New wireless standard protocols such as HSPA, LTE, and WiMax are employing multi-antennas on both the terminal and base stations sides. Multi-input Multi-output (MIMO) technology is used in order to enhance system performance in terms of higher spectral efficiency, better Quality of Service (QoS), and good link quality, especially in a strong multipath and fading environment. MIMO system performances are strongly dependent on antenna properties and the channel environment. MIMO and multi-antenna terminals cannot be reliably tested with the current Single Input Single Output (SISO) OTA methodology, which is based on 3-D radiated power or received sensitivity pattern measurements with an isotropic weighting (uniform channel model) , .
In multi-antenna communication systems, radio channel characteristics play an important role. TRP (Total Radiated Power), and TRS (Total Radiated Sensitivity) do not take into account radio channel characteristics, and hence are not suitable for testing performances of multi-antenna terminals. 3GPP, and COST2100 are evaluating possible approaches in order to come up with a standardized way for radiated performance testing. New ways of OTA testing should include a realistic radio channel, spatial and temporal effects, and user effects.
One investigated approach, and approved by 3GPP as a candidate for standardization, is presented in , , . The Spatial Fading Emulation (SFE) technique relies on an anechoic chamber, antenna array, channel emulator, and a mapping of Geometry-based Stochastic Chanel Models (GSCM) by the channel emulator. Approved GSCM like TGn model , SCM(E) , Winner , or IMT-Advanced are currently used by 3GPP for conducted testing to assess the performance of a MIMO receiver in multipath and fading environments. However, this testing does not currently incorporate the antennas’ performance.
Realistic multi-path propagation environments can be emulated in an anechoic chamber with a number of transmitting antennas surrounding the device under test (DUT), all transmitting simultaneously, and fed by a channel emulator. The geometry of the antenna array can be circular or spherical. SATIMO’s test configuration consists of 8 antenna array elements surrounding the DUT in a circular geometry. A test plan has been drafted and proposed to COST2100, CTIA, and 3GPP for OTA throughput comparison measurements . In the future, measurements results from different laboratories using the same SFE approach will be analysed in order to address specific test range requirements such as accuracy, repeatability and testing time.
SPATIAL FADING EMULATION TECHNIQUE
As highlighted, SFE is being investigated in COST2100, for suitability of standardization in 3GPP for OTA testing of multi-antenna terminals.
Multipath and fading characteristics of a realistic propagation environment are created in a controlled environment, such as an anechoic chamber with a multi-channel fading emulator and an array of probe antennas surrounding the DUT.
Figure 1 shows SATIMO’s measurement configuration:
15 probe antennas are located on a circle with a radius of 45cm. Only 8 probes are used since the multi-channel fading emulator used could only accommodate 8 outputs. Antennas are spaced 45 deg apart. Horizontal polarization of each probe is used in order to fit with the chosen channel models.
The basic principle of the test set up is to create a specific propagation environment at the DUT. The channel emulator has its outputs (8 multiple channels connected to the 8 probes surrounding the DUT) create a multipath environment including delay dispersion, fast fading, path delays, and doppler shift. The multipath signals are then transmitted to the DUT via probes.
Widely standardized channel models like Winner, SCM, and SCME can be created by using the channel emulator. A network emulator is needed in order to set up a call with the DUT, thus measuring the DUT in its operational mode, without impacting antenna performances by adding extra cables. An external antenna is needed in order to loopback the uplink signal to the Network Emulator. Usually, it is placed in the mast where the DUT will be located on SATIMO’s set up.
All the effects of antenna configuration performance, receiver, DSP, and SW are all included in the performance test with SFE. User effects can be also taken into account as per SISO testing by using a proper standardized head, and hand phantom.
MEASUREMENT SET UP
As anticipated in the previous section, a test plan has been proposed. It provide laboratories with a test procedure to be followed in case the SFE technique will be used.
The basic procedure consists of:
• Defining channel models parameters (Winner, SCM, SCME standard channel models can be used)
• Creating channel impulse responses for the multi-channel fading emulator
• Calibrating system set up
• Setting up channel models power
• Executing the measurements
Due to the lack of both HSPA/LTE MIMO radio communication tester and test devices, OTA throughput testing has been performed on a HSDPA category 9 USB stick, with a 16QAM signal on the downlink. Details of the DUT and network emulator settings will be provided in the following sections.
A. Channel models conditions
For the preliminary stage, the proposed test plan implements the SCME TDL (Tapped Delay Line) in two scenarios, Urban Micro (UMi), and Urban Macro (UMa). Taps in UMi are distributed in delay and spatial domain as depicted in Figure 2.
Taps in UMa are distributed in delay and spatial domain as depicted in Figure 3.
Full details in terms of parameters for both UMi, and UMa can be found in , but for completeness the parameters have been listed in Figure 4. The Suburban Macro scenario was not considered by channel model experts to be as important as UMi, and UMa
The angular spread for both scenarios follows a Laplacian distribution, as per Figure 5.
B. Calibration Process
Figure 6 shows the setup used for calibration.
The calibration process consists of measuring the total path loss from the input of the channel emulator to the EUT location. Usually a reference antenna, with known gain characteristics, is used as the EUT. SATIMO used a sleeve dipole (SD) as the reference antenna.
The goal of the calibration process is to ensure identical responses from each antenna by compensating for errors caused by setup nonidealities (e.g. probes placement, cable variations, etc.).
For this purpose, a one tap constant channel model is used to transmit through each probe individually, using the maximum amplitude in the fading profile. The process starts by sending a signal from each probe sequentially(1,3,5,7,9,11,13,15) to the EUT located in the center of the probes. The amplitude and phase response is then recorded via a VNA (Vector Network Analyzer), as depicted in Figure 6. The weakest signal(higher path loss) will then be selected as the starting point for calibration. Compensation for the path loss differences is accomplished by adjusting the amplitude and phase weighting on the channel emulator for each output. The calibration process ends when the amplitude and phase adjustments have been stored on the channel emulator for each probe(8 in total).
