Effective Analysis of Human Exposure Conditions with Body-worn Dosimeters in the 2.4 GHz Band

Four indoor enclosures of different sizes were selected to perform the experimental measurements, whose volumes were 63 m³ (dimensions of 12 × 1.26/3 × 2.45 m), 162 m³ (27.15 × 1.93 × 3.1 m), 57 m³ (9 × 2.56 × 2.47 m), and 63 m³ (10 × 2.56 × 2.47 m). The width of the first enclosure was not constant. In the first and second enclosures, the length of the predefined path was 12 m. In the third and fourth enclosures, the length of the predefined path was of the maximum dimension, that is, 9 and 10 m, respectively. One factor that affects the BSE is the type of materials making up the indoor enclosures, as exposure levels increase in the case of environments with conductive materials. Specifically, the enclosures we used were composed of non-reflective materials. In those conditions, the BSE becomes relevant, as the reflected rays logged by PEM under BSE are weaker than in the case of conductive materials.

The results obtained in the preliminary stage are summarized in Figure 4, which compares the logged data by the three PEMs (one at the back, another in the front, and the third situated 1 m away) while the user was walking towards and away from the AP. E-field levels logged by the worn PEM in LoS with the radiation source are very similar to those recorded by the PEM located 1 m away from the wearer, both in LoS with the radiation source, although it is appreciable that the PEM in contact with the body registers lower levels⁷. For both paths, levels collected by worn PEMs in the shadow area are lower than data collected by worn and not worn PEMs in LoS.

E-field levels logged by the PEMs in each position were very similar in both paths, but there were some differences. Considering the path away from the AP, finite-difference time-domain (FDTD) analysis showed that incident waves can bend around the body user and reach the worn PEM on the opposite side, and even the PEM situated 1 m away, where the BSE is weaker. This effect is more significant in indoor environments, as the shadowed region of the body is small. This was why the data logged by PEMs situated 1 m away from the user in both paths was similar to the exposed conditions.

Regarding the worn PEMs, the effect of the coupling with the body causes a distortion in the PEM radiation pattern (RD) that subsequently affects the logged data. However, as logged data by worn PEMs in LoS tend to be similar, but lower than logged data by PEMs situated 1 m away, it can be concluded that in LoS conditions, the human body has a negligible influence in comparison with the distortions due to the BSE.

As seen in Figure 4, in all PEM positions the E-field levels tend to be lower for the path towards the AP, where the user position is frontal to the radiation source. In the GHz range, the SAR in the whole body (SAR_WB) is slightly higher under a frontally incident plane wave because of human morphology: larger skin areas and rougher surfaces (toes, feet, chin, face) are contained on the frontal side of the body. The E-field can effectively impinge on these small body parts, which are typical peak SAR locations in the GHz range¹⁷.

The transmission from the AP is discontinuous, so many of the logged levels by the PEMs do not reach the lower sensitivity threshold, and the number of non-detects becomes too large. The percentage of non-detects considered as acceptable is below 60%, where substitution might be acceptable, as explained by Helsel¹⁸. Although in the results shown in Figure 4, the maximum number of non-detects is 50%, close to the accepted level of 60%, the tests with an AP are reliable enough to confirm that 1 m is an optimal distance to avoid the BSE.

Therefore, the position of the PEM situated 1 m away from the user is optimal to log reliable levels of exposure to the E-field, and is unaffected by the underestimation caused by the influence of the body. Taking into account these considerations, the measurements were performed in the four selected environments, in both horizontal and vertical polarizations and following the methodology described in the previous section: with two PEMs, one worn by the user and in NLoS, and the second situated 1 m away from the user and in LoS with the radiation source.

Figure 5 and Figure 6 show the E-field levels in the first and second enclosures, in a semi-logarithmic scale and in both polarizations along the path towards the radiation source comprised of a biconical antenna and a signal generator. The BSE underestimation is directly dependent on the size of the environment: the underestimation is greater in the second enclosure, and in turn, the effect is greater in outdoor, rather than indoor, enclosures. It is notable that BSE underestimation is larger with vertical than with horizontal polarization, since the polarization type of the main radiation source affects the degree of influence of the BSE. In order to avoid the high number of non-detects in the case of shadow without a further treatment of logged data, the measurements in both polarizations were repeated with a transmission power of 25 dBm (316.12 mW) in the second enclosure. Figure 6 presents the rescaled measurements to 20 dB in both polarizations, and in a semi-logarithmic scale to perceive the E-field levels in the case of shadow. In the case of horizontal polarization, the non-detects have been avoided, although in vertical polarization, the percentage is still considerable.

Measurements in both polarizations were performed in all enclosures under test conditions. Figure 5 shows the results of the first enclosure, shadowed data being similar in both polarizations. However, from the results of the second enclosure, the largest one, shown in Figure 6, the difference of shadowed data in both polarizations is more notable than in Figure 5.

