝐷 (2) 2) Mapping to the subcarriers 3) Obtaining time domain candidate signals 4) Applying correlation to signal record to get 𝑁𝐼𝐷 (1) 5) Calculating physical cell ID The sequence to define the SSS can have 1008 (336 × 3) possible values [8]. Once 𝑁𝐼𝐷 (2) is detected, the selection narrows down to 336. Each of these 336 sequences having a length of 127 must be correctly generated. Mapping to the subcarriers and obtaining the time domain candidates are the same as given for PSS. For 𝑁𝐹𝐹𝑇=4096, the cyclic prefix for short symbols (𝐶𝑃𝑠ℎ𝑜𝑟𝑡) is equal to 288, which makes 𝑆𝑆𝑆 𝑆𝑡𝑎𝑟𝑡 𝑆𝑎𝑚𝑝𝑙𝑒 𝑖𝑛𝑑𝑒𝑥=8768 samples after PSS. For such a case, a window of 15000 samples shall be taken for correlation calculation. The index of the candidate signal with the highest correlation value gives the information on 𝑁𝐼𝐷 (1). An example is shown in Fig. 42. 𝑁𝐼𝐷 (1) value of 83 and the sample at 8768 give the highest correlation. Fig. 42. Example output of correlation for different 𝑁𝐼𝐷 (1) and the position within the recorded samples Finally, the 𝑁𝐼𝐷 𝑐𝑒𝑙𝑙 is determined with the expression [8] 𝑁𝐼𝐷 𝑐𝑒𝑙𝑙=3⋅𝑁𝐼𝐷 (1)+𝑁𝐼𝐷 (2) (10) Obtaining the Cell ID is not absolutely a requirement in the decoding algorithm; however, it constitutes a way to cross check the success of the decoding. Correlation 𝑁𝐼𝐷 (1) Correlation Samples (k) 21NRM03 MEWS 54 of 78 When all these steps are properly achieved, the OFDM symbol carrying the SSS symbol can be extracted and its 4-QAM constellation diagram can be obtained (See Fig. 43). Fig. 43. When the 5G resource grid is successfully decoded and the OFDM symbol, where SSS lies, is demodulated, the resultant 4-QAM constellation diagram includes various information. The position of SSS symbols is on the real axis (denoted in blue) whereas PBCH signals have imaginary parts as well (denoted in orange), therefore they do not coincide. In the ideal case, the magnitude of all symbols (therefore, of SSS) should be 1, the measured magnitude can be therefore used to calculate the received voltage of SSS. A.9 Estimating the power of the SSS Finally, the power of the extracted SSS is used for power measurement of the received signals. The power values 𝑃𝑘,𝑙 (𝑝) (in Watt) of the SSS are simply obtained using the amplitude 𝑎𝑘,𝑙 (in Volt) of the corresponding resource elements where the index (p) represents the SSS resource element as [2]: 𝑃𝑘,𝑙=|𝑎𝑘,𝑙|2 50𝛺 (11) With this, the SSS power of the downlink 5G signal is calculated from the arithmetic mean over all the resource elements that carry the SSS values within the operating bandwidth and within a frame as [2]: 𝑃𝑓𝑟=1 𝑤∑𝑃𝑘,𝑙 𝑤 (12) where the summation goes over a total of w SSS resource elements corresponding to the given frame fr. In a similar way, the standard deviation of these power values can be estimated as follows [2]: 𝜎𝑓𝑟=√1 𝑤∑(𝑃𝑘,𝑙−𝑃𝑓𝑟)2 𝑤 (13) Imaginary Axis Real Axis VSSS 4-QAM .................................... .................................... .................................... .................................... .................................... .................................... 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Frequency … … Time 21NRM03 MEWS 55 of 78 This standard deviation is an important number, since it provides an estimate of the "quality" of the measurement and the measurement uncertainty: if the standard deviation is small, it means that all SSS have almost the same power. It suffices to estimate the power based on one radio frame of 10 ms duration. The stability in time of the power measurement can itself be estimated by estimating the standard deviation across many radio frames. A.10 Performing frequency response correction The absolute calibration of the entire conducted set-up is performed by characterizing the frequency response of the scope amplitude with an input sine wave of known power for the frequency response correction, while constantly observing the output power of sinus generator with another power meter, as given in Fig. 44. Thus, the traceability to the SI units is achieved. The frequency response of the system is calculated as follows: 𝐾(𝑓)=𝑃𝑟𝑒𝑓(𝑓) 𝑃𝐷𝑆𝑂(𝑓) (14) This factor is applied to the power values of SSS resource elements and the SI traceability is established. Fig. 44: A sine wave is passed though the proposed set-up for 5G power evaluation and recorded in the oscilloscope as well as measured using a calibrated power meter. Calibration Plane Sinus Generator Oscilloscope Local Oscillator RF Mixer Low-pass Filter Hardware Module Power Splitter -6 dB Calibrated Power Meter 2 Calibrated Power Meter 1 21NRM03 MEWS 56 of 78 A.11 Sources of Uncertainty and Measurement Uncertainty Budget There are 5 influencing factors which contribute to the measurement uncertainty according to the guidelines of the JCGM 100:2008 [14]. 