Report on an international interlaboratory comparison of new radio MIMO over-the-air system performance using project developed validated measurement methods (up to 30 GHz) for multi probe anechoic chamber (MPAC), radiated two stage (RTS) and reverberation chamber (RC) + channel emulator (CE), where applicable, for sub-6 GHz and mm wave bands taking into account ETSI TR 38.827 requirements

21NRM03 MEWS Deliverable D1 : Report on an international interlaboratory comparison of new radio MIMO over-the-air system performance using project developed validated measurement methods (up to 30 GHz) for multi probe anechoic chamber (MPAC), radiated two stage (RTS) and reverberation chamber (RC) + channel emulator (CE), where applicable, for sub-6 GHz and mm wave bands taking into account ETSI TR 38.827 requirements Organisation name of the lead participant for the deliverable: Aalborg University Authors: Ivan Bonev, Fengchun Zhang, Martin Forsberg, Fredrik Harrysson, Yunsong Gui, Jerdvisanop Chakarothai and Tian Hong Loh Due date of the deliverable: 30 September 2025 Actual submission date of the deliverable: 30 November 2025 Confidentiality Status: PU - Public, fully open (remember to deposit public deliverables in a trusted repository). 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. Deliverable Cover Sheet 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. Abbreviations Acronym Description AAU Active Antenna Unit AC Anechoic Chamber ADC Analog-to-Digital Converter AoA Angles of Arrival ATF Antenna Test Function BLER Block Error Rate BSE Base Station Emulator BS Base Station BW BandWidth B5G Beyond 5G CATR Compact Antenna Test Range CDF Cumulative Distribution Function CDL Clustered Delay Line CE Channel Emulator CFR Channel Frequency Response CIR Channel Impulse Response CRC Cyclic Redundancy Check DAC Digital-to-Analog Converter DMRS Demodulation Reference Signal Dir-VAA Directional Antenna-Based VAA DSP Digital Signal Processing DUT Device under Test EM Electromagnetic ETSI European Telecommunications Standards Institute EVM Error Vector Magnitude FF Far-Field FIR Finite Impulse Response FPGA Field Programmable Gate Array FSPL Free-Space Path Loss GCM Geometric Channel Modeling GO Geometric Optics HFC High-Frequency Convertors HPBW Half Power Beamwidth IFFT Inverse Fast Fourier Transform InO Indoor Office IQ In-phase and quadrature JADE Joint Angle Delay Estimation LF Low Frequency LO Local Oscillator LoS Line of Sight LTE Long Term Evolution MIMO Multiple-input-multiple-output 1 Acronym Description mmWave Millimeter Wave frequency band MPAC Multi-Probe Anechoic Chamber NF Near-Field NLOS Non-Light of Sight NR New Radio OTA Over-the-Air PADP Power Angle Delay Profile PAS Power Angular Spectrum PBCH Physical Broadcast Channel PDCCH Physical Downlink Control Channel PDP Power Delay Profile PDSCH Physical Downlink Shared Channel PFS Prefaded Signal Synthesis PL Path Loss PSP PAS Similarity Percentage PWG Plane Wave Generator RBs Resource Blocks (RBs) RTS Radiated Two Stage RT Ray Tracing RC Reverberation Chamber RF Radio Frequency RSRP Reference Signal Received Power Rx Receiver SISO Single-Input Single-Output SNR Signal to Noise Ratio SSB Synchronization Signal Block SS-RSARP Synchronization Signal Reference Signal Antenna Relative Phase SS-RSRPB Synchronization Signal Reference Signal Received Power per branch Sub-THz Sub-Terahertz band TDL Tapped Delay Line TDL-B Tapped Delay Line B THz-TDS THz Time-Domain Spectrometer Tx Transmitter UCA Uniform Circular Array ULA Uniform Linear Array Umi Urban Microcellular UE User Equipment VAA Virtual Antenna Array VST Vector Signal Transceiver VNA Vector Network Analyzer WP Work Package 3GPP Third Generation Partnership Project 4G 4th Generation Wireless Systems 5G 5th Generation Wireless Systems 6G 6th Generation Wireless Systems 2 Contents Abbreviations.......................................................... 1 Executivesummary....................................................... 5 1.Introduction ......................................................... 6 2 Wireless cable testing for MIMO radios: A compact and cost-effective 5G radio performance test solution . . 7 2.1Introduction...................................................... 7 2.2Problemstatement.................................................. 7 2.3Proposedsetup.................................................... 9 2.4Activethroughputmeasurements......................................... 10 3. A Generalized Wireless Cable Over-the-Air Testing Framework for Arbitrary Field Emulation in Nonanechoic RadioEnvironment.................................................... 13 3.1Introduction...................................................... 13 3.2 Proposed generalized wireless cable framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.3Measurementvalidation............................................... 15 3.4Measurementresults................................................. 17 3.5Conclusion....................................................... 20 4. Test Zone Validation for FR2 MPAC Using Directional Antenna Based Virtual Array . . . . . . . . . . . . . . . . 21 4.1Introduction...................................................... 21 4.2Proposedmethod .................................................. 21 4.3Simulationresults .................................................. 25 4.4Measurementvalidations.............................................. 26 4.5Analysisoftheresults................................................ 27 4.6Conclusion....................................................... 29 5. Dynamic Sub-THz Radio Channel Emulation: Principle, Challenges, and Experimental Validation . . . . . . . 30 5.1Introduction...................................................... 30 5.2BasicprincipleofadigitalCE............................................ 31 5.3 Challenges and enabling solutions for dynamic sub-THz radio channel emulation . . . . . . . . . . . . 32 5.4RTsimulation..................................................... 33 5.5 Sub-THz CE experimental measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 5.6Conclusion....................................................... 38 6.RadiatedTwoStage(RTS)Method............................................. 39 6.1Introduction...................................................... 39 6.2 SDR based 5G NR demodulation testbed development . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 6.3Testbedsetup..................................................... 41 6.4Subsystemverification................................................ 43 6.5Linkperformance................................................... 47 6.6Conclusion....................................................... 48 7. RMS delay spread reduction in reverberation chambers by channel model delay tap removal . . . . . . . . . 49 7.1Introduction ...................................................... 49 3 7.2Background ...................................................... 49 7.3Measurementsetups................................................. 52 7.4 Method to reduce RMS delay spread in RCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 7.5Conclusion....................................................... 54 8. Theoretical study on keyhole reduction methods in reverberation chambers . . . . . . . . . . . . . . . . . . . 55 8.1Introduction...................................................... 55 8.2Background...................................................... 55 8.3Conclusionsandfuturework............................................ 57 9. Reverberation Chamber plus Channel Emulation (RC+CE) Method . . . . . . . . . . . . . . . . . . . . . . . . . . 58 9.1Introduction...................................................... 58 9.2Two-StepClosedLoopMethod........................................... 59 9.3Subsystemverification................................................ 63 9.4MeasurementResults ................................................ 67 9.5LinkPerformance................................................... 69 9.6Conclusion....................................................... 70 10.Intercomparisons ..................................................... 71 10.1 Comparison of PDP from UMa channel model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 10.2 Intercomparison between RTS and RC+CE methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 11.Conclusion.......................................................... 79 4 Executive summary This report is the sixth deliverable D1 of the MEWS project, covering the topics included in Work Package 1 (WP1) and it merges the knowledge of the consortium on practical, traceable over-the-air (OTA) metrology for 5G/6G, from FR1/FR2 into sub-THz. The overarching aim is to reduce test cost and time while maintaining accuracy, uncertainty control, and standards alignment (Third Generation Partnership Project(3GPP)/European Telecommunications Standards Institute (ETSI)). These principles guide the methods and validations presented here. The deliverable is organized as follows: Chapter 1 is the introduction part, outlining objectives, the 3GPP/ETSI standards context, and the need for metrology that is both rigorous and deployable. Chapter 2 contains the wirelesscable multiple-input multiple-output (MIMO) test solution, while in Chapter 3 a generalized OTA field-emulation framework is presented, including calibration, implementation variants, and conducted-vs-OTA throughput results. Chapter 4 describes a directional-antenna virtual array that improves SNR, allowing wider element spacing with fewer samples, and validating emulated channels with PAS/Power Delay Profile (PDP)/AOA metrics and an efficient post-processing search. Chapter 5 presents a dynamic sub-THz channel emulation, focusing on details principles, challenges and enabling techniques. Chapter 6 describes a real-world validation of the radiated two-stage (RTS) method for the link performance testing in FR2. Chapter 7 proposes reducing RMS delay spread in reverberation chambers (RC) by removing far-end clustered taps in the emulated channel. Chapter 8 studies theoretically the capacity/rank loss when CE Rayleigh fading is compounded by RC fading. Chapter 9 presents the design of the RC-plus-CE method and its testbed implementation for single-input-single-output (SISO) and MIMO configurations in both FR1 and FR2 bands, along with performancevalidation. Chapter10 summarizesinter-laboratorycomparisonstosupporttraceabilityandreproducibility across sites, aligning measurement procedures for broader acceptance. The measurement campaigns encompass two key comparisons: (i) synthesized channel characteristics between MPAC and RC with absorber loading, and (ii) link performance testing between RTS and RC+CE. Finally, conclusions are made in Chapter 11, synthesizing findings and outlining next steps toward standardized, scalable OTA metrology for emerging wireless systems. 5 1. Introduction The global shift from fourth-generation (4G) to fifth-generation (5G), and soon towards sixth-generation (6G) mobile network systems, represents more than just an increase in data rates, it signifies a transformation in how wireless networks are conceived, deployed, and used. Future systems will operate across vastly higher frequency ranges (mmWave and sub-terahertz (sub-THz)), leverage massive antenna arrays, rely on artificial intelligence and edge computing, and serve mission-critical applications requiring extreme precision in timing and reliability [1], [2]. This technological leap introduces unprecedented complexity in the design, testing, and validation of wireless systems. Measurement science or metrology is essential to navigate these challenges, offering the traceable benchmarks necessary for innovation, standardization, regulation, and industrial deployment. However, existing metrology infrastructure and methodologies are not fully equipped to handle these emerging demands. The primary aim of the presented work in this document is to develop metrological methods that are not only scientifically rigorous, but also practical from an industrial and regulatory standpoint, i.e. minimizing cost and test time while maintaining measurement accuracy and repeatability. A key focus was placed on achieving traceability and uncertainty quantification in accordance with international standards and guidelines. The developed methods aligned with and extended relevant specifications such as [3], [4] and [5]. Thesemetrologicaladvancementssupportbothmultiple-input-multiple-output(MIMO)over-the-air(OTA)Radio Frequency (RF) conformance testing (e.g., effective isotropic radiated power (EIRP) / effective isotropic sensitivity (EIS) validation, spurious emissions, and receiver sensitivity) and end-to-end system performance assessments, including throughput, beam management, andmobility handlingunder realisticsignal conditions. Whererelevant, measurement procedures were validated through inter-laboratory comparisons and collaborative test campaigns to demonstrate repeatability and reproducibility across facilities. More specifically, the main objectives of the deliverable can be summarized as follows: • A compactand cost-effective wirelesscable testsolution has beenpresentedand analyzed. Further, a generalized wireless cable OTA testing framework for arbitrary field emulation in nonanechoic radio environment has been developed. • Test Zone Validation for frequency range 2 (FR2) multi-probe anechoic chamber (MPAC) Using Directional Antenna Based Virtual Array • Dynamic Sub-THz Radio Channel Emulation: Principle, Challenges, and Experimental Validation • Real-world validation of the Radiated Two Stage (RTS) method for the link performance testing in FR2. • Real-world validation of the two different reverberation chamber (RC) based approaches for the 5G new radio (NR) link performance testing. • Inter-comparison between multiple link performance OTA testing methods. 