Over-the-air testing of LTE-A PHY layer in reverberation chamber

Abstract

Reliable, repeatable, and flexible testing is crucial for assessing system performance and ensuring the quality of communication. In reverberation chambers (RC), real-life propagation environments can be emulated by loading absorbers, facilitating controlled testing of the system. This work presents over-the-air (OTA) testing of the LTE-A PHY layer in the RC. We assessed the performance of the LTE-A link using key performance indicators (KPIs) such as error vector magnitude (EVM), bit error rate (BER), and signal-to-noise ratio (SNR) for varying transmitter (Tx) and receiver (Rx) gains. Additionally, we have compared the results both in an empty RC and when the RC was loaded with RF absorbers. We used software-defined radio (SDR) for OTA transmission of LTE-A frames. The measurement results indicate that loading the RC with RF absorbers improves EVM, BER, and SNR. We also quantified the performance of the LTE-A link by changing the position of RF absorbers. Results showed that loading absorbers yielded up to 72.8% improvement in EVM, and placing absorbers closer to Rx helped reduce the amount of multipath, resulting in better transmission performance.

Keywords

Reverberation chamber (RC)over-the-air (OTA)error vector magnitude (EVM)bit error rate (BER)signal-to-noise ratio (SNR)

1. Introduction

The characteristic of a real propagation environment is presented in a reverberation chamber (RC) due to its stochastic nature. An RC is a highly reflective and electrically large conductive enclosure that generates a rich scattering environment. RCs have been used for electromagnetic compatibility (EMC) testing, antenna efficiency measurement, and over-the-air (OTA) testing.

OTA measurements are performed in RC to determine the noise figure of a wireless system [1], adjacent channel leakage power ratio [2], total isotropic sensitivity, antenna efficiency [3], total radiated power, and coherence bandwidth [4,5], spurious emission levels for 5G devices [6]. Studies [7–9] showed that real-world environments can be created in the RCs by loading absorbers. Root-mean-square (RMS) delay spread can be changed through the use of absorbers, and this loading has an impact on received signal quality. The performance of a real fifth-generation (5G) base station was studied by using a reverberation chamber in [10], and a real-life propagating environment was created using reconfigurable intelligent surfaces and with the addition of a large number of absorbing materials. The impact of chamber configuration on the quality of a wireless communication channel was assessed in [11] through bit error rate (BER) measurements conducted inside the reverberation chamber, considering different loadings, symbol rates, and paddle speeds. The signal quality can also be determined by measuring error vector magnitude (EVM). In [12], a comparison was made between the error vector magnitude (EVM) measurements obtained through the direct link, anechoic chamber (AC), and RC methods. The results showed that with the reduction of the multipath effect, the measured EVM approaches the ideal values measured in AC. Study [13] investigated the impact of transmitter and receiver separation distances and gains on EVM, BER, and signal-to-noise ratio (SNR) based on the OTA measurements conducted in a metal enclosure placed in an anechoic chamber. The impacts of the absorber loadings on the delay spread and EVM in the RC were studied in [14], and it was found that heavy loading leads to a better modulation quality. Although there are many studies on the determination of channel characteristics through RC measurements, to the best of the authors’ knowledge, no work has evaluated the impact of different propagation environments and varying gains on EVM, BER, and SNR.

Therefore, this study aims to fill this gap by conducting OTA measurements of LTE-A PHY and evaluating the results. LTE-A standard provides enhanced capacity and coverage and lowers delays. It also supports various types of low-power nodes, including microcells, picocells, femtocells, and relays [15]. LTE relays are easy to set up, and enhanced coverage with low interference can be provided at a low cost. Considering these advantages, this study adopts the LTE-A PHY Layer standard, which still occupies a large market share in the commercial field. Since RC measurements are repeatable, stable, and allow for the creation of different propagating environments, we performed EVM, BER, and SNR measurements in the RC to assess the system’s performance under varying conditions. The frequency response of the RC changed with the addition of RF absorbers, and we compared the performance of the RC measurements with and without RF absorbers. Furthermore, we determined the effect of the position of absorbers on the performance indicators.

The rest of this paper is organized as follows. The key performance indicators (KPIs) and measurement setup are described in Section 2, measurement results are given in Section 3, and the paper is concluded in Section 4.

