THE FIFTH-GENERATION new radio (5G NR) cellular network is the fifth generation of cellular networks. Its rollout all over the world is supposed to start within the next few years. Release 15 of the third-generation partnership project (3GPP) standard has been available since the end of 2017. It already defines the main aspects of 5G NR. Like its precursor the Long-term Evolution (LTE) standard, 5G NR will use orthogonal frequency division multiple access (OFDMA) in the downlink. Like LTE, it can be used in frequency division duplex (FDD) or time division duplex (TDD) mode. The main differences from LTE, in the sense of human exposure assessment procedures, are the use of massive, interactive, and agile beam forming and the reduction of the amount of signals transmitted independently of the current traffic load and user behavior. It seems to be a good idea to discuss exposure assessment methods of 5G NR base stations before the rollout begins.
ASSESSMENT OF CURRENT TOTAL EXPOSURE
The current total human exposure due to the thermal effects of electromagnetic fields at a specific point in space and during a specific observation time is always assessable with the general assessment method described here. This method is applicable to any kind of signal and therefore also to 5G NR signals.
The general assessment method for exposure due to the thermal effect of external electromagnetic fields is described by eqns (9) and (10) of the International Commission on Non-Ionizing Radiation Protection (ICNIRP) guidelines from 1998 (ICNIRP 1998). These equations mean that one should measure the square of the root-mean-square (rms) value of the weighted external electric and magnetic field strength. The results are two normalized exposure values. Exposure values up to unity are permissible. The frequency response of the weighting filter is the reciprocal of the reference levels, which vary with frequency. The integration time of the rms detector can be as high as 6 min for frequencies below 10 GHz. For higher frequencies, the maximum permissible integration time decreases with increasing frequency. Note that shorter integration times may be used, but they may lead to an overestimation of the current exposure. If the level fluctuations of the assessed field are fast compared to the current integration time, an overestimation will not occur.
Above 10 MHz, the reference levels for the magnetic field strength are equal to the reference levels of the electric field strength divided by the free space impedance. Below 10 MHz, the reference levels for the magnetic field strength are equal to or higher than the reference levels of the electric field strength divided by the free space impedance. The magnetic field strength is equal to the electric field strength divided by the free space impedance if the distance d to the transmitter is so far that reactive power components of the transmitter can be neglected. In International Electrotechnical Commission (IEC) 62232 (IEC 2017) this area is named source regions II and III. There are three conditions for the assumption to be in source regions II or III: d > λ, d > D, and d > D2/4λ. Therein λ is the wavelength of the transmitted field, and D is the largest dimension of the transmitting antenna. It is clear now that the assessment of the magnetic field strength is not necessary for frequencies above 10 MHz and distances to the transmitter greater than d.
Broadband field strength meters with shaped probes measure the current exposure directly and exactly as described above. They are therefore always the first choice if a reliable assessment of the current total exposure is required. Shaped probes are available for frequencies up to 50 GHz and for several standards including ICNIRP (1998) and IEEE C95.1-2005 (IEEE 2006) standards.
ASSESSMENT OF CURRENT EXPOSURE DUE TO A SPECIFIC SERVICE
For the assessment of exposure due to a specific service, often the frequency selective measurement technique is used. In the close vicinity of the base stations of a single site, it can, however, often be assumed that exposure due to other transmitters is negligible. Even broadband field strength meters with flat frequency response probes can assess the current exposure in this case. The square of the rms value of the broadband field strength divided by the square of the reference level at the transmitter frequency is the current exposure value in this case.
If transmitters at different frequencies would, however, contribute to the current total exposure significantly, then the frequency selective measurement technique is necessary to suppress their contributions.
Note that probes for broadband field strength meters are, in general, isotropic. This means their measurement result does not depend on the direction of arrival and the polarization of the measured field. The frequency selective measurement technique, however, demands antennas instead of probes. Isotropic antennas are available up to 6 GHz only. Selective meters, which support isotropic antennas, are more or less spectrum analyzers, which are often highly optimized for the assessment of human exposure.
Using a normal spectrum analyzer with a directional antenna instead of an isotropic antenna results in an assessment procedure which is more complicated. To assess the exposure at a specific point in space, it is necessary to vary the direction and polarization often enough to be sure to capture the exposure due to the maximum coupling to the field. The exposure value with the maximum coupling to the field is the relevant exposure value then.
A method called the handheld sweep method in IEC 62232 (IEC 2017) enables a practicable assessment with directional antennas. The user changes the position, direction, and polarization of the antenna randomly within a volume of interest by moving the antenna by hand. A maximum hold function in the selective meter or spectrum analyzer detects the maximum exposure during the sweeping process. An advantage is that the assessment covers not only a specific point in space but even a volume of interest. A disadvantage is, however, that the dwell time at each point in space, direction, and polarization will be very short if the assessment has to be finished within a reasonable time. Thus the rms averaging times used must be short enough to ensure that the field strength at the point with the maximum field strength is also captured with the optimal coupling to the field.
