Application Note StarMIMO RX Diversity and MIMO OTA Test Range
Contents Introduction P. 03 StarMIMO setup P. 04 1/ Multi-probe technology P. 05 Cluster vs Multiple Cluster setups Volume vs Number of probes 2/ Spatial channel emulator P. 07 Channel Models MIMO OTA testing Channel Models 3/ Radio communication tester P. 08 4/ Amplification unit P. 08 Calibration P. 09 1/ Calibration - Part 1 P. 10 2/ Calibration - Part 2 P. 10 SAM MIMO Control software interface P. 11 SAM MIMO Configuration P. 11 Conclusion P. 12 2
Introduction In Multiple Input Multiple Output (MIMO) systems, spatial correlation plays an important role. The level of correlation cannot be determined based on antenna characteristics without knowing the propagation characteristics. Therefore, a new Over-the-air (OTA) testing methodology is needed in order to take into account both antenna and propagation characteristics in the system setup when testing multiantenna devices. MIMO-OTA device testing with anechoic chambers emulates complex spatialtemporal propagation environments at the device location in a realistic and repeatable way by using a radio channel emulator connected to a circular array of probes (multiple cluster) or to an array of probes on one side of the chamber forming a single spatial cluster model. Both configurations are referred to as the Multi-probe technique. This technique addresses the need to generate precise spatial channel models at the device location for the evaluation of MIMO antenna and handset performance. Wide band cellular propogation channels exhibit the cluster behavior which can be taken into account by emulating the Spatial Channel model within the anechoic chamber. The clusters are then simultaneously mapped to the probes so that the sum of the transmitted signals at the DUT location is as defined in the model. These measurementbased channel models include all dimensions of the radio channel (time, frequency, space and polarization). Each cluster is split between several probes to enable angular spread. As a result the geometry based environment of the channel model is accurately reproduced within the anechoic chamber. The primary goal of the MIMO OTA measurement system is to be able to compare Multi-antenna/MIMO devices and algorithms based on their throughput performances when tested against spatial channel models. All critical parts of the mobile terminal design (antennas, RF front-end, baseband processing) are tested in an end-to-end configuration. This document aims to present all the most relevant technical features of the MVG StarMIMO system relative to the MIMO OTA measurement technique with anechoic chambers. 3
StarMIMO setup The StarMIMO main hardware components are highlighted in Figure 1: Data Acquisition & Processing PC Radio Communication Tester Spatial Channel Emulator MIMO Amplification Unit MV-Cal TM Calibration Unit Motion Controller Figure 1. StarMIMO setup In a MIMO OTA setup, signals come simultaneously from different directions around the device. This characteristic setup, combined with a channel emulator, enables the emulation of spatial-temporal propagation environments at the Device under Test (DUT) location. 4
Multi-probe technology N number of transmitting antennas (probes) are connected to the spatial channel emulator(s) in order to emulate a complex multipath environment at the DUT location. Figure 2 shows the StarMIMO H setup: SINGLE CLUSTER vs MULTIPLE CLUSTER SETUPS As previously introduced, the probe locations play an important role for the MIMO OTA test range. Probes can be located on a single sector or on a full circle, namely single cluster and multiple cluster approaches respectively. Figure 4, and 5 show a single cluster setup with 4 dual polarized probes and a multiple cluster setup with 8 dual polarized probes respectively. Reference position (0 deg) Figure 2. StarMIMO H test range The probes are dual polarized and there is access to both polarizations at the same time. The frequency bands of the probes are from 400 MHz up to 6 GHz. Figure 4. Single cluster Probe locations The design of the probe is shown in Figure 3: Figure 3. Probe from 400 MHz to 6 GHz Reference position (0 deg) Main parameters of the probe are specified in Table 1: PROBES REF: DP400-6000 Frequency Bands 400 MHz - 6 GHz Return Loss < 11dB Gain -10 dbi +2 dbi Cross Polarization Coupling* < -35 db to -40 db Table 1. Main parameters of the probe. *Probe is calibrated and anechoic environment is optimized. Figure 5. Multiple Cluster- Probe locations Details about the channel models used for each of the above implementations are shown in the following chapters. It can be considered that the complexity of the setup is reduced with the single cluster. 5
TEST VOLUME vs NUMBER OF PROBES The number of probes plays an important role when understanding the test volume and hence the quiet zone of the test range. The test volume is limited by the number of probes for multiple cluster setup and the size of the sector for the single cluster setup. Consequently, a high number of probes extend the ideal performance. This is demonstrated by the plot of correlation coefficient vs. distance between antennas in wavelength at 2 GHz in Figure 7: 8 PROBES For a full circle implementation (multiple clusters) with uniformly separated probes, the general rule the more probes, the larger the test volume is valid since the whole angular domain is sampled. Figure 6 shows the DUT size vs. frequency for a uniform channel with respect to the number of probes used in a multiple cluster setup (full circle): 16 PROBES Figure 6. Max DUT size vs Frequency and number of probes for a uniform distribution of probes (multiple clusters) and uniform channel model Figure 7. Correlation Coefficient for a uniform distribution of probes (multiple cluster); Uniform channel model 8 probes vs 16 probes test range Test volume is approximately 0.8l to 2.2l for the 8 probes and 16 probes cases, respectively. The single cluster setup is a special case with a limited angular coverage of the full angular domain and therefore there is no general rule for the relation between the test volume and the number of probes. There is a big impact on the channel model as the large angular spread requires the OTA probes to be located in a 6