2. Material and method

2.1. Key performance indicators

The error vector magnitude is a system-level performance metric that quantifies signal distortion of digitally modulated signals [16]. EVM is defined as the root-mean-square (RMS) value of the difference between a collection of measured symbols and ideal symbols [17] and calculated as given in (1). $\begin{eqnarray}\displaystyle \text{RMS EVM}=\sqrt{{\displaystyle \frac{\frac{1}{N}\mathop{\sum }_{k=1}^{N}((I_{k}-\tilde{I} _{k})^{2}+(Q_{k}-\tilde{Q}_{k})^{2})}{\frac{1}{N}\mathop{\sum }_{k=1}^{N}({I_{k}}^{2}+{Q_{k}}^{2})}}} & & \displaystyle\end{eqnarray}$ (1) where I_k and Q_k are ideal values of in-phase and quadrature-phase components and are received symbols, and N is the number of symbols.

Bit Error Rate is a performance metric that describes the quality of the received signal. It is a measure of the number of bits that are received incorrectly compared to the total number of bits transmitted and can be calculated as in (2). In Eq. (2), N_e represents the number of bits received in error, and N_b is the total number of received bits. $\begin{eqnarray} \displaystyle \text{BER}=\frac{N_{e}}{N_{b}} & & \displaystyle \end{eqnarray}$ (2)

The signal-to-noise ratio is a common performance metric for assessing communication quality, representing the relative measure of signal power to noise power, as in (3). $\begin{eqnarray}\displaystyle \text{SNR}={\displaystyle \frac{\text{Signal power}}{\text{Noise power}}}={\displaystyle \frac{\frac{1}{N}\mathop{\sum }_{k=1}^{N}((I_{k})^{2}+(Q_{k})^{2})}{\frac{1}{N}\mathop{\sum }_{k=1}^{N}(|n_{I,k}|^{2}+|n_{Q,k}|^{2})}} & & \displaystyle\end{eqnarray}$ (3) where I_k and Q_k are amplitudes of in-phase and quadrature-phase components, n_{I, k} and n_{Q, k} are the in-phase and quadrature noise amplitudes [17]. The relationship between the mentioned KPIs is presented in [17], making it possible to predict or, in some cases, substitute EVM in place of BER or even SNR. In this study, we utilize the relationship from [17] to estimate SNR using the RMS EVM value, as expressed in (4). $\begin{eqnarray} \displaystyle \text{SNR}\approx \frac{1}{\text{RMS EVM}^{2}} & & \displaystyle \end{eqnarray}$ (4)

2.2. Channel estimation

In order to mitigate radio channel variations and enhance the wireless communication systems, channel estimation is essential. In LTE, cell-specific reference signals (pilot symbols) are inserted in both time and frequency. These pilot symbols are utilized to estimate the channel response between each transmit and receive antenna. The estimation process involves several steps: first, pilot signals are extracted from the received grid. Then, the least-squares method is employed to calculate the channel response at the pilot symbol positions within the received grid, as given: $\begin{eqnarray}\displaystyle \tilde{H}_{P}(k)=\frac{Y_{P}(k)}{X_{P}(k)} & & \displaystyle\end{eqnarray}$ (5) where, Y_P(k) is the received pilot symbols, X_P(k) is the transmitted pilot symbol values [18].

The effects of noise on the channel estimates are reduced by averaging the least square estimates over an averaging window. Subsequently, time-frequency averaging is applied to estimate the channel response between the available pilot symbols. Finally, the complete channel response is obtained from (5) using appropriate interpolation techniques.

2.3. Measurement setup

The OTA testing of the long-term evolution advanced physical (LTE-A PHY) layer is performed in a large reverberation chamber at George Green Institute for Electromagnetics Research (GGIEMR), The University of Nottingham [19], under different loading conditions. A software-defined radio testbed is used to transmit standardized LTE-A frames for digital image transmission. We used two PlutoSDRs (Analog Devices, ADALM PLUTO, SDR Active Learning Module) for the transmission and reception of LTE-A frames. A schematic overview of the measurement setup is shown in Fig. 1a, which illustrates an electronically controlled mod-stirrer and two PlutoSDRs placed on transparent to electromagnetic wave polystyrene foam blocks. Figure 1b illustrates an outer view of the measurement setup, while Fig. 1c shows a picture of the measurement setup inside the empty RC. The inner dimensions of the chamber are 4.90 m × 3.64 m × 2.55 m. Two PlutoSDRs were connected to the same host PC, and one was used as a transmitter (Tx) and the other one as a receiver (Rx). The Tx and Rx were fixed on polystyrene foam blocks. The height of Tx and Rx from the base of the RC is 80 cm, and the distance between them is 170 cm.