When using frequency selective measurement techniques, an important issue is to select an appropriate selection filter to capture nearly 100% of the power of the channel of interest and as little power of the adjacent channels as possible. For the assessment of 5G NR base stations, this means that highly selective channel filters of selective meters are much more suitable than the Gaussian-shaped selection filters of a normal spectrum analyzer because with Gaussian filters the contributions of different 5G NR carriers cannot be separated sufficiently.
Another important issue is to set an appropriate averaging time for the rms detector. It must be selected long enough to reduce the fluctuations of the detected power to a negligible value but should not be much longer, in order to speed up the measurement process. It must also be shorter than the maximum permissible integration time of the ICNIRP guidelines.
ASSESSMENT OF THE THEORETICAL MAXIMUM EXPOSURE DUE TO 5G NR BASE STATIONS
The transmitted power of a 5G NR base station depends strongly on the current traffic load and user behavior. This is also true for Global System for Mobile (GSM) communications, Universal Mobile Telecommunications System (UMTS), and LTE. This means in practice that the current exposure measured within a specific observation time could be much lower than the theoretical maximum exposure. Regulators in some countries (e.g., Germany, Austria, and Switzerland) require, therefore, measurement of the exposure due to signal components which do not depend on the current traffic load or user behavior and then extrapolation to the theoretical maximum exposure.
The use of massive, interactive, and agile beam forming is new in 5G NR. This means in practice that it is extremely unlikely that the complete power of a 5G NR base station is concentrated on a single beam with the maximum possible antenna gain over a time span as long as the maximum permissible rms averaging time. To document this fact, manufacturers of 5G NR base stations published statistical analysis of some realistic theoretical and practical scenarios (IEC 2019). Their result is more or less that in 95% of all cases the current exposure is at least about 4 times lower than the theoretical maximum exposure. They propose to use the 95th percentile exposure as the relevant measure of the exposure because it is a much more realistic representation of exposure scenarios in the vicinity of base stations compared to the theoretical maximum exposure.
It is a regulatory decision of countries to permit base stations that meet the permissible exposure limit in 95% of all cases or even in the worst case. If the 95th percentile approach is accepted as the more realistic and relevant exposure measure, then an extrapolation from current measurements will be demanded too. It will also be necessary to agree on a reduction factor between the worst case and the 95th percentile exposure. This is true because extrapolation can only be done in a well-defined situation like the theoretical maximum exposure, for example. After extrapolation to the worst case has been done, multiplication with the agreed reduction factor would result in the 95th percentile exposure.
It is, however, not clear yet if extrapolation to the worst-case exposure is possible for 5G NR at all. The next sections describe potential methods for such an extrapolation.
Signal components transmitted by 5G NR base stations which do not depend on the current traffic load and user behavior
Beside the detailed information published by 3GPP, a good overview of the physical layer of 5G NR can be found in an application note from Keysight (Campos 2017).
For an extrapolation to the theoretical maximum exposure of a base station, there must be some signal components which are transmitted independently from the current traffic load and user behavior. In the physical layer of 5G NR, there are only the synchronization signals and physical broadcast channel (SS/PBCH) block, which fulfill this condition.
An SS/PBCH block occupies 240 contiguous subcarriers and four contiguous symbols. It contains four different types of signals, distributed within the block as shown in Fig. 1.
The primary synchronization signal (PSS) occupies 127 contiguous subcarriers in the center of the block at symbol number 0. All other subcarriers within the SS/PBCH block at symbol number 0 are not used and set to 0.
The secondary synchronization signal (SSS) occupies 127 subcarriers in the center of the block at symbol number 2.
The PBCH and the associated demodulation reference signals (DM-RS), which are necessary to demodulate the associated PBCH, occupy most of the remaining resource elements within the SS/PBCH block at symbol numbers 1, 2, and 3. Only 17 resource elements at symbol number 2 are not used and set to 0.
The user equipment (UE) uses PSS and SSS for time and frequency synchronization during the cell search. Only three different, predefined data sequences are transmitted as PSS. The actual data sequence depends on the sector identity (ID) of the cell. There are 1,008 different, predefined data sequences that are transmitted as SSS. The actual data sequence depends on the cell ID, which is 3 times its group ID plus its sector ID.
The PBCH and the associated DM-RS contain some basic information about the cell and also the block number of the specific SS/PBCH block within a SS/PBCH burst set.