6 2 Wireless cable testing for MIMO radios: A compact and cost-effective 5G radio performance test solution 2.1 Introduction Adevicemayperformpoorlyor notworkatallwithoutextensivetesting. Therefore,testingisarequiredstepduringthe stageofresearch,development,andproduction. Animportantstepcouldbethevalidationofthe5Gradioperformance in realistic propagation scenarios under laboratory conditions. In the cases of 4G long-term evolution (LTE) and new radio (NR) 5G technologies in the frequency range (FR) 1 (covering the frequency range from 410 to 7125 MHz), cable tests have been used dominantly to evaluate performance. For that purpose, RF coaxial cables are used as an interface between the testing instrument ports and the device under test (DUT) antenna ports. Thus, the needed test signals can be guided without distortion to the respective antenna ports of the DUT. Owingtothe lackof an RFconnectorforcabletestingpurposes in integratedtransceiverradiodesigns forhigher frequency 5G bands at FR2 (from 24.25 to 52.6 GHz) and beyond 5G (B5G) systems, cabled measurements are no longer practical. Therefore,OTAradiatedtestingcanbeseenasnecessarilysolution[6]. Toemulatetheradiochannelbetween the transmitter (Tx) and the receiver (Rx), radio channel emulators (CEs) have been used under laboratory conditions in both cabled setup and OTA setup solutions [7], [8], [9]. The wireless (virtual) cable solution aims to achieve identical functionality to the cable test solution, i.e. testing signals guided to the respective antenna ports with high isolation between DUT antenna ports. Recently, the wireless (virtual)cable solutionforMIMO deviceperformance testinghas attractedlotsof attentionfrom theindustry, academia and among the standardization bodies [10], [11], [12], [9], [13], [14]. Compensation of the transfer matrix between the OTA probes and the antenna ports of the DUT is done with the use of radio CEs in existing wireless cable solutions. However, use of the radio CEs for such a purpose would lead to waste of valuable CE resources and to expensive systems. Employing more probe antennas is a known and effective technique to improve the robustness and quality of the wireless cable connection. However, such a solution would further increase the cost of system implementation. This subchapter based on [15] proposes a novel wireless cable implementation, which employes a stand-alone programmable amplitude and phase control network. The presented solution is simple and cost-effective. The proposed scheme is demonstrated to work in practice by performing validation measurements obtained in the vector network analyzer (VNA)-based passive measurement setup and in the active measurement setup as well. In [15], the analogy between test solutions is discussed. 2.2 Problem statement Cabled setup and wireless cable setup are shown in Figure 1. 7 (a) (b) Figure 1: System diagram for the a) cabled setup and b) wireless cable setup for a N × M MIMO systems (with M ≥ N) Signal model derivation is presented for both setups (refer to equations (1)-(3) in [15]). x(f, t), y(f, t), s(f, t), and n(f, t) are the transmit signal vector at the base station (BS) antenna ports (Mx1), the received signal vector (Nx1), the desired testing signal vector (Nx1), and the noise vector at the DUT antenna ports (Nx1), respectively. For the cabled setup (Figure 1a), the assumption that the desired testing signal is guided to the respective DUT antenna port without any distortion can be made, i. e. y( f, t) = s( f, t). Aand Gare the unknown transfer matrix between the Kprobe antenna ports and NDUT antenna ports, and the calibration matrix, respectively. The scheme of cable setup (Figure1b) includes a BSemulator, a radio CE,multiple probe antennasplacedin an RF-shielded enclosure, and a DUT. Thenumber of probe antenna ports Kmust not be smaller than the number of the DUT antenna ports N(i.e., K≥N). Therefore, in order to achieve the same functionality of the wireless cable solution as of that of the cable solution setup, the simplified problem is to design matrix G such as |AG |=I, where I(NxM) is the identity matrix. CE is an important unit in the cabled setup for MIMO performance testing. As it is clear from Figure 1a, the required CE resource in the cabled setup includes M input interface and N output interface ports, and N × M digital fading channels for the communication link. As it is discussed in [12] and [9], the achieved isolation between the wireless cables determines the quality of the wireless cable connection. The quality and robustness of the wireless cable are ruled by the condition number of the transfer matrix A. The latter is defined by the probe antenna, multipath propagation within the RF-shielded enclosure, and DUT antennas. Usually, the DUT is considered as a black box, because the locations and radiation patterns of DUT elements are not known. Therefore, the lack of knowledge of DUT and propagation environment makes the control the condition number of the matrix Aimpossible. By using more 8 Figure 9: A diagram of the proposed generalized wireless cable framework The received field vector at the DUT antenna ports rN×1can be written as: rN×1=AN×K.GK×N.pN×1(3-1) rN×1∈CN×1isthereceivedsignalvectorattheNDUTantennaports. AN×KisthetransfermatrixfromtheKprobe antenna ports to the N DUT antenna ports. AN×Kencompasses the effects of the probe antennas, the DUT antennas, and the multipath propagation channel between the probe array and the DUT antenna array. This MIMO matrix can be directly measured using an on-off procedure, during which only one probe antenna and one DUT antenna are activated at a time, while all other probes and DUT antennas remain inactive. By following this method, we can capture the chamber matrix that describes the relationship between the probe antenna ports and the DUT antenna ports. The calibration matrix GK×N∈CK×Ncan be obtained using the pseudo-inverse operation: GK×N= (AN×K)†·IN×N(3-2) where IN×Nis the identity matrix. It is important to note that once the matrix Gis established, the physical testing environment should remain constant. The effectiveness of the proposed framework relies on the stability of the scattering from the environment, with minimal in-band interference affecting the setup. Additionally, time-varying fading in the propagation environment can be simulated by adjusting the target field vector p, while keeping Gstatic. The generation of the vector pN×1is determined by the characteristics of the target field. In contrast to PWG and CATR methods, which can only simulate line-of-sight (LOS) paths, the proposed framework can replicate arbitrary target fields. For more information, the reader can consider equation (2) and Fig. 3 in [23]. 3.3 Measurement validation Two measurement setups were utilized in the validation campaign, as shown in Figure 10. The x-axis is oriented along the boresight direction of the DUT antennas and is directed toward the probes. The positive x-axis points directly at the probes, while the y-axis is oriented perpendicular to the ground, with the positive y-axis extending vertically upward. The AoA θ is measured from 0° at the positive x-axis and increases in a clockwise direction. In Setup I, the reconstructed fields at the four antenna ports of the DUT were directly measured using a VNA and compared to 15 target values to validate the proposed concept. In Setup II, the objective is to demonstrate the DUT’s beam-steering performance using the reconstructed fields. This setup incorporates four programmable phase shifters and a combiner connected to the testing antenna ports, simulating a beamformer. The phase shifters are adjusted according to the desired beam-steering angle. (a) I (b) II Figure 10: Illustration of the measurement setup a) I and b) II with the 2-D coordinate system The validation measurement system, illustrated in Figure 11, consists of the following components: a VNA, two uniform linear arrays (ULA), a four-way RF power splitter (used on the OTA probe side) along with four digital phase shifters and four attenuators (creating the desired field vector for the DUT ports), four digital phase shifters and a four-way RF powersplitter(employedtosimulatea beam-steeringcapableDUT). TheVNA generatesa continuouswave frequency signal at 3.55 GHz and the two ULA with decoupled patch antenna elements serve as OTA probes and as the DUT antenna arrays. The measurement was conducted at a range of 10.5 cm, with the spacing between ULA elements set at 0.5 wavelengths (λ) at 3.55 GHz. The two ULA were arranged facing each other during the measurements. Due to constraints related to available phase shifters and attenuators, only the four central antenna elements from both ULA were employed in the measurements. It is important to note that the 10.5 cm measurement range is significantly smaller than the far-field (FF) distance for the four-element ULA, which is 38 cm at 3.55 GHz. 16 Figure 11: Photo of the measurement setup in the laboratory condition The initial phase of the measurement campaign involved recording and calibrating the four RF paths associated with the probe antennas. This step was crucial to eliminate any discrepancies among the RF chains using a programmable phase shifter and attenuator. Following this calibration, the transfer matrix Awas directly measured in ”on-off” mode using a VNA. It is important to note that 50 Ω loads were added to properly terminate the ”off” links during the ”on-off” measurement, which helped to minimize potential port reflections. For ”Setup I”, two measurement scenarios were conducted: 1. A plane wave was directed at a ULA with AoA at 5°, 15°, 25°, and 50°. The ULA was comprised of four isotropic antennas spaced 0.5λ apart, which was designated as the DUT. This DUT antenna configuration can be loaded. 2. Another plane wave was directed at the same ULA with AoA values of 5°and 25°. In this scenario, a faulty antenna element was incorporated into the ULA, serving as the DUT (the 3rd DUT element for the 5°condition and the 2nd DUT element for the 25°condition, respectively). It is noteworthy that the ”faulty element” was simulated in the target field while all physical connections remained intact and functional. For ”Setup II”, a plane wave was considered to be impinging on the ULA at an AoA of 0°. This ULA also consisted of four isotropic antennas spaced 0.5λ apart. At the DUT side, the phase shifters were adjusted to steer the beam direction from -30°to +30°in 10°increments, resulting in a total of seven beam-steering directions. The peak response was anticipated only when the beam-steering directions configured by the DUT aligned with the 0°direction of the plane wave generated by the arbitrary wave emulator. Ultimately, the accuracy of the field emulation was assessed by comparing the effects observed at each of the NDUT antenna ports with the target field vector pN×1and the measured received signal vector r′ N×1. 3.4 Measurement results The results of the measured complex field at the four DUT antenna ports, subjected to a plane wave with varying impinging angles, are presented in Figure 12. 17 (a) (b) (c) (d) Figure 12: The measured fields at the four DUT ports for a plane wave with impinging angle a) 5°, b) 15°, c) 25°and d) 50° There is a remarkable correlation in both amplitude and phase for all the considered angles of incidence, showing deviations between the emulated and target fields of up to ±0.75 dB and ±1.5°for an impinging angle of 5°, ±0.8 dB and ±6°for 15°, ±0.7 dB and ±6°for 25°, and ±0.25 dB and ±2.5°for 50°, respectively. The measured phase and power for the first DUT element are normalized against the target values. These results clearly illustrate that a plane wave with any given impinging angle can be accurately reproduced at the DUT antenna ports. The observed deviations may stem from uncertainties associated with the programmable shifter and attenuator, quantization errors in implementing the equivalent field vector, and non-ideal conditions in the measurement environment. In the context of non-stationary testing, our measured results strongly align with the theoretical predictions, effectively validating our proposed scheme, as demonstrated in Figure 13. 18 (a) I (b) II Figure13: ThemeasuredfieldsatthefourDUTports(witha faultyDUT element)foraplanewavewithvariousimpinging angle The amplitude and phase deviation between the emulated and target fields for functioning elements is within ±0.9 dB in amplitude and ±0.25° in phase for an impinging angle of 5°. For an impinging angle of 25°, the deviations are ±0.15 dB in amplitude and ±1° in phase, respectively. In the scenario with faulty elements, an isolation of over 25 dB can be achieved to simulate the malfunctioning antenna elements for both cases (i.e., when ”DUT element 3 is off” and ”DUT element 2 is off”). It is important to note that isolation is practically limited by the coupling between the DUT elements. Additionally, the phase deviation is larger for the faulty element since its output power is significantly weaker, which is an expected condition. For the proposed measurement setup II, the measured DUT beam-steering pattern is presented in Figure 14. Figure 14: The measured and target beamforming power pattern of the ULA under the reconstructed plane wave path with 0°impinging angle ThebeamformingpatternoftheDUTalignscloselywiththetargetpatterninthereconstructedfield. Anoticeable peak at 0° indicates that this angle corresponds to the direction of the target plane wave. As the beam-steering angle diverges from the target path direction, there is a significant drop in power, which is as expected. The measured results confirm that the DUT performed similarly to its testing conditions under a physical plane wave field. 