Fig. 1.

The RC measurement setup (a) schematic overview (b) an outer view (c) empty (d) middle-loaded (e) corner-loaded.

For all measurements, the same image file is transmitted using a PlutoSDR. Another PlutoSDR receives the transmitted signal. The image file is formatted as a PNG (Portable Network Graphics) and has dimensions of 512 × 384 pixels. It is encoded with RGB channels, resulting in a total of 24 bits per pixel. Pixel values are stored as unsigned 8-bit integers (uint8). Subsequently, the size of the image to be transmitted is scaled down to 256 × 192 in order to reduce the number of frame transmissions. LTE-A baseband waveform generation and processing were performed using MATLAB’s LTE-A Toolbox. Received waveforms are processed using an LTE-A baseband receiver, which performs important signal processing and compensation for RF hardware impairment. First, the number of frames of the transmitted LTE signal is captured. Then, time synchronization and frequency synchronization are used to align the received symbols with the transmitter’s symbols, compensating for timing and frequency offsets introduced by the channel. Channel estimation techniques are applied to estimate the characteristics of the wireless channel, which are used to optimize subsequent signal processing algorithms. Interference mitigation techniques are applied to mitigate the effects of co-channel interference and noise on the received signal, improving overall system performance. The transmitted data from the received signals is extracted by decoding the Physical Downlink Shared Channel (PDSCH) and Downlink Shared Channel (DL-SCH). A signal with a 10 MHz bandwidth, a center frequency of 5 GHz, and a 64QAM modulation scheme is used in measurements. RMS EVM, BER, and SNR are calculated for gains ranging from −20 dB to 0 dB with an increment of 5 dB for Tx and from 20 dB to 60 dB with an increment of 2 dB for Rx. These measurements were repeated in the presence of RF absorbers with dimensions of approximately 87 cm × 64 cm (117 cones in total) at two locations: one in the middle and one in the left corner of the RC, as shown in Figs 1d and 1e.

3. Measurement results

We conducted far-field measurements in both empty and loaded RC scenarios, using PlutoSDRs for varying Tx/Rx gains. The empty RC represents a rich scattering environment, and its quality factor can be tuned by adding absorbers. Therefore, we were able to perform repeated measurements several times in a stable RC environment. We evaluated the impact of loading on the RMS EVM, BER, and SNR and conducted a performance comparison between the empty and loaded measurement scenarios.

Figure 2a shows the RMS EVM in an empty RC for different Tx/Rx gains. As seen from Fig. 2a, the RMS EVM ranges between 85.73% and 81.47%, even for higher Tx gains in the case of the empty RC. It can be clearly concluded from the figure that, under a rich scattering environment condition, the received constellation points deviate from their ideal position, resulting in a higher RMS EVM.

Fig. 2.

The RMS EVM versus Rx gain in the (a) empty (b) middle-loaded (c) corner-loaded RC.

Figure 2b depicts the RMS EVM under the scenario when RF absorbers were placed in the middle of the RC. In the case of loading absorbers in the middle of the RC, the RMS EVM is significantly improved, especially for 0 dB Tx gain. However, there was not much change in the RMS EVM levels at lower Tx gains.

Figure 2c shows the RMS EVM when RF absorbers were placed in the corner of the RC. As with the absorbers placed in the middle, the RMS EVM is also reduced with the loading of absorbers in the corner of the RC. This improvement is more pronounced at higher Tx gains, but the effect is less significant at lower Tx gains. Additionally, positioning absorbers near the Rx enabled achieving lower RMS EVM values even with lower Rx gains. For example, the RMS EVM at 30 dB Rx gain is 81.69% in the case of the empty RC. With the addition of absorbers in the middle, the RMS EVM is reduced to 44.86%, while this value is 11.82% when the absorbers are placed in the corner. The overall evaluation of the results shows that adding absorbers has a significant impact on RMS EVM, and these findings align with those obtained in [14].

