Power Density Case Study for 5G mmWave Array Antennas

As 5G millimeter wave (mmWave) wireless device involves some new technologies, such as beamforming, the radiofrequency (RF) exposure compliance test for the 5G mmWave wireless device is significantly complicated. In order to shorten the compliance period for 5G mmWave terminals, the relevant regulatory authorities recommend a combination of numerical simulation and measurements to demonstrate compliance. To verify the feasibility of this method, the RF exposure test conducted in this paper was a reverse procedure according to IEEE (the Institute of Electrical and Electronics Engineers) standards. First, actual measurements under various conditions, including different beam configurations, different test distances, different input power levels, different duty cycle, and nonpeak directions, were performed, and the changing trend of PD over testing conditions was analyzed. Then one dual-polarized patch antenna array was selected for simulation analysis. The feasibility of the method proposed in IEEE standards was proved through the comparison of the results experiment and numerical analysis.


Introduction
The 3rd Generation Partnership Project (3gpp) protocol divides the 5G frequency spectrum into two FR (Frequency Range). FR1 is a low-frequency band up to 7125 MHz, and FR2 ranges from 24.25 GHz to 52.6 GHz. Most of the electromagnetic wavelengths in this FR2 are in the millimeter level. As mmWave technology can achieve a faster speed and lesser extent latency than 4G network, 5G wireless device has attracted massive attention in recent years. In the meantime, it also has raised the public concern regarding the biological effects on the human body resulted from the exposure to such high-frequency electromagnetic field (EMF). To control the effect in a safe range, standard organizations such as the International Commission for the Protection of Non-Ionizing Radiation (ICNIRP) and IEEE have established the permissible exposure limits for electromagnetic radiation [1].
Specific absorption rate (SAR) and power density (PD) are usually adopted to determine exposure compliance. SAR defines the amount of energy absorbed by human tissue per mass unit, and it is currently preferred in exposure com-pliance determination. However, when higher frequencies are involved, the skin depth decreases, and energy absorption is usually limited to the skin surface. Studies have shown that when the frequencies are at 10 GHz-100 GHz, about 90% of the energy will be absorbed by the skin's epidermis or dermis [2]. In such circumstance, it is difficult to build a meaningful volume for SAR. The mmWave devices operating at such high-frequency range should be evaluated with power density, as it is expressed per unit area for near-field RF exposure evaluation, instead of per unit mass in SAR. Whereas power density measurement is a time-consuming procedure; therefore, combining simulation and measurement has been proposed as an efficient method during a compliance test.

Device under Test
The device used in this testing is a prototype board that operates at 28 GHz, and the bandwidth is 200 MHz. There are five 5G antenna arrays at different sides of the device as shown in Figure 1. The details of the device antenna are as follows: (1) One 2 × 2 patch antenna array, dual-polarized, is located at the left corner of DUT. Beam  The antenna arrays cannot work simultaneously. While one antenna is active, the rest of the arrays will be disabled. As for beamforming, the beam configuration of an antenna array can be changed by software.

Power Density Assessment
The power density testing employs the mmWave module of cDASY6 from SPEAG, Switzerland. The mmWave   2 Wireless Communications and Mobile Computing module supports a measurement at distances as small as 2 mm and to automatically determine the optimal measurement grid based on the frequency modulation, evaluation distance, and extent of the transmitter array. The phantom of this mmWave module approximates free-space conditions and allows evaluating not only the antenna side of the device under test (DUT) but also any opposite-radiating side of the devices operating above 10 GHz.
The probe used in the measurement is an E-field mmWave probe. It can perform at the frequency range from 750 MHz to 110 GHz. According to the current draft of IEC/IEEE 63195, for general near-field measurements, the magnitudes and phases of both E-filed and H-field are necessary to correctly determine the power density. As an E-field probe is adopted during measurement, the field reconstruction is used to derive the H-field from the direct measurement of E-field. All components of the E-field, the H-field,

Test Settings
During the power density measurement, the mmWave signal in the DUT is configured by an engineering testing software, and the test mode is nonsignaling. This testing is implemented in 5 different situations: (1) The device is operated to transmit continuously at 100% transmission duty with different antenna beams. The test distance is 5 mm, and the input power is specified for each antenna array (2) For certain beams, 3 different test distances are set from 10 mm to 20 mm with 5 mm increments. The

Measurement Results
The PD shown as follows is computed under the averaging area of 1 cm 2 and 4 cm 2 . As the antenna arrays operate at 28 GHz, according to ICNIRP RF Guidelines, the data of concern is averaged over 4 cm 2 in this paper.  Most of the beams of the single-polarized antenna array at the left edge are below the exposure limits, and the maximum power density is 25.4 W/m 2 found from beam 20 (as shown in Figure 4).
Results in Figure 5 demonstrate that the PD of the antenna at the top edge does not exceed the current ICNIRP basic limit of 20 W/m 2 at 4 cm 2 .  The results shown in Figure 6 are measured from the single-polarized antenna array at the right edge. Out of the total five beams, three beams are larger than 20 W/m 2 . The largest value is measured from beam 27 at 34.3 W/m 2 . Figure 7 is the measured results of the 1 × 4 patch antenna array beams. They indicate that beam 30-33 comply with the RF radiation exposure limits of the ICNIRP.