19 3.5 Conclusion This subchapter presented a generalized wireless cable framework designed to emulate arbitrary fields at the antenna ports of the DUT. The framework was experimentally demonstrated in an open laboratory environment using plane waves with varying angles of incidence, ranging from 5°to 50°, for a DUT equipped with four antennas and a short measurement distance. The framework also successfully demonstrated non-stationary fields, achieving a difference of more than 25 dB between the “faulty” and functional DUT elements. Furthermore, a plane wave with a 0°AoA was emulated, enabling the four-element DUT to perform beam-steering operations in the presence of this emulated field. The measured results exhibited excellent agreement with the targeted results. 20 4. Test Zone Validation for FR2 MPAC Using Directional Antenna Based Virtual Array 4.1 Introduction Validatingand testingthe performanceof 5G radiodevicesprior toroll-outsiscrucial as thecommercializationprocess of 5G picks up in speed [25], [18]. The traditional method of conducting performance testing involves bypassing the DUT antennas and using RF cables to guide the test signals directly to the antenna ports of the DUT [26], [27]. The FR2 offers extensive spectrum sources for 5G applications [28]. However, because of the highly integrated transceiver design of the 5G NR devices, OTA testing techniques have become an unavoidable trend for performance testing [29], [30], [31] and [32]. In the available literature, the proposed OTA techniques includes wireless cable solution [33], [12], [34], reverberation chamber (RC) [35], and 3-D MPAC [20], [31], [32]. The 3GPP has chosen the 3-D MPAC method as the reference standard methodology for FR2 MIMO capable devices among a variety of OTA testing techniques. However, prior to actual measurements, the fidelity of the emulated spatial channels in a 3-D MPAC setup must be assessed and verified. Assessing the target channel models’ ability to be reconstructed in the test zone is the goal of test zone validation. The traditional technique for verifying a 3-D MPAC setup’s channel emulation performance relies on a virtual antenna array (VAA) made up of omnidirectional antennas. Validating the emulated channels for FR2 3-D MPAC is difficult, though, for a number of reasons. A low SNR in the measurements due to the high path loss (PL) in the FR2 frequency band could introduce errors into the estimated parameters of emulated channels. Furthermore, the large number of measurement samples needed in the test zone to accurately estimate the 3-D channel spatial profiles may necessitate a lengthy measurement period. Additionally, in order to evaluate the emulated channels using various metrics, such as PDP, AoA, or power angle delay profile (PADP), power spectra in both the angular and delay domains are needed. This subchapter is based on the work [36]. In the latter, the authors suggest a test zone validation technique for the FR2 3-D MPAC configuration that is based on a virtual array made up of directional antennas. The suggested method requires fewer measurement samples (i.e., virtual array elements) in the test zone and can produce a higher SNR in the measurements when compared to the current method. The post-processing technique for estimating the reconstructed channel spatial profiles in the test zone is an extended sequential search algorithm. Since a directional element pattern is advantageous for suppressing both the grating and sidelobes of the virtual array beam pattern, a larger element spacing greater than half wavelength could be used to decrease the number of measurement samples. 4.2 Proposed method As previously stated, the MPAC methodology is a common approach for testing the end-to-end performance of MIMO devices OTA [37]. In order to test 4G or 5G MIMO devices under real-world propagation conditions, multiple probes are used to reconstruct spatial fading channels defined in 3GPP 38.901 [38] with the aid of radio CE. As shown in Figure 15a, the several probes are linked to a fading CE and turned on simultaneously. In order to recreate the target channel conditionsinthetestzone,thedifferentprobesemitfadingchannelprofilesthataresuitablyconstructed. Aspreviously mentioned, a test zone validation measurementshould beused to assesshow well thetarget channelenvironmenthas been replicated within the test zone prior to carrying out the actual testing. Data collection and signal post-processing are usually the two primary steps in the test zone validation process. The measurement setup for data acquisition is depicted in Figure 15b, where a mechanical positioner is used to move a test antenna (also known as a calibration antenna in some works) to various spatial locations to form the virtual array. The VNA records the channel frequency 21 (a) (b) Figure 15: a) 3-D MPAC setup for MIMO OTA performance testing b) Measurement system for test zone validation in 3-D MPAC response (CFR) at each site. The post-processing process extracts the necessary evaluation metrics and the channel characteristics from the measured data. MIMO performance testing can begin after the target channel environment inside the test zone has been verified. More precisely, the following is a summary of the comprehensive steps involved in test zone validation with the suggested approach. 1) In accordance with the virtual array configuration, record the CFRs for every channel snapshot at every predetermined location. 2) Using the extended sequential search algorithm and the information gathered in step 1, estimate the channel characteristics (i.e., AoAs, path powers, and PDP from each probe). 3) Determine the channel’s spatial profile by assessing metrics (such as Power Angular Spectrum (PAS) and PAS Similarity Percentage (PSP)). The signal model for the signals received in step 1 is provided in Section II-B in [36]. The extended sequential search algorithm, which is used to determine the channel characteristics based on the CFRs that the virtual array receives, is explained in Section II-C in [36]. The reader can find the detailed explanation in the aforementioned sections. AssumethatthePdirectionalantennaelementsthatmakeupthe3-Dvirtualarrayinsidethetestzoneradiate in a radial direction. Since directional antenna patterns can suppress the array pattern’s grating lobes, the element 22 spacing between adjacent elements can be chosen to be greater than λ/2 [39]. In the MPAC setup, N channel snapshots stepped in the CE are used to synthesize the desired channels using Q OTA probes. The OTA probes’ configuration cannot be altered because they are a part of the standardized MPAC setup design. The purpose of the current work is to verify the channel environment radiated by the OTA probes. Keep in mind that the suggested approach can, in theory, be used with any 3-D virtual array configuration. Figure 16: Diagram of the locations of OTA probes and the 3-D virtual array for test zone validation Figure 16 provides an example of where the virtual array and OTA probes are located. Although our work can be directly extended fordual-polarized scenarios, the antenna polarization is disregarded for simplicity’s sake. Equations (1) to (5) in [36] define the CFRs received by P elements. All of the OTA probes are situated in the mid-field region of the DUT and radiate toward the test zone center because the measurement range, or the distance between the OTA probes and the DUT, is set at 0.75 m in the standard for cost-saving reasons. Since the view angle differences between any two virtual array elements and a given OTA probe are negligible, the antenna gain of any OTA probe can be roughly regarded as a constant value for virtual array elements. Figure 16 shows an example of the transmission between the fifth OTA probe and the p-th element. In order to assess the channel environment for test zone validation in FR2 3-D MPAC, the joint angle delay power profiles of the emulated channels are necessary. MUSIC is a high-angle estimation resolution algorithm that is based on subspace. Separating the spatial covariance matrix into signal and noise subspaces and calculating the AoAs based on the orthogonality between the signal and noise subspaces is the fundamental concept. Only AoAs’ information is gathered, though. Joint angle delay estimation (JADE) could also be accomplished by extending the traditional MUSIC algorithm, which is known as the JADE-MUSIC algorithm [40]. By stacking the signals that the array elements receive at all frequencies into a high-dimension vector, this algorithm takes advantage of the channel’s spatial and temporal characteristics. Joint angle delay power information can be obtained by performing the eigen decomposition on the covariance matrix of that high-dimension vector. However, the direct 3-D brute-force search in the elevation angle domain, the azimuth angle domain, and the delay domain, as well as the large matrix dimension of the covariance 23 matrix, result in an extremely high algorithm complexity. The implemented algorithm is an extension of the N twostage sequential search algorithm proposed in [41]. Equations (9) - (11) in [36] propose steering vector solution for estimation of AoAs. By using classical beamforming to steer the array beam to the target probe direction, the CFR and channel impulse response (CIR) at each probe direction observed by the 3-D virtual array can be determined based on the estimated AoAs and the angular location of the OTA probes. However, when the target probe and some of the other probes are in close proximity, the signals from other OTA probe directions may reduce the estimation accuracy. In order to lessen the interference from the other probe directions, nulling operations are implemented. More precisely, nulls at the other five probe directions are simultaneously formed whenever the virtual array’s main beam is directed to one target probe direction. In general, the required angular resolution and the angle range of the spatial multipath profile determine the 3-D virtual array configuration. For arbitrary 3-D spatial profiles, a sphere array that fills the entire space might be a good general option. The necessary spatial resolution determines the sphere’s radius and the array’s total number of elements. It is possible to optimize the 3-D virtual array configuration for a particular 3-D channel model in order to balance the number of measurement samples (i.e., the number of antenna elements in the virtual array) and the estimation accuracy. The measurement time is determined by the quantity of measurement samples. As a result, it’s critical to correctly arrange the array’s components so that it can offer adequate angular resolution with the fewest possible measurement samples. Clustered Delay Line C (CDL-C) urban microcellular (UMi) and indoor office (InO) channel models are typically defined by 3GPP as reference multipath channels for performance testing in NR FR2 MIMO OTA testing. The 3-D MPAC systeminthisinstancehassixdual-polarizedOTAprobespositionedonasectorthatisatleast0.75metersfromthetest zone’s center, while the test zone’s radius is 0.1 meters [42]. The OTA probes are situated in the elevation and horizontal planes at angles of 5°and 25°, respectively. The 3-D virtual array can be specifically created to attain adequate spatial resolution with fewer virtual array elements for the purpose of validating the reference channels. In Section III, the precise locations of the probe antennas and virtual array elements will be provided. Because of the high directivity of the element pattern, a directional-antenna-based virtual array can have element spacing greater than 0.5. However, when the element spacing is set too large and the directive element pattern is no longer able to adequately suppress the grating lobes, it would be challenging to achieve good nulling results and accurate AoA estimation. In theory, the virtual array can be created using a variety of directional antenna types. For practical measurements, however, directional antennas with stable phase centers and constant complex radiation patterns across the entire measurement frequencies are more advantageous. Additionally, for a path with a specific incident angle, the directional antennas with a relatively wide beamwidth will increase the 3-D virtual array’s effective array elements. The same virtual array configuration will be used to increase the angular resolution with more efficient array elements. Corrugated horn antennas [39], which are frequently used as feeds for reflector antennas and anechoic chambers, are a good example of a suitable directional antenna. To attain the same angular resolution as directional antennas with wide beamwidth, the comparatively narrow beamwidth directional antennas may need a larger array aperture, i.e., a larger radius for Uniform Circular Array (UCA) and more virtual array elements. On the other hand, narrow beamwidth directional antennas typically have higher antenna gain, which can result in better SNR for mmWave measurements. An unstable phase center within the frequency bands of interest, difficulties aligning polarization and main beam direction, higher costs for directional antennas with consistent antenna gain, and continuous maintenance are some 24 • Covering topics like carrier frequency, system BW, CE resource, channel models, and interface between testing devices and sub-THz radios, the essential requirements for sub-THz channel emulation are identified. • The frequency extension scheme for downand up-converting signals from/to sub-THz bands, the bandstitching concept to improve the emulation BW, the tap resource optimization scheme to lower the required CE resource, the site-specific dynamic channel modeling taking blockage and channel transition into consideration, and the wireless cable concept for establishing an OTA connection are some of the enabling solutions that are applicable for sub-THz radios. • Using a commercial CE platform, an experimental demonstration of fading channel replay for measured subTHz channels is made. In the experimental setup, dynamic channels were simulated and verified using a sub-THz radio that could beam-steer and align. 