1. Uncertainty of power reference: In the setup, two different calibrated power meters are used for establishing the SI traceability. In the calibration certificates of the power meters, the uncertainties of these power references are given. Depending on the type and the quality of the power meter, this uncertainty changes. It was found out that the power meter used for FR1 has lower standard uncertainty (0.025 dB) than the one used in FR2 (0.060 dB). 2. Resolution of oscilloscope: The oscilloscope used in the measurement chain has 8-bit resolution. The uncertainty in the voltage amplitude measured by the oscilloscope has a rectangular distribution with distribution factor of 1.73. The amount of the uncertainty is calculated as follows: 𝑢Scope(dB)=20∙log10(1+ 1 2 × 28)=0.017 dB (15) 3. Frequency response of the measurement chain: The characterization of the measurement chain using conventional measurement methods such as a VNA or similar would yield the uncertainty contribution from the frequency response of the measurement chain. This effect includes also the broadband characteristics of the chain. In this application, a signal having a bandwidth up to 50 MHz is measured using the chain. The RF mixer and the bandpass filters do not have the same conversion loss throughout the band. This source of uncertainty also considers this factor. The values of 0.03 dB and 0.08 dB for FR1 and FR2 respectively are obtained from the calibration of the chain and the conversion loss characteristics of RF mixer and bandpass filters. 4. Stability of SSS measurements: This factor is obtained by the overestimation of the standard deviation of 50 different SSS measurements. 5. Reproducibility: This source of uncertainty has been investigated and the corresponding contribution of 0.01 dB has been found by repeating measurements for the decoding of the whole resource grid for the exact same configuration. The measurement uncertainty of the measurements for different carrier frequencies are given in Table 37. 21NRM03 MEWS 57 of 78 Number Source of Uncertainty Uncertainty [dB] Distribution type Distribution factor Standard uncertainty [dB] 1 Uncertainty of power reference 0.025 for 800 MHz and 3500 MHz 0.060 for 26955 MHz Normal 2 0.013 for 800 MHz and 3500 MHz 0.030 for 26955 MHz 2 Resolution of oscilloscope (8 bits) 0.017 Rectangular 1.73 0.010 3 Frequency response of the measurement chain 0.030 for 800 MHz and 3500 MHz 0.080 for 26955 MHz Rectangular 1.73 0.017 for 800 MHz and 3500 MHz 0.046 for 26955 MHz 4 Stability of SSS measurements 0.010 for 800 MHz 0.015 for 3500 MHz 0.020 for 26955 MHz Normal 1 0.010 for 800 MHz 0.015 for 3500 MHz 0.020 for 26955 MHz 5 Reproducibility 0.010 Normal 1 0.010 Total (k=1) 0.028 for 800 MHz 0.030 for 3500 MHz 0.060 for 26955 MHz Total (k=2) 0.056 for 800 MHz 0.060 for 3500 MHz 0.120 for 26955 MHz Table 37. Measurement Uncertainty Budget for 5G Measurements (given for all selected center frequencies) 21NRM03 MEWS 64 of 78 Fig. 53. Power Histogram for Case 6 These figures prove the validity of the method. The power difference between the traffic REs (peaks with changing amplitudes) and the SSS REs (overlapping peaks of smaller amplitude) are in perfect agreement with the case definitions for all traffic situations. The same observation is performed for the traffic peaks, which can be identified by their traffic-dependent amplitudes. As a numerical example, the measured power of SSS for Case 1 is -53.5 dBm whereas the measured powers of traffic REs for all traffic conditions are around -59.5 dBm, which yields the expected difference of 6.0 dB. Moreover, the measured SSS power (-53.5 dBm) is also matching well with the generator setup given for Case 1 in Table 39. Furthermore, the number of REs is also increasing proportionally with the incident traffic in all cases as expected. For example, in Case 2, the number of REs corresponding to 10% Traffic is around 750 REs, whereas, this number is doubled to 1500 REs for 20% Traffic case and quintupled to 3750 REs for 50% Traffic, as expected. 