wider sector than the small angular spread. This method will better sample the Laplacian shaped cluster power angular spectrum. For example, in the case of +/-45 and +/-55 sectors the Laplacian function of 35 angular spread is better sampled using the sector of +/-55. Figure 8 below (Laplacian PAS with 35 degree angular spread) shows that 1/6 of the power is located > 45 degree and < -45 degree angles. Spatial channel emulator A test signal from a radio communication tester, goes through a spatial channel emulator which emulates the radio channel according to a pre-defined channel model. The most typically emulated parameters are path loss, multipath fading, delay spread, Doppler spread, polarization and of course spatial parameters such as Angle of Arrival (AoA), and Angular Spreads (AS). CHANNEL MODELS The channel models used for MIMO OTA testing are Geometry-based Stochastic Channel Models (GSCM) in which the radio channels are defined by: Transmit Antenna location and pattern Propagation Characteristics (delay, Doppler, AoA, Angular Spread at the transmitter (ASD), Angular Spread at the receiver (ASA), and polarization) Mobile velocity and direction of travel Receiver antenna location and pattern A number of large scale parameters Figure 8. Power Angular Spectrum with 35 angular spread The larger the test volume, the more accurate the tails of the Power Distribution Function (PDF) have to be modeled. Table 2 reports some simulated figures of the Test Volume vs Number of probes when using Spatial Channel Model Extended (SCME) Urban Micro and Urban Macro channel models for both single cluster and multiple cluster setups: TEST UNIFORM PROBES SETUP VOLUME (MULTIPLE CLUSTER) 0.75l 8 2.25l 16 5.25l 32 TEST SECTOR (+/-45 ) PROBES SETUP VOLUME (SINGLE CLUSTER) 0.75l 3 0.75l 4 1.75l 8 TEST SECTOR (+/-55 ) PROBES SETUP VOLUME (SINGLE CLUSTER) 0.75l 3 0.75l 4 2.75l 8 Table 2. Test Volume Vs Number of probes when using SCME Urban Micro and Macro channel models These are measurement-based channel models that have all the dimensions in, space, time, polarization, and frequency. Space and polarization are crucial parameters for determining the spatial correlation. The GSCM family includes the 3 rd Generation Partnership Project (3GPP) standardized channel model, SCME, Winner II, Union Internationale des Telecommunication (ITU), and International Mobile Telecommunications (IMT) Advanced channel models. MIMO OTA TESTING CHANNEL MODELS In the StarMIMO setup, the clusters are simultaneously mapped to the probes so that the sum of the transmitted signals in the center of the array is as defined in the propagation model. The mapping is done in the spatial channel emulator. In this case, the geometry based environment of the channel model is transferred to the anechoic chamber. The most common channel emulators have a specific tool for generating the impulse response of each probe based on the number and location of the probes. 3GPP and the international association for the wireless telecommunication industry (CTIA) have agreed on a set of channel models to be used for the MIMO OTA testing: SCME Urban Micro-cell SCME Urban Macro-cell 7
3GPP TR 37.977 is the reference document for more details about the channel model parameters. Table 3 reports the parameters of the SCME Urban Micro-cell for the multiple cluster case: SCME Urban micro-cell CLUSTER # DELAY [NS] POWER [DB] AOD [ ] AOA [ ] 1 0 5 10-3.0-5.2-7.0 6.6 0.7 2 285 290 295-4.3-6.5-8.3 14.1-13.2 3 205 210 215-5.7-7.9-9.7 50.8 146.1 4 660 665 670-7.3-9.5-11.3 38.4-30.5 5 805 810 815-9.0-11.2-13.0 6.7-11.4 6 925 930 935-11.4-13.6-15.4 40.3-1.1 Delay spread [ns] 294 Cluster AS AoD / AS AoA [ ] 5 / 35 Cluster PAS shape Laplacian Total AS AoD / AS AoA [ ] 18.2 / 67.8 Mobile speed [km/h] / Direction of travel [ ] 3,30 / 120 XPR - NOTE: V & H components based on assumed BS antennas 9 db Mid-paths Share Cluster parameter values for: AoD, AoA, AS, XPR Table 3. SCME Urban Micro-cell channel model In the defined single cluster case, all AoA are assumed to be set to 0. Hence, the model has just one cluster in the spatial domain. In terms of the delay positions, the single cluster is the same as the multiple cluster. Other parameters such as Cross Power Ratio (XPR), direction of travel, and mobile velocity are also the same for both implementations. One option in the single cluster is related to the AS AoA, which can be set to 35 or 25 in order to enable a range of spatial correlations based on the type of the DUT. Radio communication tester The radio communication tester generates the wireless signals that are sent to the spatial channel emulator. In 2x2 MIMO OTA testing, two radio frequency (RF) outputs will be used while in Single-input Mulitple Output (SIMO) (RX diversity) testing, only one RF output is connected to the spatial channel emulator. The RF IN/OUT port is connected to the link antenna in order to keep the data connection and be able to measure the Block Error Rate (BLER) and hence the throughput for a given power at the DUT location. Amplification unit In order to increase the total power at the DUT location, each output of the spatial channel emulator is amplified before being sent to the DUT via the probe. A power level of - 50 dbm/15 KHz can be emulated at the DUT location when a channel model is simulated by the channel emulator. This power level is enough for testing LTE protocol at high modulation coding scheme (MCS) such as 64 QAM, and full bandwidth (BW) allocation. One amplification unit is composed of 8 channels/amplifiers. 8