A frame in the 5G NR cellular network has a length of 10 ms. A half-frame has a length of 5 ms. Some half-frames contain SS/PBCH blocks in burst sets. The UE may assume that the periodicity of SS/PBCH burst sets is two frames. An SS/PBCH burst set may contain up to 4 (f < 3 GHz), 8 (3 GHz < f < 6 GHz), or 64 (f > 6 GHz) SS/PBCH blocks and is transmitted within a single half-frame. Each SS/PBCH block in a burst set is transmitted on antenna port number 4000 but on different beams.
The UE detects which block within a burst set is received with the maximum signal strength and reports the associated block number to the cell. The use of different beams for SS/PBCH blocks with different block numbers has the objective of finding a first estimate of the optimal beam to service a specific UE.
There are two principle methods for extrapolating to the theoretical maximum exposure based on SS/PBCH block measurements. The first method is based on a frequency selective measurement of the field strength of the SS/PBCH block, and the second method is based on a code selective measurement of the field strength of the SS.
Frequency selective extrapolation
The first method is called frequency selective extrapolation here. A similar method has already been proposed for LTE in IEC 62232 (IEC 2017). The first precondition for this method is that resource elements outside the SS/PBCH blocks are never transmitted with a higher power and a higher antenna gain due to beam forming compared to the nonzero resource elements used in SS/PBCH blocks. The second precondition is that all nonzero resource elements used in SS/PBCH blocks are transmitted with a constant power and a constant antenna gain due to beam forming. It is not clear yet under which circumstances these two preconditions are met in real 5G NR systems. If both preconditions are met, the maximum theoretical exposure index EImax can be calculated with eqn (1):
The maximum field strength ESSblock is measured with a channel filter, which is 127 subcarriers broad and centered at the center frequency of the SSS and PSS. The averaging time of the applied rms detector must be shorter than the length of four symbols to be able to settle within the length of an SS/PBCH block completely. On the other side, the averaging time should not be much shorter to attenuate short time-level fluctuations as much as possible.
The parameter Eref is the reference level for the electric field strength at the center frequency of the SSS and PSS.
The parameter Nsc is the total number of subcarriers used in the associated 5G NR channel.
The parameter kTDD is the ratio of the maximum downlink time within a frame to the total frame length.
The parameter ksystem is a parameter which is set to unity by default but might be different for special system settings. For example, it must be set to 0.5 if the resource elements of the SS/PBCH block are transmitted with twice the power of the other resource elements.
Note that the power of SS/PBCH blocks from different cells, which coincide in frequency, can’t be separated by this method. The extrapolation will extrapolate to the total exposure due to all received cells in this case.
SS demodulation-based extrapolation
The second method is called SS demodulation-based extrapolation here. The basic idea behind it is that the received signal strength of PSS and SSS can be measured separately for each cell if a part of the cell search procedure is implemented in a measurement device. The advantages compared to the frequency selective method are that the first precondition of the frequency selective method is not necessary here and that the exposure due to different cells can be separated. Note that the second precondition of the frequency selective method is also necessary here. The maximum theoretical exposure index EImax can then be calculated with eqn (2):
Using this method, one would implement the initial part of a cell search procedure in a measurement device and then identify the SS/PBCH blocks with the highest field strength EPSS and ESSS of the PSS and SSS.
If the same sector ID is used by more than one of the received cells, the assessment of cell-specific exposures based on PSS measurements is not possible. This can happen if cells from different sites are receivable with relevant field strength levels at the location of interest. So in the general case it will probably be more accurate to assess the exposure based on SSS measurements only:
Note that the parameter ksystem is unity if the transmitted power per resource element is the same for SS and the traffic load and if the beams used for SS and traffic load are exactly the same. But in the general case, ksystem is a product of up to five factors reflecting five potentially relevant aspects of the system settings and the environment, which may affect SS and traffic load differently. The first aspect is the difference of the transmitted power per resource element. The second aspect is the gain difference of the used beams. The third aspect is the difference of the use of two polarization planes. The fourth is the difference of the combined envelope of the beams. The fifth aspect is the different influence of reflectors. It will be difficult in practice to get precise values for all of these five factors, but probably some of them will be available from the network providers and some of them will be estimable.
The current human exposure to the electromagnetic fields transmitted by 5G NR base stations is assessable with broadband field strength meters or with common spectrum analyzers using general assessment methods. Extrapolation to the theoretical maximum exposure can, however, be assessed only if certain preconditions are met. There are two potential methods for the extrapolation process, namely the frequency selective method and the SS demodulation-based method. At the least, the preconditions for the second method are likely to be met in practice. If the estimation of the 95th percentile exposure is necessary too, regulators must agree to a reduction factor between the theoretical maximum exposure and the 95th percentile exposure. The 95th percentile exposure is the theoretical maximum exposure multiplied by the agreed-upon reduction factor. It is not clear yet which extrapolation methods for 5G NR, if any, will be included in future releases of IEC 62232.