0 200 400 600 800 1000 1200 -66 -64 -62 -60 -58 -56 -54 -52 -50 -48 -46 -44 -42 -40 Number of Resource Elements () Power (dBm) Power Histogram for Case 6 10% Traffic 20% Traffic 50% Traffic 21NRM03 MEWS 65 of 78 Annex C – Analysis of Different Threshold Values for Method 5 and Different Measurement Bandwidths The goal of this study is to analyze different parameters for the measurement of traffic beams and for the quantification of the exposure from these beams. It is aimed with this study to find the optimum settings that would result in the correct quantification of exposure from traffic beams. C.1 Explanation of Analysis Basics For the analysis of the traffic measurements with 5G NR RE Statistics method, from five suggested methods, Method 5 found out to be the most suitable analysis method, proven by the consistency analysis with the traceable SSS measurements in the case where no additional traffic is generated in the vicinity of the measurement point. This method considers the average power of the REs, the total number of which corresponds to the 10% of all REs (Threshold Value) and having the highest power. It assumes that the traffic generation produces at least 10% traffic on the base station. The algorithm is as follows: 1. Get the recordings where 2. Integrate from 𝑃𝑚𝑎𝑥𝑑𝐵−30 to 𝑃𝑚𝑎𝑥𝑑𝐵 3. Average Fig. 54. Method 5 To estimate the effect of the threshold value in Method 5, numerous measurements with different thresholds are performed in LOS conditions. In these measurements, the threshold value was selected to be 5%, 10% (original suggested value) and 25%. Both the additional traffic case with iPerf and the no traffic case were measured for all these settings. Moreover, the measurement bandwidth was also varied to see the possible effect on the decisive value from the analysis method. The selected bandwidths for the analysis are 10 MHz, 20 MHz, 50 MHz (around the SSS) and 100 MHz (the whole bandwidth). Reducing the measurement bandwidth has certain advantages such as increasing the speed of captures, yielding in more frames per a recording of 5 minutes. However, staying focused on a subband of the whole bandwidth around SSS could result in missing certain traffic occurring in the communication channel (See Fig. 55 for the traffic blocks outside of the measurement bandwidth). This trade-off is aimed to be evaluated within the scope of this study. # of REs Power Bins PmaxdB -30 PmaxdB ∑𝑅𝐸𝑖 𝑃𝑚𝑎𝑥𝑑𝐵 𝑖=𝑃𝑚𝑎𝑥𝑑𝐵−30 ≥10% 𝑜𝑓 𝑎𝑙𝑙 𝑅𝐸𝑠 21NRM03 MEWS 66 of 78 Fig. 55. Sample full resource grid with PSS, SSS, PBCH and Traffic. The selected measurement bandwidth was placed around the center frequency of SSS. In the following tables and graphs, the measurements for iPerf Traffic and No Traffic cases are performed with 5G NR RE Statistics and analyzed with Method 5. Moreover, for the consistency analysis, the comparison with traceable SSS measurements is also given for No Traffic cases. In the charts, the error bars around the mean values correspond to the standard deviations. C.2 Measurements for Threshold Value of 5 % Measurement Bandwidth (MHz) Number of Measurements () Mean Value (V/m) Standard Deviation (V/m) 10 20 0.28 0.14 20 21 0.20 0.07 50 15 0.14 0.05 100 17 0.16 0.05 Table 40. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 5% for Method 5 21NRM03 MEWS 67 of 78 Fig. 56. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 5% for Method 5 for iPerf Traffic Measurements Measurement Bandwidth (MHz) Number of Measurements () Mean Value (V/m) Standard Deviation (V/m) Traceable SSS Measurements (V/m) 10 15 0.10 0.01 0.06 20 22 0.10 0.02 0.06 50 15 0.08 0.02 0.06 100 16 0.07 0.02 0.06 Table 41. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 5% for Method 5 for No Traffic Measurements Fig. 57. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 10% for Method 5 for No Traffic Measurements 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 10 MHz 20 MHz 50 MHz 100 MHz Field Intensity (V/m) Measuring Bandwidth iPerf Traffic Measurements with Different Bandwidths Measurement values Mean values with error bars 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 10 MHz 20 MHz 50 MHz 100 MHz Field Intesity (V/m) Measuring Bandwidths No Traffic Measurements with Different Bandwidths Measurement values Mean values with error bars Traceable SSS measurements 21NRM03 MEWS 68 of 78 C.3 Measurements for Threshold Value of 10 % Measurement Bandwidth (MHz) Number of Measurements () Mean Value (V/m) Standard Deviation (V/m) 10 20 0.19 0.08 20 21 0.16 0.05 50 15 0.10 0.02 100 14 0.14 0.04 Table 42. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 10% for Method 5 for iPerf Traffic Measurements Fig. 58. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 10% for Method 5 for iPerf Traffic Measurements Measurement Bandwidth (MHz) Number of Measurements () Mean Value (V/m) Standard Deviation (V/m) Traceable SSS Measurements (V/m) 10 15 0.09 0.02 0.06 20 22 0.10 0.02 0.06 50 15 0.07 0.02 0.06 100 16 0.06 0.02 0.06 Table 43. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 10% for Method 5 for No Traffic Measurements 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 10 MHz 20 MHz 50 MHz 100 MHz Field Intensity (V/m) Measuring Bandwidth iPerf Traffic Measurements with Different Bandwidths Measurement Values Mean values with error bars 21NRM03 MEWS 69 of 78 Fig. 59. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 10% for Method 5 for No Traffic Measurements C.4 Measurements for Threshold Value of 25 % Measurement Bandwidth (MHz) Number of Measurements () Mean Value (V/m) Standard Deviation (V/m) 10 20 0.11 0.02 20 19 0.11 0.02 50 5 0.07 0.01 100 9 0.10 0.02 Table 44. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 25% for Method 5 for iPerf Traffic Measurements Fig. 60. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 25% for Method 5 for iPerf Traffic Measurements 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 10 MHz 20 MHz 50 MHz 100 MHz Field Intensity (V/m) Measuring Bandwidths No Traffic Measurements with Different Bandwidths Measurement values Mean values with error bars Traceable SSS measurements 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 10 MHz 20 MHz 50 MHz 100 MHz Field Intensity (V/m) Measuring Bandwidth iPerf Traffic Measurements with Different Bandwidths Measurement Values Mean values with error bars 21NRM03 MEWS 70 of 78 Measurement Bandwidth (MHz) Number of Measurements () Mean Value (V/m) Standard Deviation (V/m) Traceable SSS Measurements (V/m) 10 15 0.07 0.02 0.06 20 21 0.07 0.02 0.06 50 9 0.06 0.02 0.06 100 7 0.05 0.03 0.06 Table 45. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 25% for Method 5 for No Traffic Measurements Fig. 61. Different Measurement Bandwidths, Number of Measurements, Mean values and Standard Deviation of Measurements with a Threshold Value of 25% for Method 5 for No Traffic Measurements C.5 Summary A summary of measurements for decisive values using different settings are given in Table 46. Bandwidth Threshold 5 % Threshold 10 % Threshold 25 % SSS (V/m) Traffic (V/m) SSS (V/m) Traffic (V/m) SSS (V/m) Traffic (V/m) 10 MHz 0.10 0.28 0.09 0.19 0.07 0.11 20 MHz 0.10 0.20 0.10 0.16 0.07 0.11 50 MHz 0.08 0.14 0.07 0.10 0.06 0.07 100 MHz 0.07 0.16 0.06 0.14 0.05 0.10 Table 46. Summary of Measurements for Decisive Values using Different Settings Measured increase of exposure for different settings and the corresponding standard deviations are given in Table 46. Bandwidth Ratio Traffic/SSS for 5% Traffic () Ratio Traffic/SSS for 10% Traffic () Ratio Traffic/SSS for 25% Traffic () Standard Deviation () 10 MHz 2.80 2.11 1.57 0.50 20 MHz 2.00 1.60 1.57 0.20 50 MHz 1.75 1.43 1.17 0.24 100 MHz 2.29 2.33 2.00 0.15 Table 47. Measured Increase of Exposure for Different Settings and Corresponding Standard Deviations 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 10 MHz 20 MHz 50 MHz 100 MHz Field Intensity (V/m) Measuring Bandwidths No Traffic Measurements with Different Bandwidths Measurement values Mean values with error bars Traceable SSS measurements 21NRM03 MEWS 71 of 78 No strong correlation was found between the selected bandwidth and the decisive exposure values. When different thresholds are considered, a measurement bandwidth of 100 MHz results in more stable measurements (i.e., the lowest standard deviation). Moreover, the ratio of Traffic to SSS would show how much the exposure increases with the generated traffic. This value is highest for the 10 MHz / 5% whereas the second highest value occurs for 100 MHz / 10%, which are the originally selected values for this analysis. Moreover, the consistency between the traceable SSS measurements and the no traffic measurements with 100 MHz bandwidth is very good for all selected threshold values. Regarding all these facts, the selection of 100 MHz bandwidth for the base station under concern and the threshold value of 10% would be confirmed to be the best selection to be used in Method 5 to quantify the exposure arising from the traffic beams. 21NRM03 MEWS 72 of 78 Annex D – Emails sent as Evidence of Submission to Standards Bodies 21NRM03 MEWS 73 of 78