The BER curves for empty and loaded RC scenarios are illustrated in Fig. 3. As seen in Fig. 3a, for the empty RC, very high BER values are obtained due to a rich scattering environment, similar to the EVM results. The mean BER values are 0.2275, 0.1952, 0.1771, 0.1689, and 0.1691 at selected gains ranging from −20 dB to 0 dB, respectively. The minimum BER is obtained as 0.1607 at −5 dB Tx and 52 dB Rx gain, while the maximum is 0.3994 at −20 dB Tx and 20 dB Rx gain.

Fig. 3.

The BER versus Rx gain in the (a) empty (b) middle-loaded (c) corner-loaded RC.

Figures 3b and 3c depict the BER values for the scenario in which RF absorbers were placed in the middle and corner of the RC, respectively. As seen in the figures, BER values are significantly affected by the loading of absorbers. Placing absorbers close to the Rx helps reduce the destructive effects of frequency selectivity, resulting in better transmission performance. For instance, at 0 dB Tx gain and 30 dB Rx gain, the BER value is 0.0164. With absorbers added to the middle, it decreases to 0.0067, and it reaches 0 when absorbers are placed in the corner.

The SNR values estimated for both the empty and loaded RC are depicted in Fig. 4. Similar to RMS EVM and BER values, the addition of absorbers also has an effect on SNR values. The increments in SNR values are significant, especially when absorbers are placed near the Rx. For instance, the ratios between the maximum SNR values in the middle-loaded and corner-loaded cases, compared to the empty RC case, are 7.96 and 11.98, respectively. As seen from the figures, the SNR values are 1.76 dB for the empty case, 6.96 dB for the middle loaded case, and 18.55 dB for the corner loaded case, all at 0 dB Tx and 30 dB Rx gain.

Fig. 4.

The SNR versus Rx gain in the (a) empty (b) middle-loaded (c) corner-loaded RC.

In order to further assess the impact of absorber loading on performance metrics, the minimum, and maximum values are tabulated and presented in Table 1. As can be concluded from the table, loading the RC significantly decreases the EVM; this improvement reaches up to 72.8%, particularly in the corner-loaded case. These enhancements are also evident in the BER and SNR values. The addition of absorbers helped to receive symbols almost without error at certain gains and yielded a 19.57 dB improvement in SNR.

Table 1

Statistics of the KPIs

	RMS EVM (%)		BER		SNR (dB)
Scenario	Min	Max	Min	Max	Min	Max
Empty	81.47	85.73	0.1607	0.3994	1.34	1.78
Middle-loaded	19.55	86.15	0.0	0.4386	1.29	14.17
Corner-loaded	8.59	84.88	0.0	0.4366	1.42	21.35

The RMS EVM, BER, and SNR values are averaged over Tx gain, and the statistical evaluations of these averaged values for the three loading scenarios are illustrated in Fig. 5 as a box plot. In the figures, the red lines indicate the median and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The red crosses denote the outliers, and the black diamonds show the mean values. Figure 5a displays the box plot for the averaged RMS EVM values, while Figs 5b and 5c show the averaged BER and SNR values for all loading cases.

Fig. 5.

Box plots for averaged (a) RMS EVM (b) BER (c) SNR.

As seen from Fig. 5a, the averaged RMS EVM values are very high for the case with no absorbers, with a mean value of 82.24%. However, with the addition of absorbers, the averaged RMS EVM values decrease significantly, with mean values of 66.34% and 48.27% for the middle-loaded and corner-loaded cases, respectively. Additionally, the figure clearly shows the difference in median values for different cases of loading. The box plot for the corner-loaded case is much longer than the others, indicating that the RMS EVM values are more spread out, and the Tx gain has a greater effect on the RMS EVM value. Conversely, changing the Tx gain has no significant effect on RMS EVM for the empty RC case. The results demonstrate that loading positively affects the characteristics of the channel and contributes to improving the performance of wireless communication systems.

Figure 5b displays the averaged BER values for different loading cases. As observed from the figure, the averaged BER values correlate with the RMS EVM values. Outliers are present in all cases, and the mean BER value for the no-absorber case is very close to the upper quartile. In the absence of absorbers, regardless of the Tx gain, the received image was not recovered well. However, absorber loading resulted in an almost perfect recovery of the received image. The maximum averaged BER values are 0.2772, 0.3050, and 0.2975 for the empty, middle-loaded, and corner-loaded cases, respectively, while the minimum values are 0.1677, 0.0290, and 0.0172.