5.2 Basic principle of a digital CE A CE is specialized testing apparatus (hardware) made to simulate radio wave propagation. A simulation platform for simulating propagation channels and producing related CIRs is the channel simulator (e.g., geometry-based stochastic channel, 3GPP 38.901) or RT (usually software). In other words, channel emulation simulates real-world conditions by physically simulating radio channels in a lab setting. A standard channel simulator, RT simulation, or channel measurements are some of the methods that can be used to generate the CIRs loaded in the CE (implemented with the finite impulse response (FIR) in the baseband). For radio performance verification, the CE offers the ability to establish repeatable and controllable propagation conditions, which is frequently difficult to accomplish through field testing. Figure 19 shows an example of the digital CE. Figure 19: Diagram of a sub-THz CE for single-input single-output (SISO) transmission, with the band stitching scheme using N sub-channels Digital-to-analog and analog-to-digital converters (ADC) are represented by the symbols ADC and digital to analog converter (DAC) in this figure. These devices are then digitally processed, more precisely convolved, with the timevariant multipath channel profiles in real-time using digital signal processing (DSP) units. After processing, the signal is coupled into the Rx after being upconverted in the CE. Notably, probe antennas are used to transmit and receive testing signals to the CE OTA in sub-THz frequency ranges. Because they may implement flexible and standard multipath channels based on FIR filter structure, digital CEs based on tapped delay line (TDL) structure are widely used. 31 Keep in mind that bi-directional emulation would require twice the fading channel resources, whereas a MIMO CE with N input ports and M output ports would require N × M fading channels. 5.3 Challenges and enabling solutions for dynamic sub-THz radio channel emulation The challenges and enabling solutions have been defined in details in [49]. They will be discussed briefly here. They are: • Extensiontosub-THz carrierfrequencyband: Usingamixertoup-convertanddown-convertthesignalfrequency, existing commercial digital CEs, which were first created for sub-6GHz applications, can now cover mmWave frequency bands [8]. Frequency extension modules, which are frequently utilized in sub-THz channel sounders [44], are utilized for a wider frequency range, as illustrated in Figure 19. By appropriately adjusting the local oscillator (LO) frequency, the IF in the CE, and the multiplier in the frequency extender, the RF chain’s sub-THz carrier frequency can be freely designed. • Scheme for enhancing BW: Sub-THz BW will be used in vast quantities. The costly ADCs and DACs in current commercial CEs, which are intended for testing of current systems like 4G LTE, 5G NR have restricted system BW. In [50], a mmWave CE enabled by a software-defined radio platform achieved an instantaneous BW of 3 GHz. Recent works [11] and [47] showed how successful the band-stitching approach is. • CE resource optimization: The delay tap resource is practically constrained for commercial digital CEs. Due to computational constraints in the DSP units, the number of delay taps per channel is decreased when additional fadingchannelsareneeded. Furthermore,thesamplingfrequencyofcommercialCEsdeterminesthediscreteness ofthedelays. However,thenumberofdelaytapsandpositionsinpracticalpropagationchannelscanbearbitrary. As a result, it is essential to optimize the CE’s tap resource, including the number of delay taps, their positions, and the complex coefficients that are assigned. The CE will be able to replicate the target channel characteristics as closely as feasible thanks to this improvement. • OTA interface connection: In conventional cabled configurations, RF coaxial cables are used to route testing signals from testing instrument ports (such as CE output or input interfaces) to the DUT antenna ports, avoiding the need of integrated DUT antennas. For sub-6 GHz radios, where testing via reserved antenna connectors is practical,thistechniquehasbeenthemostpopular. However,theusageofextremelylossy,costly,andbrittlesubTHz connectors is not feasible for sub-THz radios, which are distinguished by their tiny and integrated designs. The predominant technique will therefore be OTA radiated testing, which uses sub-THz integrated antennas as direct interfaces for signal transmission and receiving. • Models of radio channels that are appropriate for simulating sub-THz: Based on the idea of channel clusters— groups of unresolvable multi-path components—geometry-based stochastic channel modeling has gained popularity in 4G and 5G standardization. Because 4G and 5G radios have limited latency and spatial resolutions, clustering makes modeling for channels with rich multipath components easier. It might not work as well with sub-THz radios, though. There are few dominant pathways in the sparse, specular sub-THz propagation channels. Highly directional antennas are used by sub-THz radios to further filter channels spatially, producing even more sparse characteristics. Furthermore, because of their large BW and electrical apertures, sub-THz devices provide excellent latency and spatial resolution. 32 5.4 RT simulation Due to slow mechanical positioning and orientation systems, recent channel measurements employing both frequency domain [44] and time-domain channel sounders [51] have concentrated on sub-THz propagation channels, usually in static situations. Dynamic spatial channel measurements are an open problem, because to the complexity and expense of real-time sub-THz channel sounders. Another method is RT simulation, which models the propagation of electromagnetic (EM) waves as they interact with materials, surfaces, and objects in complicated environments by applying the concepts of geometric optics (GO). It enables the analysis and simulation of multiple scenarios with varying configurations and environmental factors, including the location of the antenna, the characteristics of the material, and the existence of obstructions. By modifying parameter values, RT also provides flexibility for examining dynamic channel properties. For 6G [45], RT shows potential in modeling sub-THz propagation characteristics. As previously mentioned, beam-steerable and high gain antennas are essential for keeping constant alignment with the dominant propagation channel, guaranteeing good connectivity even in spite of changing transceiver placements and environmental conditions. Dynamic channels caused by the mobility of communication ends and objects must be taken into account when evaluating realistic sub-THz communication links. Moreover, real-time steering towards the most prominent propagation path requires adaptive beamforming at both communication links. RT simulation is used to model a scenario going from non-LOS (NLOS) to LOS zones in order to illustrate dynamic channel emulation capabilities. Additionally, we used practical CEs with constrained BW and tap resources to simulate wideband dynamic channels by implementing band-stitching and CE resource optimization techniques. 33 (a) (b) (c) (d) (e) Figure 20: Dynamic fading channel emulation: (a) RT simulation scenario, (b) the CIRs of the target channels over the trajectory, the CIRs of the emulated channel over the trajectory by using (c) 24 taps, (d) 12 taps, and (e) 6 taps, respectively Two omnidirectional antennas were used as the Tx and Rx antennas in the RT simulation. They have a gain of 0 dB and a HPBW of 30° in the elevation. The height of both antennas was 1.25 meters. Figure 20a depicts the simulated scenario, in which the Rx antenna moves through 31 places in 0.5 m increments, moving from NLOS to LOS and back to NLOS zones, as indicated by the blue arrow, while the Tx antenna is at a fixed location. As seen in Figure 18a, the Rx locationsarenumbered 1 to31. The100 strongestpaths fromthe RTsimulation werechosenforeach locationand used as the target fading channels. The Wireless InSite 3D program was used to do the simulations. Furthermore, 3.5 GHz was selected as the RT’s simulation frequency. The CE’s performance is independent of the carrier frequency because all of its algorithms were implemented in the baseband. We chose the D-band at 140 GHz as the carrier frequency in the channel emulation simulation. As an example, we use four sub-bands, each spanning a BW of 135 MHz, to build a wideband CE with a BW of 500 MHz. The optimization used the measured system frequency response for each subchannel at 135 MHz in the CE bypass mode [8]. To simulate fading channels over the intended 500 MHz BW, we use three number of taps in the CE in the simulation: 24, 12, and 6. We use the tap selection and coefficient optimization strategy technique [52] for every sub-band. Theoretically, we can establish a single virtual digital channel with a total BW of 4 × 135 = 540 MHz by employing 4 sub-bands, each with a frequency of 135 MHz. To assess the quality of the emulated channels, Figure 20b illustrates the CIRs of the target dynamic channels over the trajectory (i.e., 31 locations). As the Rx moves from location 1 to location 31, it is evident that the CIR progressively changes. The channel conditions changing from NLOS (at places 1–7) to LOS (at positions 8–17) and back to NLOS (at locations 18–31) is the reason of this 34 variation. It is obvious that for the first arrival path at each Rx location, the NLOS conditions show greater delay and lower power than the LOS conditions. Figure 20c-Figure 20e show the emulated CIRs over the trajectory with 24, 12, and 6 taps, respectively. A single color bar for Figure 20c-Figure 20e indicates that the power levels of the target channels (Figure 20b) and emulated channels (Figure 20c)-Figure 20e) have the same dynamic range. Although the emulated and target CIRs for dominant paths closely match, there are considerable differences, particularly for weaker paths when the tap count is decreased. These differences demonstrate that emulated and target channels are not the same and are caused by restrictions in commercial CEs, such as discrete delays, distortion from band stitching, system BW limitations, and a limited number of delay taps because of DSP computational limitations. To reduce insertion delay and additional latency that would arise when switching between the frequency and time domains, commercial CEs often carry out channel emulation in the time domain rather than the frequency domain. As a result, it is crucial to optimize the CE’s delay taps, complex coefficients, and locations. Within the limitations imposed by commercial CEs, this optimization aims to correctly depict dominant paths while compromising in simulating weaker paths. The results of the simulation show that the tap selection and tap coefficient optimization method in [52] can successfully simulate dynamic fading channels over a targeted wide band stitched together from many sub-bands, even with a restricted number of taps. This is accomplished by using sub-band calibration in the digital channels of CEs for band stitching and fading channel emulation. 5.5 Sub-THz CE experimental measurements Experimentalmeasurementsconfirmedtheconceptofchannelemulationinthesub-THz range. Intheseinvestigations, acommercialCE operatingat sub-6GHz was usedtoreplicatefadingchannels overa wideBW.Figure21showsa picture of the measurement setup. Figure 21: The photo of the measurement setup ThemeasurementsetupconsistsofacommercialKeysightPropsimFS16(usedasCE),aKeysightVNAPNA-X(used as VNA) and a control PC (used to configure and control the CE for emulating specific channels). The validation of subTHz channel emulation included into account two different kinds of dynamic channel models: “blockage scenario” and “transitionscenario”. Asshown in Figure 22a (LOS)and Figure 22b (NLOS withLOS obstructedby thehuman blocker), the ”blockage scenario” uses static Tx and Rx locations and purposefully adds a ”human blocker” to simulate environment changes brought on by a moving object. Assuming isotropic antennas at both ends, RT simulation is used to extend 35 the original measurement-based CIR in order to model this situation. As explained in [53], the RT is simulated at 140 GHz, while the channel sounding measurement spans 140-144 GHz. When human blocker moving at 1 m/s and paths approach within 0.8 meters of the human blocker’s center, a blockage event takes place. By scaling individual path gains with a time-variant attenuation factor, the human blocker’s attenuation profile, which was derived from tests at 140 GHz within a 2 GHz BW as described in [54], is incorporated into double-directional data. (a) (b) (c) Figure 22: ”Blockage scenario”: (a) the RT simulation scenario in the LOS case (with the blocker at position 1), (b) the RT simulation scenario in the NLOS case (with the blocker at position 2) and (c) measured and target CIRs for both cases In order to ensure link stability, beam-steering is used at both communication ends to monitor and align with the strongest path. As a result, Figure 22a and Figure 22b, respectively, show the LOS and NLOS paths. When one of the link ends—the Tx or the Rx—moves from the NLOS region to the LOS region while the environment stays the same, this is known as the ”transition scenario” and the channel changes. In order to track the most prominent propagation path, beamforming is used at both communication ends. The primary objective is to simulate the Tx/Rx’s movement in a propagation scenario, where the LOS conditions change. 