In order to quantify the difference of shadowed data in both polarizations in each enclosure, Table 2 presents the polarization factor (PF) that relates the ratios between the means of non-shadowed and shadowed data in both polarizations, as is shown in (1):

(1)

From Table 2 it can be deduced that the larger the enclosure is, the greater the differences found between non-shadowed and shadowed data for vertical polarization. The results of this study show a more significant underestimation in vertical than in horizontal polarization, because for frequencies around 2,100 MHz, the localized SAR in limbs and head/trunk is higher for vertical polarization, in a standing position, and when waves impinge on the body from the front or back¹⁷. In addition, the user is not small in comparison to the wavelength, so the vertical polarization is at a worst-case level in terms of absorption of the incident wave²⁴. When the major axis of the human body is parallel to the electric field vector (which happens when the polarization of the biconical antenna is vertical), the specific absorption rate (SAR) of the human body reaches maximum values¹⁹. Theoretically, the vertically polarized waves are largely shielded by the human body, in comparison to the horizontally polarized waves. This is due to the fact that in vertical polarization, the E-field oscillates parallel to the long axis of the wearer⁸. As the polarization of the antenna is a key factor in the BSE, the proper polarization is vertical, in order to detect the maximum influence of the presence of the user on the measurements of the worn PEM and in NLoS²⁰.

The exposure levels obtained in the four enclosures under test conditions are shown in Figure 7 in a semi-logarithmic scale. The simulation results are shown together with the measurements at each point of the predefined route, demonstrating that both types of data vary in the same way in relation to their distance from the radiation source.

Table 3 summarizes the measured and simulated E-field levels, respectively. For each indoor enclosure the mean, standard deviation, and the maximum and minimum values are provided. It is worth noting the similarity between the statistical values of the experimental and simulated data. The similarity between each pair of experimental and simulated data series has also been checked in terms of the p-value obtained with the Kolmogorov-Smirnov (KS) test. The p-values are shown in Table 3. The p-values were always greater than the significance level of 0.05, so there is an adequate match between each pair of experimental and simulated data series. In addition, it has also been confirmed using the KS test that the cumulative distribution function (CDF) of each series, experimental or simulated, always follows the lognormal statistical distribution in both polarizations.

Figure 7 shows the measured and simulated data in the indoor enclosures being used for testing and the compliance with the thresholds established in the European legislation based on the ICNIRP, which forms the basis of many exposure standards presently applied worldwide in general, domestic, and occupational contexts. In the case of the general population, the limit of exposure to non-ionizing radiation at the 2.4 GHz frequency is 61 V/m. The value of 61 V/m established in the ICNIRP is not the most restrictive limit in terms of human exposure. Other standards exist around the world: in North America, IEEE establishes less restrictive limits: 66.7 V/m for uncontrolled environments, the equivalent for the general public in the ICNIRP. In addition, more restrictive regulation exists in Eastern Europe, such as the case of Russia where the strictest limit for the general population is 3.14 V/m. In Figure 7, the measurements compared with the ICNIRP threshold are not affected by the uncertainties of the PEM, providing reliability in the extracted conclusions with regard to regulation compliance.

Figure 1: Location of the PEMs during the experiment.

Figure 2: Predefined paths of control tests, towards and away from the radiation source, and position of the three dosimeters.

Figure 3: Predefined path of the measurement performed in the four enclosures, towards the radiation source, and positions of the dosimeters. The length of the test area within the first and second enclosures, 12 m, is shown.

Figure 4: CDFs of the results of the three PEMs in different positions. Results are shown 1 m away, worn by the user in LoS, and worn by the user in NLoS for both predefined paths-towards and away from the radiation source.

Figure 5: Experimental data obtained in the 63 m³ first enclosure. Data are shown for (a) vertical and (b) horizontal polarization, with and without body influence, with a transmission power of 100 mW. The data are shown in function of the number of samples logged by the PEM while the user is walking towards the source. The results are shown in a semi-logarithmic scale.

Figure 6: Experimental data obtained in the 162 m³ second enclosure. Data are shown for (a) vertical and (b) horizontal polarization, with and without body influence, with a transmission power of 25 dBm (316.12 mW) and rescaled to 20 dBm (100 mW). The data are shown as a function of the number of samples logged by the PEM while the user is walking towards the source. The results are shown in a semi-logarithmic scale.

Figure 7: Measured and simulated levels of the E-field for vertical polarization. Levels are shown for the (a) first (63 m³), (b) second (162 m³), (c) third (57 m³), and (d) fourth (63 m³) enclosures. The levels are shown as a function of the percentage of the ICNIRP exposure limit of 61 V/m for the general population and for the 2.4 GHz band. The data are shown as a function of the number of samples logged by the PEM while the user is walking towards the source.

Table 1: Electromagnetic parameters used in the simulation.

Table 2: Polarization factor for each enclosure, calculated as the relation between the means of non-shadow and shadow data. The sizes of the enclosures are indicated.

Table 3: Main statistical values of the experimental and simulated results in the four enclosures under test conditions for vertical and horizontal polarization. The sizes of the enclosures are indicated.