Power Density Measurement at Different Test Distances.
Data from Figures 8-15 show that the power density decreases as the testing distance increases from 5 mm to 20 mm. For test distance at 10 mm, the PD of beam 3, beam 20, and beam 27 complies with the limits set by ICNIRP. Figures 9 and 11 demonstrate that the PD of the two normal beams of the 2 × 2 dual-polarized patch antenna array is still over the basic restriction even at the 15 mm test distance.   Figures 16 and 17 show that the measurement values go down when the input power levels are reduced by -3 dB and -5 dB, respectively, for both beam 14 and 23. The power density of beam 14 is finally below 20 W/m 2 at 4 cm 2 as the input power is decreased from -1 dB to -4 dB.

Power Density Measurements under Different Duty Cycle.
As shown in Figure 18, there is a significant trend in the measurement result change under different duty cycle, i.e., when the duty cycle is 0.6, the measured power density is approxi-mately 60% of that at its maximum duty cycle, and the same is true when the duty cycle is 20%.

Power Density Measurements at the Nonpeak Direction.
Beam 20 is generated by the single-polarized dipole antenna located at the left corner. Figure 19 shows that the power density at the back (nonpeak direction) is close to the highest simulated power density at the left edge, while the power density of beam 23 generated by the left corner antenna at the back (nonpeak direction) is approximately equal to the highest power density at the top edge (as shown in Figure 20).

Wireless Communications and Mobile Computing
Beam 2, beam 6, and beam 8 are generated by 2 × 2 patch antenna array that is located at the back. As can be seen from Figures 21-23, the measured power density of each beam at the nonpeak directions is much less than that on the major radiation direction.

Simulation and Measurement
Results Comparison 6.1. Simulation Environment. For providing case study for power density analysis, we perform simulation and  To explore the consistency of simulation and measurement, the simulated results of 2 × 2 patch antenna array will be discussed.

Power Density Simulation
Results. In order to save computational resources, the DUT model is truncated and only the 2 × 2 patch antenna array located at the left corner is modeled and simulated in this paper.
The detailed simulation results of the selected 2 × 2 patch antenna array beams are shown in Table 1. The DUT is modeled in free-space and the same configurations as the (1) testing condition mentioned in chapter 4, where the averaging area is 4 cm 2 and distance is 5 mm from the back side.
The simulation results have been normalized to the same PA output power level for comparison with the measurement   11 Wireless Communications and Mobile Computing results. The insertion loss difference of each panel has also been considered for the simulation and measurement results normalization.

Analysis
(1) The deviations between PD simulation and measurement results of the selected 2 × 2 patch antenna array beams were verified to be within ±2.4 dB (2) From simulation and measurement results, we could check the correlation of power density variation. The PD trends between simulation and measurement may not ideal one-to-one correspondence. But the top four highest beams from PD simulation at most can still cover the worst case of PD measurement Figure 24 shows the cross-sectional power density plot of the normal and offset beams. The PD measurement results show that the normal beam (e.g., beam 5) is the worst case in the panel. While in the simulation, the PD simulation result of the offset beam (e.g., beam 4) is slightly higher than the normal beam.
In the near field simulation region, there may be a case where the offset beam PD is slightly higher than the normal beam due to the influence of the shell and the close distance. The different array elements may occur in the beam aggregation region. It is probably because the truncated DUT model is used, and the surrounding shells are not accurately modeled.

Conclusion
For mmWave array antennas, when practical, the combination of numerical simulation and measurements to demonstrate the power density compliance is a practicable assessment approach and a considerable reduction in the measurement period. As (1) from power density measurement results of multiple factors (different beam configurations, test positions, input power levels, duty cycles, main lobe and side lobe beam directions, etc.), we could check the correlation of power density variation. It would be a good reference when developing IEC standard (2) from power density simulation and measurement results comparison, it is worth to consider the power density results from the simulation in term of showing compliance

Data Availability
The measurement and simulation data used to support the findings of this study are included within the article.

Conflicts of Interest
The authors declare that they have no conflicts of interest.