36 (a) (b) Figure 23: ”Transition scenario”: (a) an illustration of the RT simulation scenario and (b) the measured CIRs at 0 m, 1.04 m and 2.08 m As shown by the blue and orange dots in Figure 23a, respectively, double directional sub-THz channel measurements covering a frequency range from 110 to 170 GHz were carried out in the corridor scenario at two Rx locations, one in the NLOS and one in the LOS location in [55]. RT simulation was used to interpolate the double directional channels for the transitional areas (shown by the gray dots) between these two locations based on measurement data. A total of seven Rx locations with a 0.35 m separation distance and a 2.08 m length were represented in the RT simulation. The RT simulation produced three channel models, with the Rx antenna positioned at the left end (0 m), center (1.04 m), and right end (2.08 m), as shown in Figure 23 as an example. In Figure 22c, the target channel models are compared with the measured CIRs of the emulated channels in the ”blockage scenario” for both ”blocker” positions. ”Gain” in the figure denotes the channel gain in dB, which reflects the CIR’s amplitude. ”Measurements” indicates the CIRs of the channel that the CE emulated and measured using a VNA, while ”Model” refers to the CIRs derived from the RT simulation. We find observable differences in the channels at the two ”blocker” positions. Compared to the LOS example (position 1), the NLOS scenario (position 2) has fewer paths within the 40 dB dynamic range, and the paths are less dispersed. The blocking effect and the Rx antenna’s main beam direction change towards the strongest NLOS path are the two primary reasons of these variations. Multiple scattering and reflections occur when the signal deviates from a direct LOS path due to environmental factors and obstructions in NLOS situations. The CIRs of the target and emulated channels in Figure 22c were compared, and the dominating paths in both scenarios showed a satisfactory alignment between the target and emulated channels. There are some apparent delay variances for specific paths, mostly due to the discrete delays in the CE with a resolution of 5 ns. The LOS path shows a 47.1 ns delay, indicating that the Tx and the Rx are around 14.1 m apart. For three different locations, as shown in Figure 23b, we compare the measured results with the target models in the ”transition scenario”. For alllocations, there isa verygood alignment betweenthe target channeland theemulated channel that was measured. Some changes in the channel characteristics may be seen as the Rx antenna moves from the NLOS condition (at 0 m) to the LOS conditions (at 1.04 m and 2.08 m). In particular, when the direct LOS path takes dominance in the LOS conditions, the gain of the strongest path in the CIR rises. Overall, the observations show that, in both the blockage and transition scenarios, the emulated channels successfully replicate the characteristics of the target channels. The measured results demonstrate how well the emu37 lated channel models can replicate wireless propagation conditions, since they closely match the expected behaviors in real-world wireless communication scenarios. 5.6 Conclusion It is essential to assess wireless systems’ performance in real-life propagation channels. The requirements, difficulties, and solutions for dynamic sub-THz channel emulation in beam-steerable sub-THz radios are discussed in this subchapter. In order to satisfy the limited CE tap resource requirements and the wideband system BW, we additionally investigate ideas like band stitching and CE tap resource optimization. We experimented with a commercial CE to simulate sub-THz channel models with a 1.4 GHz BW, concentrating on beam-steerable sub-THz radio transitions and indoor human blocking scenarios. Our experiment results showed excellent emulation accuracy. 38 6. Radiated Two Stage (RTS) Method 6.1 Introduction The establishment of measurement traceability and the improvement of measurement accuracy will enable manufacturers to provide confidence in their specifications. This plays a key role in the customer/supplier relationships, for which products need to be demonstrated as ’meeting specification’, regardless of who is carrying out the test or when/where the test is being performed. With the industrial exploitation and adaptation of complex 5G and beyond mobile network NR signals and multi-antenna technologies, such as, MIMO, at different radio frequency bands in emerging wireless systems, the verification of emerging NR products has become both complex and time-consuming [56]. Furthermore, as wireless communication technologies continuously evolve, performance testing through cablebased measurement methods [57] becomes increasingly difficult, not only because the antenna cannot be separated from its RF front end, but more importantly, as the number of antennas increases, especially in the case of the massive MIMO, there is an unsustainable increase in the required measurement duration. Therefore, the trend in measurement of such systems has shifted toward OTA based methods. Efficient OTA testing measurements of 5G NR systems with many advanced features, is needed to efficiently and cost effectively assess complex new radio system performance and product conformance to specifications. International standard development bodies, such as the 3GPP [58] and ETSI [59] have been working on the OTA performance test metrics for MIMO wireless communication system, measurement methodologies, and validation procedure evaluation of 5G NR UE (e.g., 3GPP TR 38.827 [57] for MIMO 5G NR end-to-end performance). This plays an important role in ensuring the multi-antenna mobile UE meets the desired performance requirements as defined in the 3GPP protocol. Several OTA test methods were proposed originally in the sub-6GHz 4G mobile network LTE era, which includes the MPAC [57], radiationtwo-stage (RTS)[60], and reverberation chamber pluschannel emulation(RC+CE) [60]methods. In 3GPP TR 38.827 v16.8.0 (2022-09) [57], only the MPAC and RTS methods have been considered for 5G NR performance testing. The RTS method aims to address the core challenge of link performance measurement faced by the wireless communication system, i.e. mimicking the real-world propagation channel with its inherent special domain characteristics in a controllable measurement environment. When compared with the MPAC method, which reconstructs the multipath wireless channel in a fully anechoic chamber through an array of precisely coherent probes. The multiple duplicates of the multipath signals with different phases, amplitudes, and timing are generated by a sophisticated centralized RF signal generator instrument, distributed to and transmitted by the multiple probes typically located at the circumference of the circle surrounding the UE and combined to form the designed RF distribution at the location of the UE. The RC+CE method is a much more cost-effective solution due to its simpler measurement setup compared with MPAC and RTS methods, but has not been included in [57] due to its limited channel emulation capability. In recent years, the suitability of various test methods has been explored at different 5G NR frequency bands, including frequencyrange1 (FR1, 0.41 GHzto 7.125 GHz), frequency range2 (FR2, 24.25 GHzto 71.0 GHz)and frequency range 3 (FR3, 7.125 GHz to 24.25 GHz) bands for evaluating the OTA performance of 5G NR MIMO system. With millimeter-wave (mm-wave) MIMO antenna technologies increasingly emerging in 5G NR mobile devices, OTA characterization under the ideal far-field condition is very costly to realize since the large-scale phased arrays and beamforming technologies are likely used in the mm-wave frequency to counter the large path loss, which requires a large size of chamber to meet the far-field condition. Currently, for a 5G mm-wave UE, only a three-dimensional MPAC solution has been selected by 39 3GPP [56], due to limited validation of RTS methods at mm-wave frequencies. Although mm-wave channel models can be emulated by the RTS method in principle, the RTS method has been mainly validatedfor 2 × 2 MIMO mobile handsets in sub-6 GHz. Under current 3GPP standards, the RTS method is not yet fully applicable for NR systems operating with mm-wave frequency bands due to the lack of real measurement results. Additionally, the interest in MIMO near-field communications has grown significantly over recent years, especially towards 5G-advanced and 6G communications [61], [62], [63], [64] and [65]. Furthermore, there are higher expectations for near-field measurement than far-field measurement due to advantages, such as lower chamber realization cost, smaller footprint, and lower propagation losses [66], [67]. To verify whether the RTS method is applicable in a size-limited anechoic chamber for the mm-wave frequency band, this section presents a preliminary near-field evaluation of an RTS-based mm-wave 5G NR MIMO OTA test system newly developed at the UK National Physical Laboratory (NPL) [68]. In this section, we present our work through the following sub-sections. In Section 6.2 , we will briefly introduce the general RTS principle and our approach to address the challenge in its realization at mm-wave frequencies. In Section 6.3, we provide an overview of the measurement testbed and key details in the design. The subsystem verification and the measurement results are presented, respectively, in Sections 6.4 and 6.5. Section 6.6 concludes this section. 6.2 SDR based 5G NR demodulation testbed development For FR2, the RTS method has been extended from the conventional cable-based methods [5]. As its name implies, the RTS method consists of two stages of measurement. In the first stage, the RF signals at the receiver side are generated from a 5G NR base station emulator (BSE) and imposed a channel fading by a CE through the cable-based method. In the second stage, the RF signal of the receiver side is transmitted to the DUT through the OTA “wireless cable” method [13]. Unlike the MPAC method, the OTA step in the RTS method is only responsible for delivering the RF signals to the antenna array at the DUT in the wireless cable method. The merits of this kind of method include low complexities in testbed implementations and associated low uncertainties, as well as its less requirement for the anechoic chamber (AC) compared with the MPAC method. However, there are two key challenges to be addressed to evaluate the realworld MIMO performance of the UE. Firstly, the three-dimensional (3-D) antenna radiation pattern characteristics need to be measured and applied in the wireless channel emulator. Secondly, the crosstalk between multiple Tx probes at the testbed side and Rx antennas at the DUT side needs to be precisely measured and eliminated through the signal processing method to emulate a direct cable-based connection. Regarding measuring the antenna radiation pattern for the antenna arrays attached to UE, one significant problem to address is the lack of accessible antenna test function (ATF) within UE available. The ATF functions allow access to the reference signal received power per branch (RSRPB) and reference signal antenna relative phase (RSARP) which are defined in 3GPP TS 38.215 [69], but are optional functions for UE. Due to the scarcity of test mobile phones with fully supported ATF software UE based on a software-defined radio (SDR) system has been developed in this project, as depicted in Figure 24. Its details will be further described in Section 6.3. A particular RSRPB and RSARP measurement function block has been developed in the software UE to measure the received downlink power corresponding to each antenna and the phase differences between the reference antenna and the rest of the antenna elements, by exploring the synchronization signal block (SSB) signal, which is the downlink broadcast signal with a typical repetition period ranging from 20 to 160 milliseconds in the time domain and 20 resource blocks (RBs) in the frequency domain. 40 (a) (b) (c) (d) Figure 31: Frequency domain channel response after applying inverse matrix: (a) Tx0 to Rx0, (b) Tx0 to Rx1, (c) Tx1 to Rx0, (d) Tx1 to Rx1. 6.5 Link performance This section presents the MIMO OTA performance testing results comparing different Modulation and Coding Scheme (MCS) for 2 × 2 MIMO OTA performance test at FR2 frequency band (see Figure 32) under clustered delay line (CDL) Urban Microcell (UMI) non-line-of-sight (NLOS) environment scenario for a moving user equipment (UE) with velocity of 15 kmh. Three MCS which represent low, median, and high modulation orders were chosen to assess the effect of signalto-noise (SNR) towards the MIMO OTA link performance. An Additive White Gaussian Noise (AWGN) noise generation module embedded within the CE was used to produce a variable-level AWGN signal alongside the desired signal, based on the target receive SNR. Due to a bandwidth limitation caused by the RTS implementation, a maximum 20 MHz bandwidth is allocated to the PDSCH channel, corresponding to 14 RBs at a 120 kHz subcarrier spacing. The Block Error Rate (BLER) results were calculated based on decoded Cyclic Redundancy Check (CRC) results of 16000 transmission blocks (TBs). The throughput results were calculated based on the BLER results and theoretical peak rate under the TB configuration. The BLERexhibits relatively slow convergence across all threeMCS levels, whichcan be attributedto thecharacteristics of the CDL UMI NLOS channel model. 47 (a) (b) Figure 32: Measured link performance for 2 × 2 MMO OTA performance test at FR2: (a) Block Error Rate (BLER); (b) Throughput 6.6 Conclusion The section presents a real-world evaluation of an RTS-based mm-wave 5G NR MIMO OTA test system. An SDR-based software UE emulator has been developed, which contains a fully functioning baseband processing unit compliant with the 5G NR physical layer protocol. A sophisticated system controlling program to coordinate different functional blocks in the testbed including BSE, CE, mode stirrer, and UE emulator control has been developed. Some key factors which can significantly influence the performance of the RTS such as the ATF have been carefully studied and evaluated by performing a measurement campaign in a fully anechoic chamber in the FR2 frequency band. Finally, the performance evaluation of the antenna radiation pattern and H-matrix for the RTS method in the near field is presented through these preliminary measurement results. 