The averaged SNR values for all loading cases are shown in Fig. 5c. As with the RMS EVM values, outliers are present only for the empty RC case, and loading has improved the SNR values as expected. The SNR values vary within a larger range for the corner-loaded case. The mean values of the averaged SNR values are 1.69 dB, 4.24 dB, and 9.43 dB for the empty, middle-loaded, and corner-loaded cases, respectively.

In order to perform further evaluation on the averaged values and assess whether the means of these values are equal or if there is significant variation between them, a one-way analysis of variance (ANOVA) test is applied. The ANOVA yielded highly significant F values of 67.09 (p < 0.001) for RMS EVM, 12.58 (p < 0.001) for BER, and 70.89 (p < 0.001) for SNR values, indicating substantial variation among the measurement scenarios. These results suggest significant differences in the means of the groups, and the loading of the RC has a strong impact on the quality of the system.

Fig. 6.

Channel state information for (a) empty (b) middle-loaded (c) corner-loaded RC.

The channel state information (CSI) for the empty, middle-loaded, and corner-loaded cases at 0 dB Tx and 30 dB Rx gain is illustrated in Fig. 6a, b, and c, respectively. The CSI estimation is obtained using LS estimation, time-frequency averaging, and cubic interpolation. Figures show that the channel exhibits high-frequency selectivity for the empty RC case, but this selectivity is reduced with the addition of RF absorbers, resulting in improved link performance. The empty RC represents a rich scattering environment, and loading the RC with absorbers affects the RMS delay spread and, thus, the coherence bandwidth of the wireless channel. The RMS delay spread, and the coherence bandwidth is inversely proportional to each other. Loading of the RC reduces RC’s spatial uniformity, resulting in a decrease in RMS delay spread and, consequently, an increase in the coherence bandwidth of the wireless channel. If the channel coherence bandwidth is larger than the transmitted signal bandwidth, flat fading occurs. On the other hand, if the channel bandwidth is narrower than the signal bandwidth, the channel is considered frequency-selective. The study [11] demonstrates that the RMS delay spread exhibits a significant frequency dependence in the absence of absorbers, whereas this frequency dependence diminishes with the increasing number of absorbers. Therefore, the frequency selectivity of the channel can be reduced by loading absorbers. These results support the findings in [20] that the multipath degree can be tuned by varying the amount and the position of the loading. Therefore, different propagation environments can be emulated in the RC, allowing flexible and reliable testing.

Figure 7 depicts transmitted and received images for all measurement cases at 0 dB Tx and 60 dB Rx gain. In the case of the empty RC, errors were present in the received image, whereas, with the addition of absorbers, the images were obtained with quite low errors: 0.2147, 0.00013, and 2.12 × 10⁻⁵ for the empty, middle-loaded, and corner-loaded cases, respectively.

Fig. 7.

(a) The transmitted image (b) the received image in the empty RC (c) the received image in the middle-loaded RC (d) the received image in the corner-loaded RC.

4. Conclusions

In this study, OTA testing and end-to-end system performance of the LTE-A communication link are performed. We focused on the KPIs such as EVM, BER, and SNR. The measurements were conducted in both an empty RC and an RC loaded with RF absorbers. A comparison of the measurement results shows that adding RF absorbers to the RC has reduced the impact of a rich scattering environment on EVM and, consequently, on SNR and BER. Moreover, the performance was further improved when the absorbers were placed in the corner close to the Rx. By conducting controlled tests and measurements, this study provides valuable insights into the optimal placement of RF absorbers for reducing the impact of a rich scattering environment on signal quality. Furthermore, we showed that various types of real-world wireless propagation mediums could be simulated inside an RC, and the quality of a wireless communication channel can be adjusted to accurately replicate diverse environmental conditions and assess the performance of communication systems under realistic scenarios.

Footnotes

Acknowledgements

Begum Korunur Engiz’s study at The University of Nottingham, GGIEMR, was supported by TUBITAK 2219 International Postdoctoral Research Fellowship Programme under 1059B192100872.

References

Stek

Hubrechsen

Prinsloo

D.S.

and Johannsen

, Over-the-air noise figure characterization of mm-wave active integrated antennas using a reverberation chamber, IEEE Trans. Microw. Theory. Tech.71(3) (2023), 1093–1101.

Xue

Sun

Yang

and Chen

, Reverberation chamber techniques for determining the adjacent channel leakage power ratio of wireless devices, IEEE Trans. Microw. Theory. Tech. (2022), (early access), doi:10.1109/TMTT.2022.3230598.