Calibration The goal of the calibration process is to ensure equal responses from each output/path, both amplitude and phase, by correcting the errors caused by the setup, such as probe misplacement and differences in gain and phase caused by cables. Figure 9 shows the calibration setup: Data acquisition & processing PC SEVERAL PATHS S21 VNA Spatial Channel Emulator MIMO Amplification Unit MV-Cal TM Calibration Unit Figure 9. Calibration setup The total path loss is measured for each path (channel) from the input of the spatial channel emulator to the DUT location. The calibration is performed by using a static channel model, so called 1 tap model. Both the V and H components of the transmitted signal are calibrated. Usually a reference antenna with known gain characteristics is used, sleeve and loop dipoles are the main choice respectively for V and H components. Gain and phase compensation are then stored on the spatial channel emulator for each channel. All the active components in the test range circuitry are also calibrated by using this process. The calibration process is time consuming mainly due to the following drawbacks: 1. Dipoles are narrow band. 2. Dipoles are single polarized. 3. Active component circuitry, such as amplifiers, mixers, oscillators, etc. are temperature dependent making calibration necessary several times per week and before starting any testing campaign. 4. The probe array is not calibrated. The radioelectric axis of each probe should still be calibrated for high quality testing. All the above can be mitigated by splitting the calibration procedure in two parts and using the MVG patented MV-Cal calibration solution. 9
CALIBRATION - PART 1 All the active components in the setup are calibrated. MV-Cal closes the loop and measures the path loss on each channel. Gain and phase compensation are stored in the MV-Cal unit. The whole process takes only a few seconds. Figure 10 shows the components which are calibrated in part 1 of the calibration process: CALIBRATION - PART 2 This part of the calibration concerns only the probe array. The procedure is the same as for a SISO system in which calibration is performed approximately once a year and is usually part of a maintenance contract. VNA Spatial Channel Emulator MIMO Amplification Unit MV-Cal TM Calibration Unit Figure 10. Calibration Part 1 10
SAM MIMO Control software interface The StarMIMO setup, the components and third party equipment are all managed by the control software. The main tasks are: 1. Control the Spatial Channel Emulator(s), the Base Station Emulator, and Network Analyzer 2. Automatize the gain calibration process, part 1 of the calibration by controlling the MV-Cal unit 3. Upload the standardized channel models, i.e. SCME Urban Micro, SCME Urban Macro, and Winner II 4. Data acquisition and plotting of the Throughput vs. Power curve SAM MIMO CONFIGURATION The SAM-MIMO configuration interface will guide you through all the steps necessary to configure the used HW connected to the StarMIMO test range easily. Figure 12 shows the SAM MIMO configuration wizard: This control software is needed for a MIMO OTA test range turnkey solution. Figure 11 shows a screenshot of the software interface: Figure 12. SAMMIMO Configuration Interface Calibration coefficients can be uploaded by using the SAM MIMO configuration interface. Figure 11. SAM MIMO GUI Figure 13 shows how it is done: Figure 13. Uploading calibration coefficients In Figure 13, there are two sets of coefficients for the calibration that are independently uploaded on the SAM MIMO configuration as documented in the earlier section of Calibration. 11
Conclusion MIMO is all about correlation. Correlation has a major impact on the throughput performance of a multi-antenna wireless device. Since correlation is both a function of antenna characteristics and propagation channel, MIMO OTA testing enables direct measurement of the true terminal performances. AUTHORS Alessandro Scannavini, MVG Nicolas Gross, MVG The StarMIMO setup is capable of emulating any spatialtemporal characteristics environment at the device under test (DUT) location by utilizing MVG s multi-probe technology with a spatial channel emulator. This is implemented by placing the probes at different location and emulating the MIMO channel with the spatial channel emulator. In a single measurement, it is possible to characterize the performance of the device at the system level, including baseband, RF front-end, and antennas by measuring the throughput in downlink. StarMIMO provides an efficient end-to-end testing of MIMO devices and is a major asset in the design cycle and product validation of mobile devices. For more information about StarMIMO, please visit our site: http://www.satimo.com/content/mimo-testing-solutions Or email us: sales@microwavevision.com Graphic design: www.ateliermaupoux.com, pictures: all rights reserved