48 7. RMS delay spread reduction in reverberation chambers by channel model delay tap removal 7.1 Introduction When performing OTA radio tests in RCs [71], the inherent delay spread in the RC can affect the signalling performance for measurements involving throughput measurements. One way to deal with this influence from the chamber is to load the RC with Radio Frequency (RF) absorbers in order to reduce the Q-factor, which is the underlying factor for the long inherent delays, normally characterized by the Root Mean Square (RMS) delay spread. When large devices such as vehicles are tested with OTA techniques in RCs, large chambers are needed and consequently a large amount of RF absorbers to limit the RMS delay spread to limits stated by standards like [72] (3GPP). In this section we have investigated the possibility to alter the emulated channel to get a better compliance with the stated RMS delay spread requirements for OTA measurements in RC’s. 7.2 Background 7.2.1 Channel model In this section, we used the Urban Microcell channel model (UMi), which emulates radio propagation in urban environments. It is a clustered delay line model and is defined by 3GPP in [72]. The model contains six clustered delay taps (three Rayleigh paths for each cluster), with separate amplitude profiles to simulate the multipath EM urban environment. It also contains a Doppler model to simulate moving UE. 7.2.2 RMS Delay Spread RMS delay spread is calculated from the PDP, and definitions are found in [72]. The RMS delay spread στis computed by: στ=sPnP(τn)τ2 n Pn(Pn)−PnP(τn)τn PnP(τn)2(7-1) where P(τ)is the PDP, given by P(τ) = 1 T T X t=1 |h(t, τ)|2(7-2) Here h(t, τ)is the inverse Fourier transform of the of the channel matrix H(t, f)which is captured by sequential VNA S21 measurements. 7.2.3 Reverberation Chambers RCs are used for EMC and radio testing. They usually consists of a shielded cavity with conducting walls, roof, and floor with some installations to achieve alterations in the EM boundary conditions (mode stirrers), that will be moved throughout of a performed test. When stirring is performed correctly, the EM environment will be statistically isotropic and uniform [73]. 49 TwoRCsarementionedinthisreport: TheRCtent,whichisalargeRCmadeofconductivetextilewithdimensions 16x8x6 m, and a smaller fixed RC, the Fourier RC, with dimensions 3.7x2.6x4.5 m. Ref. [72] proposes a RMS delay spread of maximum 55ns. However, this can be hard to reach in large RCs (see Figure 33) due to a huge number of RF absorbers needed. For the RC tent in this example, approximately 200 absorbers would be needed to reach this limit. Figure 33: Delay spread vs. number of RF absorbers in the large RC tent. 7.2.4 Inherent RMS delay spread of RCs When a channel model like the UMi is transmitted through the RC, the delaytaps become ”smeared out” by the Q factor of the chamber and it affects the RMS delay spread. This is called the inherent delay spread and is illustrated by the PDP measurements in Figure 34 (Fouier RC). In Figure 35 the inherent delay spread is shown together with an emulated channel. The black line is a UMi channel measured with a conducted setup, and the blue and red lines are UMi channels, measured through the Fourier RC under different loading conditions. 50 Figure 34: Inherent delay spread of the Fourier RC. Figure 35: PDP and delay spread for a UMi channel in a conducted setup, and in the Fourier RC under different loading conditions. 51 Table 2: Measurement equipment. 7.3 Measurement setups For the measurements a conducted setup was used as reference, and this can be seen in the upper part of Figure 36. The RC setup is shown in the same figure in the lower part. The instruments are shown in Table 2. Figure 36: Conducted measurement setup above and RC measurement setup below. 7.4 Method to reduce RMS delay spread in RCs By removing far-end cluster taps in the UMi channel, the resulting RMS delay spread through the RC is reduced, since trailing delay contributions are removed. This approach was tested for the removal of several clustered delay taps and vary the number of RF absorbers. The results can be seen in Table 3 and the Figure 37 - 38 shows some example plots. 52 Table 3: RMS delay spread measurements with different number of delay taps, absorbers and centre frequencies. 53 Figure 37: RMS delay spread at 1 GHz in Fourier RC. 12 taps seems like a good choice. Figure 38: RMS delay spread at 2 GHz in Fourier RC with 4 RF absorbers. 12 taps seems like a good choice. 7.5 Conclusion This document investigates an additional method to tune the RMS delay spread for RC measurements involving CDL channel models. This method can reduce the number of absorbers needed to get an RMS delay spread that meets the specifications of certain testing standards. This is especially important for large RCs. 54 8. Theoretical study on keyhole reduction methods in reverberation chambers 8.1 Introduction The theoretical study shown in this section is part of activity 1.1.4 in the 21NRM03 MEWS project. The keyhole effect can appear in poor channel conditions and can reduce the rank of the communication link. ThiseffectcanalsobeencounteredwhenperformingcommunicationtestinginRCswithCEswhenusingMIMOsignaling (see [74] and [75]). This is due to that the signal is affected by more than one Rayleigh fading. In our case one Rayleigh fading is applied in the CE and another one is inherently applied in the RC. This faulty reduces the channel capacity and hence introduces an error during equipment testing. One way of reducing this effect is to increase the number of down-link antennas in the RC. For higher order MIMO this becomes costly and impractical as a lot of antennas will be needed. As an entry to the work on efficient communication testing in the MEWS project, a method of reducing the keyhole effect and the number of RC antennas needed is theoretically investigated in this report. 8.2 Background The MIMO channel capacity is often explained as in [76] and can be summarized as C(i) = log2detINR+ρ NTHH´ (8-1) where C(i)is capacity per channel realization (e.g., RC stirrer state), Iis the identity matrix, ρis the SNR, NTis the number of transmitters, NRis the number of receivers and His the channel matrix. In normal channel emulation both straight and cross Radio Links (RL) are present in the RF connection setup (see Figure 39). The proposed idea is to reduce the Rayleigh content in the channel model by removing the cross-RL and simultaneously introduce a LOS component by increasing the K-factor. Figure 39: CE port configuration with cross-RL (left) and without (Right). 55 To investigate the method, a set of simulations were performed. Several channels were simulated (different H) and the results for 2x2 MIMO can be seen in the plots below. In Figure 40 the Cumulative Distribution Function (CDF) is plotted. Rayleigh is the theoreticalreference (e.g., only a CE). Keyhole = 2 and Keyhole = 4 shows the signal when the CE is connected to the RC with two or four keyholes. Here the keyhole effect is present. CE 2 straight RL shows the CDF when the cross RL are removed. CE Rician, 2 straight RL shows the CDF when the cross RL are removed and a K-factor of approximately 10 dB is applied. CE Rician, 2 straight RL is the proposed model and aligns well with the theoretical Rayleigh signal. Figure 40: CDF of the norm of the channel matrix (H). 56 Figure 45: Two-stage testbed setup and measurement scheme using a PNA-X for mm-wave multiple channel scenarios 9.3 Subsystem verification All the measurements presented in this section were performed within the RC of the UK NPL. 9.3.1 Channel sounding measurement To characterize the propagation channel between transmitter and receiver in the RC, a high accuracy channel sounder and the related estimation algorithm were developed. Figure 46 shows the block diagram for realization of the channel sounder system. As depicted in Figure 46, the channel sounder system was designed based on a software-defined-radio (SDR) platform. In this work, a pair of National Instrument (NI) vector signal transceivers (VSTs) PXIe-5840 are used as the RF front ends at both transmitting (Tx) and receiving (Rx) ends respectively, with up to 1 GHz RF bandwidth to provide high accurate delay estimation. Meanwhile, two FPGA based baseband processing units have been used as signal generator at Tx end and data streaming at Rx end. Figure 46: Block diagram for the channel sounder system and setup in RC To enable synchronization of the measurement platform, a rubidium clock unit was employed as the reference signal source which provide both Tx and Rx highly stable 10 MHz reference signal with less than 1 PS of timing error accuracy. The reference sharing scheme between the Tx and Rx combined with a hardware trigger in the controller 63 provides the higher synchronization and very low jitter between transmitter and receiver. At the baseband of the Tx end, a Pseudo-Noise sequence of length 4095 is firstly up-sampled to a two-fold sampling rate and passes through the root raised cosine (RRC) pulse shaping filter to achieve multipath delay resolution to less than 20 ns for the channel estimation of the RC channel. At the Rx end, a post-processing module is developed to estimate the CIR of the RC from the received streamed IQ data sequence. The post-processing module includes an 8190-points length slide correlation module, a windowing module, and a calculating module of the equalizer filter. The time-domain slide correlation was conducted to achieve the CIR of the RC, and a square window was used to significantly reduce the impacts from the noise sample within the CIR. Finally, an equalizer filter was estimated by performing calculations with the help of domain transformation and was output to the channel model synthesis step. In addition, a cable-based calibration procedure was performed at the beginning of the measurement, allowing the effects of the Tx and Rx system response to be measured and compensated based on the calibration results in the postprocessing stage. A pair of directional linearly polarized antennas were used at both Tx and Rx ends. Table 4 lists the key system settings and parameters. 9.3.2 Setup of reverberation chamber Figure 47 shows a photograph of the measurement setup inside the RC at NPL. The dimension of the RC is 6.55 m × 5.85 m × 3.5 m, it contains one vertically installed paddle stirrer. The transmit and receive antennas are the ETS-Lindgren’s model 3117 double-ridged waveguide horn antennas, which are located at two corners of the RC three meters apart. The paddle stirrer is located between the two antennas and blocks the LoS path between them. Table 4: Key parameters of the channel sounder Name of parameter (unit) Value Centre frequency 3.5 GHz Peak output power 30 dBm Ref accuracy 1 ps Symbol rate 100 MS/s PN sequence length 4095 Pulse shaping filter 13-order RRC filter, roll-off factor 0.25 Max capture length (cycles per snapshot) 81920 (10 cycles per snapshot) Antenna gain @3.5 GHz 6.8 dBi Half-power beamwidth @3.5 GHz E-plane: 68 degree; H-plane: 80 degree 64 Figure 47: RC measurement setups and the antenna pair 9.3.3Setupofchannelemulator The CE used in the work was based on the vector signal transceiverPXIe-5644R which is set to 100 MHz channel sampling. Two typical channel models were selected for verification: 1) the Pedestrian-B model defined of in the 3GPP SCM channel model [86] which is a scenario with fewer multipaths; 2) the tapped delay line B (TDL-B) model with a 300 ns of delay spread scaling defined in the 3GPP 5G channel model [70] which is rich in multipaths. The two models are presented in Table 5 and Table 6. Limited by the SDR equipment minimum sample interval of 10 ns, several multipaths are indistinguishable and are merged in the CE. As shown in Figure 48, the TDL-B model with rich multipaths and relatively small delay intervals is set in the CE. The Rayleigh distribution can be set for each path in terms of its LoS and NLoS settings. Doppler spread can also be imposed on each path according to Jakes spectrum model determined by the velocity of the motion experienced by the DUT. Table 5: Parameters of TAPS defined in Pedestrian-B in [86]. Tap # delay [ns] Power [dB] Fading distribution Coerced delay∗[ns] 1 0.0000 0 Rayleigh 0 2 200 -0.9 Rayleigh 200 3 800 -4.9 Rayleigh 800 4 1200 -8.0 Rayleigh 1200 5 2300 -7.8 Rayleigh 2300 6 3700 -23.9 Rayleigh 3700∗ *Note that due to the limitation of CE software, the last path at 3700 ns could not be set in the experiment. 65 Table 6: Parameters of TAPS defined in TDL-B in [70]. Tap # Normalized delay [ns] Power [dB] Fading distribution Coerced delay∗[ns] 1 0 0 Rayleigh 0 2 0.03216 -2.2 Rayleigh 30 3 0.06285 -4.0 Rayleigh 60 4 0.06465 -3.2 Rayleigh 60 5 0.08610 -9.8 Rayleigh 90 6 0.08958 -1.2 Rayleigh 90 7 0.11043 -3.4 Rayleigh 110 8 0.11091 -5.2 Rayleigh 110 9 0.11256 -7.6 Rayleigh 110 10 0.15165 -3.0 Rayleigh 150 11 0.15849 -8.9 Rayleigh 160 12 0.17100 -9.0 Rayleigh 170 13 0.33063 -4.8 Rayleigh 330 14 0.38268 -5.7 Rayleigh 380 15 0.46422 -7.5 Rayleigh 460 16 0.55326 -1.9 Rayleigh 540 17 0.65057 -7.6 Rayleigh 610 18 0.84882 -12.2 Rayleigh 850 19 0.96857 -9.8 Rayleigh 910 20 1.08561 -11.4 Rayleigh 1090∗ 21 1.23201 -14.9 Rayleigh 1230∗ 22 1.28370 -9.2 Rayleigh 1280∗ 23 1.43502 -11.3 Rayleigh 1440∗ *Note that due to the limitation of CE software, the last 4 paths could not be set in the experiment. 66 Figure 48: The PDP setting of CE 9.4 Measurement Results The following subsections show, respectively, the measurement results for SISO and MIMO setups. 