Hubrechsen

, The effect of noise on reverberation-chamber measurements of antenna efficiency, IEEE Transactions on Antennas and Propagation69(12) (2021), 8744–8752, doi:10.1109/TAP.2021.3083822.

Huang

Xue

Ren

and Chen

, Signal and coherence bandwidth effects on total radiated power measurements of LTE devices in reverberation chambers, IEEE Trans. Instrum. Meas.71 (2022).

Horansky

R.D.

and Remley

K.A.

, Flexibility in over-the-air testing of receiver sensitivity with reverberation chambers, IET Microw. Antennas Propag.13(15) (2019), 2590–2597.

Catteau

, Measuring radiated spurious emissions from a 5G device in a reverberation chamber, in: 2022 International Symposium on Electromagnetic Compatibility–EMC Europe, 2022, pp. 42–46.

Chen

Kildal

P.S.

Orlenius

and Carlsson

, Channel sounding of loaded reverberation chamber for over-the-air testing of wireless devices: coherence bandwidth versus average mode bandwidth and delay spread, IEEE Antennas and Wireless Propagation Letters8 (2009), 678–681, doi:10.1109/LAWP.2009.2025149.

Chen

and Kildal

P.S.

, Theoretical derivation and measurements of the relationship between coherence bandwidth and RMS delay spread in reverberation chamber, in: 3rd European Conference on Antennas and Propagation, Berlin, Germany, 2009, pp. 2687–2690.

Rehammar

, Load optimization in reverberation chambers, in: 15th European Conference on Antennas and Propagation (EuCAP), Dusseldorf, Germany, 2021, pp. 1–4. doi:10.23919/EuCAP51087.2021.9410935.

10.

Micheli

Test of 5G system in the reverberation chamber at mm-wave, in: 17th European Conference on Antennas and Propagation (EuCAP), Florence, Italy, 2023, pp. 1–4. doi:10.23919/EuCAP57121.2023.10133586.

11.

Genender

Holloway

C.L.

Remley

K.A.

Ladbury

J.M.

Koepke

and Garbe

, Simulating the multipath channel with a reverberation chamber: application to bit error rate measurements, IEEE Transactions on Electromagnetic Compatibility52(4) (2010), 766–777, doi:10.1109/TEMC.2010.2044578.

12.

Fang

Xia

Xing

and Xu

, A comparative study of error vector magnitude measurement in a reverberation chamber, IEEE Antennas Wirel. Propag. Lett.21(12) (2022), 2547–2551.

13.

Korunur Engiz

Lodro

Greedy

and Thomas

D.W.P.

, OTA testing of near-field and far-field communication links, Electronics Letters59(19) (2023), e12967.

14.

Zhou

, Impacts of absorber loadings on simulating the multipath channel in a reverberation chamber, International Journal of Antennas and Propagation2017 (2017), 3078103, 7 pages, doi:10.1155/2017/3078103.

15.

Ghosh

Ratasuk

Mondal

Mangalvedhe

and Thomas

, LTE-advanced: next-generation wireless broadband technology, IEEE Wireless Communications17(3) (2010), 10–22, doi:10.1109/MWC.2010.5490974.

16.

Acar

, How error vector magnitude (EVM) measurement improves your system-level performance, Analog Devices Tech. Art. (2021), 1–6. Available from: https://www.analog.com/en/technical-articles/how-evm-measurement-improves-system-level-performance.html.

17.

Shafik

R.A.

Rahman

M.S.

and Islam

A.R.

, On the extended relationships among EVM, BER and SNR as performance metrics, in: 2006 International Conference on Electrical and Computer Engineering, Dhaka, Bangladesh, 2006, pp. 408–411. doi:10.1109/ICECE.2006.355657.

18.

Engiz

B.K.

Kurnaz

and Sezgin

, Approach for determining the optimum pilot placement in orthogonal frequency division multiplexing systems, IET Communications2015(15) 1915–1923, doi:10.1049/iet-com.2015.0104.

19.

GGIEMR. Available from: https://www.nottingham.ac.uk/research/groups/ggiemr/facilities.aspx (accessed on 20.02.2024).

20.

Primiani

V.M.

, Reverberation chambers for testing wireless devices and systems, IEEE Electromagnetic Compatibility Magazine9(2) (2020), 45–55, 2nd Quarter, doi:10.1109/MEMC.2020.9133241.