9.4.1ForSISOsetup In Figure 49a, the PDP of RC at one stirrer position is presented which follows a typical continuous exponential delay profile with maximum delay value of approximately 2500 ns. After channel estimation and postprocessing as described above, the equalizer filter was derived and output to the convolution module in the channel model synthesis step. To evaluate the effectiveness of the equalizer filter, a channel measurement which includes complete RF path in the channel model synthesis step and bypassing the CE is performed. From result of the Figure Figure 49b, it can be seen that one can achieve the promising cancellation effectiveness. In Figure 49c and Figure 49d, the effectiveness of the two-step approach for two typical standard channel model is evaluated. In Figure 49c, five discrete multipaths can be observed clearly with correct delay and power profile. Similar result could also be observed in simulation of rich multipaths scenario TDL-B of 3GPP 5G channel model from Figure 49d, where clear discrete PDP can be seen and correspond correctly to the setting in the CE. 67 (a) (b) (c) (d) Figure 49: (a) The PDP of RC; (b) the PDP of RC after convolving with the derived equalizer filter; (c) the PDP of the synthesized channel for model Pedestrian-B; (d) the PDP of the synthesized channel for model TDL-B with a 300 ns delay spread scaling 9.4.2 For MIMO setup Figure 50 shows the comparison of the measured PDP decoupling mode based on an inversematrix approach [87], by-pass mode and decoupling mode based on singular value decomposition of RC+CE for 2 × 2 MIMO channels. The two approaches—namely, the inverse-matrix method and the singular value decomposition (SVD) method—differ only in how they handle matrices with a high condition number. As we can see from Figure 50b and Figure 50c, a good cancellation effectiveness for the multipaths (except the first path) in every cochannel (e.g. T1R1 and T2R2) and all multipaths in all cross-channels (e.g. T1R2 and T2R1) can be observed in both methods. The inversematrix method showed a slightly better performance than SVD-based method. Due to fluctuation in the RF channel responses estimate caused by PDP decoupling module, obvious residuals can also be observed in the results of both methods. 68 (a) (b) (c) Figure 50: Measured PDP of RC+CE setup for 2 x 2 MIMO channels (a) for decoupling mode based on inverse-matrix approach; (b) measured PDP for by-pass mode and (c) measured PDP for decoupling mode based on singular value decomposition 9.5 Link Performance This section presents the MIMO OTA performance testing results comparing different MCS for 2 × 2 MIMO OTA performance test at FR2 frequency band (see Figure 51) using RC+CE method. Three MCS which represent low, median, and high modulation orders were chosen to assess the effect of SNR towards the MIMO OTA link performance. An AWGN noise generation module embedded within the CE was used to produce a variable-level AWGN signal alongside the desired signal, based on the target receive SNR. Due to a bandwidth limitation caused by the RC+CE implementation, a maximum 10 MHz bandwidth is allocated to the PDSCH channel, corresponding to 7 RBs at a 120KHz subcarrier spacing. The BLER results were calculated based on decoded CRC results of 16000 transmission blocks (TBs). The throughput 69 results were calculated based on the BLER results and theoretical peak rate under the TB configuration. The BLER exhibits relatively slow convergence across all three MCS levels, which can be attributed to the characteristics of the CDL UMI NLOS channel model. The planned 4 × 4 MIMO OTA performance test at FR1 frequency band has not been carried out due to link establishment between the BSE and mobile UE cannot be established. (a) (b) Figure 51: Measured link performance for 2 × 2 MIMO OTA Performance Test at FR2 using RC+CE method: (a) BLER; (b) Throughput 9.6 Conclusion This chapter presents a novel two-step closed-loop method for accurate simulation of 3GPP 5G channel or SCME channels in an RC environment. From actual measurements in the RC, even if the channel sample rate of the CE is not very high, the efficiency of eliminating the inherent continuous channel response can be demonstrated. After evaluating the effectiveness of simulating two typical channel models in a RC environment, the method demonstrates the ability to extend the RC to performance tests defined in standards such as 3GPP, and the results can be further compared with other candidate 5G OTA methods such as MPAC and RTS. Further work to be considered includes how a reverberation chamber can emulate the specific MIMO spatial correlation. 70 10. Inter comparisons 10.1 Comparison of PDP from UMa channel model 10.1.1 Measurement of PDP in RC for UMa channel model The PDP was measured in the RC Fourier at the RISE EMC testing facility in Borås, Sweden for the A1.1.5 activity in the Euramet 21NRM03 MEWS Project. 10.1.2 Measurement setup The PDP was calculated with MATLAB form VNA S21 measurements (20 measurements) according to the PDP definition in 3GPP. The setup is schematically shown in Figure 52. The UMa channel model was implemented in a CE with a modulation bandwidth of 40 MHz. The signalling was configured for a 2x2 MIMO use case and the measurements were performed at a center frequency of 3 GHz. To decrease the delay spread influence of the RC, eight RF absorbers were placed at random positions in the RC. The UE speed was set to 30 km/h in the CE. Figure 52: PDP measurement setup. Table 7: Measurement equipment. 10.1.3 Results In Figure 53 the measured PDP can be seen together with the nominal path delays from the UMA channel model. 71 Figure 53: Measured PDP in RC with a UMA channel model at center frequency 3 GHz with 8 absorbers 10.2 Intercomparison between RTS and RC+CE methods This subchapter presents the inter-laboratory comparison for the OTA performance testing of the device under test DUT, the software-defined radio (SDR) system using RTS and RC+CE methods. 10.2.1 Introduction Antenna measurement facility intercomparison activities have become mandatory to achieve or maintain official accreditation such as ISO 17025 [64]. The main goal of such activities is to provide opportunities for the participants to validate and document their laboratory proficiency and competence by comparison with other facilities. They constitute a crucial foundation for validation of measurement accuracy among participants and provide an important prerequisite for certification of facilities as well as inputs to standards and research on measurement uncertainties [88]. The following sections present the details of the intercomparison carried out between the RTS and RC+CE methods at the UK National Physical Laboratory for the same DUT. 10.2.2MeasurementSetup Thissectionshowstherelevantmeasurementsetupsandequipmentsettingsfortheinterlaboratory comparison. 10.2.2.1 Equipment settings This subsection presents the equipment settings for BSE and channel emulator (CE) for different DUT, namely, the mobile UE and the SDR system. 1. Base Station Emulator (BSE) Configuration 72 11. Conclusion This deliverable gathered, advanced, and validated a set of metrology methods that make OTA testing practical, traceable, and scalable as industry moves from 5G toward sub-THz 6G. The overarching goal has been to reduce cost andtesttime whilepreservingaccuracy andrepeatability—firmlygroundedinstandards—andtoensurethatlaboratory evaluations are representative of real-world performance. These aims guided all activities and were framed explicitly in the introduction, including the emphasis on uncertainty, traceability, and alignment with established 3GPP/ETSI specifications. A first contribution is a compact, cost-effective “wireless cable” solution for MIMO radio testing. By establishing stable port-to-port OTA links inside a small RF enclosure and calibrating the probe–DUT transfer matrix, we reproduced cabled functionality without consuming scarce channel-emulator resources. The approach delivered strong isolation across multiple links and—critically—produced throughput results that track conducted measurements across several MIMO channel ranks. Peak rates on representative rank-1, rank-2, and rank-4 cases closely matched their conducted counterparts at comparable RSRP levels, supporting wireless cable as a viable, industry-ready path for 5G FR1/FR2 device performance testing. Building on this principle, we generalized the wireless-cable concept into a framework for arbitrary field emulation in non-anechoic environments. Rather than synthesizing a full quiet-zone field as in classic array-based methods, the proposed approach measures the chamber matrix Abetween Kprobes and NDUT ports and then loads a “target field” at the DUT via a calibration matrix G=A†IN, effectively reconstructing the desired receive vector directly at the DUT ports. This reduces hardware complexity, relaxes anechoic requirements, and enables emulation of plane waves with varying AoA, non-stationary fields, and beamforming scenarios. Laboratory demonstrations verified robustness in a standard room environment, linking metrology-grade reproducibility with pragmatic test setups. To ensure that MPAC setups faithfully realize 3GPP spatial channels at FR2, we developed and validated a DirVAA for test-zone validation. Relative to traditional omnidirectional virtual arrays, the Dir-VAA improves SNR, reduces required samples through larger inter-element spacing (thanks to sidelobe suppression), and shortens measurement time. Measurements showed high agreement between emulated and target channels: probe-power deviations remained well within a tight tolerance window, the PAS similarity was high, and downlink throughput behavior remained consistent under canonical CDL profiles. These findings confirm suitability for real MIMO performance testing while keeping measurement budgets under control. Looking ahead to 6G, we investigated dynamic sub-THz channel emulation. We outlined an end-to-end chain for digital channel emulation at sub-THz, including frequency extension, band-stitching for wide BWs, and tap-resource optimization to stay within practical DSP limits—while still capturing dominant paths under blockage and transition dynamics. Using a commercial platform, we emulated dynamic sub-THz channels with multi-gigahertz effective BWs and validated the approach with beam-steerable radios and human-blockage scenarios, demonstrating accurate reproduction of target CIRs and path dynamics within emulator constraints. Two reverberation-chamber (RC) studies addressed pressing OTA challenges. First, RMS delay-spread reduction was tackled by shaping the emulated channel: removing far-end clustered delay taps of the UMi model lowers delay spread after the RC’s inherent Q-induced smearing, thereby reducing the amount of absorber material needed to 79 meet specification limits (e.g., 55 ns that are otherwise costly to achieve in large chambers). Measurements across frequencies and absorber loads confirmed the efficacy of this “tap-removal” strategy. Second, a keyhole-reduction strategy for RC-based MIMO testing was proposed and validated in simulation: suppressing cross radio links in the emulator and introducing a LOS component (higher Rician K-factor) reduced keyhole effects and improved capacity without adding antennas—bringing the distribution closer to ideal Rayleigh behavior. This offers a practical pathway to higher-rank channel realizations in RCs with minimal hardware overhead, with physical validation and higher-order MIMO scaling set as future steps. The report also emphasizes the analysis and comparison of multiple candidate link performance testing methods, focusing on design challenges, key performance impact factors, and real-world results comparison. Regarding the RTS method, a practical testbed based on the software-defined radio (SDR) platform was developed. Key functions and their impacts—such as antenna radiation pattern measurement using actual 5G NR downlink SSB reference signals, estimation and cancellation of the H-inverse matrix—were thoroughly investigated. Regarding the RC+CE approach, a novel two-step method based on RC PDP decoupling was proposed and extended from SISO to MIMO. Likewise, a practical implementation using a multi-channel SDR was developed and evaluated, with particular emphasis on FR2 MIMO link performance. Finally, an intercomparisonof the twomethods—RTSand RC+CE—based onreal-world measurements demonstrated generally good agreement in link performance under selected scenarios. Throughout the report, we kept a strong emphasis on traceability and standards alignment, with methods designed to dovetail with 3GPP/ETSI frameworks and validated, where relevant, via inter-laboratory comparisons to demonstrate reproducibility across sites. That emphasis underpins the metrological credibility required for regulatory acceptance and industrial uptake. By unifying compact wireless-cable testing, generalized field emulation outside anechoic constraints, rigorous FR2 test-zone validation, dynamic sub-THz channel emulation, and RC-based delay/rank conditioning, this work advances a practical, standards-aware measurement science for 5G/6G radios. The resulting toolbox is both scientifically rigorous and industrially deployable—ready to be scaled, standardized, and transferred into conformance and performance test workflows. 80 References [1] ITU-R Report IMT.2030. “Framework and Overall Objectives of the Future Development of IMT for 2030 and Beyond”. In: ITU (2023). [2] Studyon ScenariosandRequirementsforNextGenerationAccessTechnologies.“3GPP TR 38.913V15.0.0”.In: 3GPP (2018). [3] Study on test methods. “TR 138 827 V16.0.0”. In: European Telecommunications Standards Institute (ETSI) (2020). [4] BaseStation(BS)andUserEquipment(UE)conformancetestingaspects.“TR138810V16.1.0”.In:EuropeanTelecommunications Standards Institute (ETSI) (2020). [5] Base Station (BS) Conformance Testing Part 2: Radiated Conformance Testing. “Technical Specification, V17.7.0”. In: 3GPP (2022). [6] H Kong et al. “Midfield Over-the-Air Test: A New OTA RF Performance Test Method for 5G Massive MIMO Devices”. In: IEEE Transactions on Microwave Theory and Techniques 67.7 (2019), pp. 2873–2883. [7] M Hofer et al. “Real-Time Geometry-Based Wireless Channel Emulation”. In: IEEE Transactions on Vehicular Technology 68.2 (2019), pp. 1631–1645. [8] Wei Fan et al. “A Flexible Millimeter-Wave Radio Channel Emulator Design With Experimental Validations”. In: IEEE Transactions on Antennas and Propagation 66.11 (2018), pp. 6446–6451. [9] Wei Fan et al. “MIMO Terminal Performance Evaluation With A Novel Wireless Cable Method”. In: IEEE Transactions on Antennas and Propagation 65.9 (2017), pp. 4803–4814. [10] New Radio Study on Test Methods. “document 3GPP 38.810”. In: 3GPP (2017). [11] Yilin Ji et al. “Virtual Drive Testing Cver-the-Air for Vehicular Communications”. In: IEEE Transactions on Vehicular Technology 69.2 (2020), pp. 1203–1213. [12] Fengchun Zhang, Wei Fan, and Z Wang. “Achieving Wireless Cable Testing of High-Order MIMO Devices With A Novel Closed-Form Calibration Method”. In: IEEE Transactions on Vehicular Technology 69.1 (2021), pp. 478–487. [13] Wei Fan et al. “Wireless Cable Method for High-Order MIMO Terminals Based on Particle Swarm Optimization Algorithm”. In: IEEE Transactions on Antennas and Propagation 66.10 (2018), pp. 5536–5545. [14] P Shen et al. “A Decomposition Method for MIMO OTA Performance Evaluation”. In: IEEE Transactions on Vehicular Technology 67.9 (2018), pp. 8184–8191. [15] Wei Fan et al. “Wireless cable testing for MIMO radios: A compact and cost-effective 5G radio performance test solution”. In: IEEE Transactions on Antennas and Propagation 71.10 (2023), pp. 8239–8249. [16] New Radio Study on Test Methods. “[Online]. Available: http://www.3gpp.org/DynaReport/38810.htm”. In: Version 16.7.0, Sophia Antipolis, France, 3GPP Rep. (TR) 38.810 (2023). [17] Test Plan for: “2x2 Downlink MIMO and Transmit Diversity Over-The-Air Performance”. In: version 1.1.1 (2018). [18] H Gao et al. “Over-The-Air Performance Testing of 5G New Radio User Equipment: Standardization and Challenges”. In: IEEE Communications Standards Magazine 6.2 (2022), pp. 71–78. [19] FengchunZhangetal.“PlaneWaveGeneratorinNon-AnechoicRadioEnvironment”.In:IEEEAntennasandWireless Propagation Letters (2023). [20] P Kyosti, T Jamsa, and J-P Nuutinen. “Channel Modelling For Multi-Probe Over-The-Air MIMO Testing”. In: International Journal of Antennas and Propagation 1 (2012). 81 [21] P. Zhang et al. “A Survey of Testing for 5G: Solutions, Opportunities, and Challenges”. In: China Communications 16.1 (2019), pp. 69–85. [22] Wei Fan et al. “A Step Toward 5G in 2020: Low-Cost OTA Performance Evaluation of Massive MIMO Base Stations”. In: China Communications 59.1 (2016), pp. 38–47. [23] Chunhui Li et al. “A Generalized Wireless Cable Over-the-Air Testing Framework for Arbitrary Field Emulation in Nonanechoic Radio Environment”. In: IEEE Antennas and Wireless Propagation Letters (2024), pp. 1–5. [24] W Yu et al. “Radiated Two-Stage Method for LTE MIMO User Equipment Performance Evaluation”. In: IEEE Transactions on Electromagnetic Compatibility 56.6 (2014), pp. 1691–1696. [25] Y Qi et al. “5G Over-The-Air Measurement Challenges: Overview”. In: IEEE Transactions on Electromagnetic Compatibility 59.6 (2017), pp. 1661–1670. [26] M Rumney. “Testing 5G: Time to Throw Away the Cables”. In: Microwave Journal 59.11 (2016), pp. 10–18. [27] Wei Fan et al. “Over-The-Air Radiated Testing of Millimeter-Wave Beam-Steerable Devices in a Cost-Effective Measurement Setup”. In: IEEE Communications Magazine 56.7 (2018), pp. 64–71. [28] User Equipment (UE) Radio Transmission and Reception. “3GPP TS 36”. In: Version 10 (2011). [29] Y Li et al. “Dual Anechoic Chamber Setup for Over-The-Air Radiated Testing of 5G Devices”. In: IEEE Transactions on Antennas and Propagation 68.3 (2020), pp. 2469–2474. [30] X Chen et al. “Throughput Modeling and Validations for MIMO-OTA Testing with Arbitrary Multipath”. In: IEEE Antennas and Wireless Propagation Letters 17.4 (2018), pp. 637–640. [31] M G Nilsson et al. “Measurement Uncertainty, Channel Simulation, and Disturbance Characterization of an OverThe-Air Multiprobe Setup for Cars at 5.9 GHz”. In: IEEE Transactions on Industrial Electronics 62.12 (2015), pp. 7859– 7869. [32] Y Ji et al. “On Channel Emulation Methods in Multiprobe Anechoic Chamber Setups for Over-The-Air Testing”. In: IEEE Transactions on Vehicular Technology 67.8 (2018), pp. 6740–6751. [33] Pekka Kyosti and J Kyrolainen. “Systems and Methods for Radio Channel Emulation of a Multiple Input Multiple Output (MIMO) Wireless Link”. In: U.S. Patent 10 601 695 (2020). [34] H Gao et al. “Over-the-Air Testing for Carrier Aggregation Enabled MIMO Terminals Using Radiated Two-Stage Method”. In: IEEE Access 6 (2018), pp. 71622–71631. [35] WXue,F.Li,andXChen.“EffectsofSignalBandwidthonTotalIsotropicSensitivityMeasurementsinReverberation Chamber”. In: IEEE Transactions on Instrumentation and Measurement 70 (2021), pp. 1–8. [36] Mengting Li et al. “Test Zone Validation for FR2 MPAC Using Directional Antenna Based Virtual Array”. In: IEEE Transactions on Instrumentation and Measurement 72 (2023). [37] P Kyosti et al. “On Radiated Performance Evaluation of Massive MIMO Devices in Multiprobe Anechoic Chamber OTA Setups”. In: IEEE Transactions on Antennas and Propagation 66.10 (2018), pp. 5485–5497. [38] Study on Channel Model for Frequencies From 0.5 to 100 GHz. “Available: https://www.3gpp.org”. In: document TR 38.901 (2022). [39] Data Sheet for. “ASY-CWG-S-265, Available: https://asysol.com/portfolio/circularcorrugatedfeeds/”. In: March 4 2021 (2021). [40] I Jaafar et al. “Joint Angle and Delay Estimation of Point Sources”. In: IEEE International Conference on Electronics, Circuits and Systems (2005). [41] Fengchun Zhang et al. “Millimeter-Wave New Radio Test Zone Validation for MIMO Over-The-Air Testing”. In: IEEE Transactions on Antennas and Propagation 70.2 (2022), pp. 1569–1574. 82 [42] Study on Radiated Metrics and Test Methodology for the Verification of Multi-Antenna Reception Performance of NR User Equipment. “Available: https://www.3gpp.org”. In: document TR 38.827 (2021). [43] HKrim and MViberg.“TwoDecadesofArraySignalProcessingResearch:The ParametricApproach”.In: IEEE Signal Processing Magazine 13.4 (1996), pp. 67–94. [44] A Ghosh and M Kim. “THz channel Sounding and Modeling Techniques: An Overview”. In: IEEE Access (2023). [45] C Han et al. “Terahertz Wireless Channels: A Holistic Survey on Measurement, Modeling, and Analysis”. In: IEEE Communications Surveys and Tutorials 24.3 (2022), pp. 1670–1707. [46] F8800b propsim F64 radio channel emulator. “[Online]. Available: https://www.keysight.com/ie/en/assets/31222164/data-sheets/F8800BPROPSIM-F64-Radio-Channel-Emulator.pdf”. In: Data Sheet, Keysight Technology USA (2022). [47] J Cao, F Tila, and A Nix. “Design and Implementation of a Wideband Channel Emulation Platform for 5G MmWave Vehicular Communication”. In: IET Communications 14.14 (2020), pp. 2369–2376. [48] J Sozanski and T Grosch. “Radio Frequency Multipath Channel Emulation System and Method”. In: US Patent 9,130,667 (2015). [49] Fengchun Zhang et al. “Dynamic Sub-THz Radio Channel Emulation: Principle, Challenges, and Experimental Validation”. In: IEEE Wireless Communications Magazine 31.1 (2024), pp. 10–16. [50] Real-Time Massive MIMO Channel Emulation. “[Online]. Available: http://wireless.engineering.nyu.edu/realtimemassive-mimo-channel-emulatio”. In: NYU WIRELESS, Technical Report (2024). [51] S Rey et al. “Channel Sounding Techniques for Applications in THz Communications: A First Correlation Based Channel Sounder for Ultra-Wideband Dynamic Channel Measurements at 300 GHz”. In: International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (2024), pp. 449–453. [52] A W Mbugua et al. “Efficient Pre-Processing of Site-Specific Radio Channels for Virtual Drive Testing in Hardware Emulators”. In: IEEE Transactions on Aerospace and Electronic Systems (2022). [53] M F De Guzman, P Koivumäki, and K Haneda. “Double-Directional Multipath Data at 140 GHz Derived from Measurementbased Ray-Launcher”. In: IEEE Vehicular Technology Conference (2022). [54] P. Zhang et al. “Measurement-Based Characterization of D-Band Human Body Shadowing”. In: European Conference on Antennas and Propagation (2023). [55] J Kokkoniemi et al. “Initial Results on D Band Channel Measurements in LOS and NLOS Office Corridor Environment”. In: European Conference on Antennas and Propagation (2022). [56] T. H. Loh. “Metrology for 5G and Emerging Wireless Technologies”. In: The Institution of Engineering and Technology (2021). [57] Study on radiated metrics and test methodology for the verification of multi-antenna reception performance of NR User Equipment (UE). “TR 38.827 V16.8.0 (2022-09)”. In: 3GPP (2022). [58] The 3rd generation partnership project. “https://www.3gpp.org/”. In: 3GPP (2025). [59] European Telecommunications Standards Institution. “https://www.etsi.org/”. In: ETSI (2025). [60] Verification of radiated multi-antenna reception performance of User Equipment (UE). “TR 37.977 V19.0.0 (202509)”. In: 3GPP (2024). [61] S Phang et all. “Near-Field MIMO Communication Links”. In: IEEE Transactions on Circuits and Systems I: Regular Papers 65.9 (2018), pp. 3027–3036. [62] S Olenik et all. “The future of near-field communication-based wireless sensing”. In: Nat. Rev. Mater 6 (2021), pp. 286–288. 83 [63] M Cui et all. “Near-Field MIMO Communications for 6G: Fundamentals, Challenges, Potentials, and Future Directions”. In: IEEE Communications Magazine 61.1 (2023), pp. 40–46. [64] J An et all. “Near-Field Communications: Research Advances, Potential, and Challenges”. In: IEEE Wireless Communications 30.3 (2024), pp. 100–107. [65] Z Zhang et all. “Dynamic MIMO Architecture Design for Near-Field Communications”. In: IEEE Transactions on Wireless Communications 23.10 (2024), pp. 14669–14684. [66] MHaider et all. “Near-Field MIMO SystemAssessment”.In: 2nd URSIAtlanticRadioScienceMeeting (AT-RASC 2018) (2018). [67] P Shen et all. “An RTSBased Near-Field MIMO Measurement Solution—A Step Toward 5G”. In: IEEE Transactions on Microwave Theory and Techniques 67.7 (2019), pp. 2884–2983. [68] Y Gui et all. “Near-Field Evaluation of an Radiated-Two-Stage-Based Millimeter-Wave 5G New-Radio MultipleInput-Multiple-Output Over-the-Air Test System”. In: 19th European Conference on Antennas and Propagation (EuCAP 2025) (2025). [69] Physical layer measurements. “TS 38.215 V18.3.0 (2024-06)”. In: 3GPP (2024). [70] Study on channel model for frequencies from 0.5 to 100 GHz. “TR 38.901 V19.1.0 (2025-09)”. In: 3GPP (2024). [71] Xiaoming Chen et al. “Reverberation Chambers for Over-the-Air Tests: An Overview of Two Decades of Research”. In: IEEE Access 6 (2018), pp. 49129–49143. [72] Universal Terrestrial Radio Access (UTRA) and Evolved Universal Terrestrial Radio Access (E-UTRA); Verification of radiated multi-antenna reception performance of User Equipment (UE). “Technical Report (TR) 37.977”. In: 3GPP (2018). [73] IEC 61000-4-21:2011. “Electromagnetic compatibility (EMC) - Part 4-21: Testing and measurement techniques - Reverberation chamber test methods”. In: IEC (2011). [74] Kristian Karlsson et al. “On the Keyhole Effect in Over-The-Air Testing of Higher Order MIMO Systems”. In: 2021 15th European Conference on Antennas and Propagation (EuCAP). 2021, pp. 1–5. DOI: 10.23919/EuCAP51087.2021. 9410976. [75] Charlie Orlenius and Mats Andersson. “Connected reverberation chambers with variable channel rank for measurement of MIMO antenna system performance”. In: 2009 IEEE Antennas and Propagation Society International Symposium. 2009, pp. 1–4. DOI: 10.1109/APS.2009.5172100. [76] Mohinder Jankiraman. Space-Time Codes and MIMO Systems. Artech House, 2004. [77] M Rumney et all. “MIMO over-the-air research, development, and testing”. In: Intenational Journal of Antennas and Propagation 23.10 (2012), pp. 1–8. [78] T Loh et all. “OTA Test Methods and Candidates for 5G and Beyond”. In: The Institution of Engineering and Technology (IET) 23.10 (2020), pp. 14669–14684. [79] D Sanchez et all. “Validation of 3GPP SCME channel models emulated in mode-stirred reverberation chambers”. In: 3GPP Tsg-ran Wg (2012). [80] C Li et all. “Evaluation of chamber effects on antenna efficiency measurements using non-reference antenna methods in two reverberation chambers”. In: IET Microwaves, Antennas and Propagation 11.11 (2017), pp. 1536– 1541. [81] T Loh et all. “An Assessment of the Adaptive Routing Performance of a Wireless Sensor Network using Smart Antennas”. In: IET Wireless Sensor Systems 4.4 (2014), pp. 196–205. 84 [82] A Cozza et all. “Accurate Radiation-Pattern Measurements in a Time-Reversal Electromagnetic Chamber”. In: IEEE Antennas and Propagation Magazine 52.2 (2010), pp. 186–193. [83] T Nguyen et all. “Emulation of 3GPP SCME power-delay profiles for characterisation of multiple-input multipleoutput antenna in reverberation chamber”. In: IET Microwaves, Antennas and Propagation 12.11 (2018), pp. 1828– 1833. [84] XGuo et all. “Onthe RealizationChallengesforAccurateSCMEChannelImplementationinRC”.In: XXXIVthGeneral Assembly and Scientific Symposium of the International Union of Radio Science (URSI GASS 2021) (2021). [85] D Baum et all. “An interim channel model for beyond-3G systems: extending the 3GPP spatial channel model (SCM)”. In: Proc. IEEE 61st Veh. Technol. Conf. (VTC ‘2005) 5 (2005), pp. 3132–3136. [86] Spatial channel model for Multiple Input Multiple Output (MIMO) simulations. “TS 25.996 V19.0.0 (2025-10)”. In: 3GPP (2025). [87] Y Gui et all. “A realisation of channel emulation in a reverberation chamber method for over-the-air compliance testing in support of 3GPP standardisation”. In: 17th European Conference on Antennas and Propagation (EuCAP 2023) (2023). [88] TLohetall.“The EurAAPWorkingGrouponAntennaMeasurements:Highlights overTwoDecades”.In:46thAnnual Meeting and Symposium of the Antenna Measurement Techniques Association (AMTA 2024) (2024). [89] Measurement of radiated performance for Multiple Input Multiple Output (MIMO), multi-antenna reception for High Speed Packet Access (HSPA), and LTE terminals. “TR 37.976 V19.0.0 (2025-10)”. In: 3GPP (2025). 85