Figure 7 shows a path compensation table:
It is worth noticing that the state of the channel emulator should be identical during calibration and measurement.
C. Link budget analysis
The link budget of the test range depicted in Figure 6 is detailed in Figure 8.
When the considered channel emulator is fed with the maximum input power of -15dBm, the power at the DUT is -42dBm.
D. Channel model power settings
Based on the proposed test plan, throughput in test mode is measured using three different power levels.
The power level refers to the measured power at the center of the probes when a constant channel model is transmitted via the probes. Power levels are set to -63dBm, -69dBm, and -75dBm by adjusting the amplitude compensation of each path, including the correction factors shown in Figure7.
E. Network Emulator settings
When testing an HSDPA device , the following parameters should be chosen properly:
• Downlink Modulation
• Fixed Reference Channel
• HSDPA physical channel relative power
Figure 9 shows the HSDPA channel power settings:
The goal of the testing campaign was to estimate the accuracy of the channel models, and to validate the reference test setup.
F. Spatial Correlation results
A Uniform Linear Array was used as the EUT. It is comprised of 3 dipoles with a resonant frequency of 2050MHz, spaced apart. Figure 10 shows the EUT setup
The spatial correlation has been measured and compared with theoretical values for angular spreads of 10° and 35°. The results are shown in table 1.
As can be seen, the spatial correlation matches the theoretical values when the angular spread is equal to 10°, and tends to diverge for the higher angular spread of 35°.
G. Feasibility Study
The objective of this study is to understand if 8 probes are sufficient for generating a field at the center of the array with a distribution that matches the theoretical distribution that would be obtained from an infinite number of probes (a Bessel function of the first kind). A SD2050 dipole has been stepped along the X-axis from - to with 1 cm steps. The impulse responses was recorded for each step using a VNA.
Figure 11 shows the results of the comparison, and good correlation is observed between the two curves.
H. Throughput testing
The test configuration is shown in Figure 6, except the VNA has been replaced by a radio communication tester capable of setting up a WCDMA/HSDPA call. As anticipated in subsection E, some parameters need to be configured on the network emulator. Mainly, the downlink channel was set to 10562(low channel), the fixed reference channel to H-set 3, and the power of the HSDPA physical channel as per the table shown in Figure 9.
The channel model power has been set to -63, -69, and -75dBm by following the procedure detailed in subsection D.
A USB stick (HSDPA category 9) has been used as the DUT, and placed at the center of the probes as per Figure 12.
Throughput testing has been performed while rotating the DUT with 45° steps. In order to be consistent with the SFE set up where the probes are located on the phi plane, the channel models have been rotated on the test setup, instead of rotating the DUT. In Figure 13, throughput versus angle of rotation is shown with both the UMi, and UMa scenarios implemented.
In Figure 14, average throughput versus channel model power is shown. The average has been calculated from the throughput results taken at each rotation step.
SATIMO is currently developing a commercial solution based on the SFE technique. In Figure 15 our StarMIMO system is shown. It will be the system which it will be used for the COST2100/3GPP/CTIA round robin testing .
StarMIMO can be equipped with up to 32 dual polarized probes. Only 8 dual polarized probes will be used for the round robin testing as it was agreed in 3GPP RAN4 .
SATIMO is also looking at an easy way to upgrade existing SISO OTA measurement systems to MIMO OTA testing range.
SATIMO’s proposed solution has two arrays of probes, one in elevation, and one in azimuth for standard SISO, and MIMO measurements respectively. Figure 16 shows the mentioned implementation.
CONCLUSIONS AND NEXT STEPS
This paper gives an overview of the Spatial Fading Emulation technique for testing multi-antenna terminals in a multipath and fading environment. It also discusses some practical considerations about SATIMO’s implementation of the SFE technique, while providing results of a testing campaign.
Attention was dedicated mainly to throughput testing. The results show that the throughput changes when rotating scenarios, highlighting the dependence of the DUT performance on the DUT’s antenna configuration(i.e., radiation pattern dependence). The feasibility study also shows that 8 probes are sufficient to generate a field at the center of the array with a distribution that matches the theoretical distribution that would be obtained from an infinite number of probes (a Bessel function of the first kind).
Further studies are planned to compare throughput results gathered from different laboratories using the same SFE technique in order to address accuracy and repeatability of results. The above assumes all the laboratories are going to follow the proposed test plan depicted in section III of the present paper. Further research is required to investigate the effects of using both polarizations of the probes in the multipath environment.
The authors would like to thank NOKIA OULU, and Elektrobit OULU for their support during the test bed configuration and the testing campaign.
 3GPP TR 25.914 v7.0.0 (2006-06), 3 rd Generation Partnership Project (3GPP); Technical Specification Group Radio Access Network, Measurements of radio performances for UMTS terminals in speech mode (Release 7), Tech. Rep., 67p.
 Cellular Telecommunication & Internet Association (CTIA), “CTIA test Plan for Mobile Station Over the Air Performance, Revision 2.2” , CTIA, Washington DC, Tech. Rep., Nov. 2006, 173p.
 Pekka Kyosti, et.al “OTA Testing for Multiantenna Terminals” , COST2100 TD (08)670, Lille France
 Joonas Krogerus, et al., “Towards an applicable OTA test method for multi-antenna terminals”, COST2100 TD(08)671, Lille France
 Per O. Iversen, et al., “Radiated Perfromance Testing of Diversity and MIMO Enabled Terminals”, COST2100 TD(08)618, Lille France
 V. Erceg, L. Schumacher, P. Kyristi, et al., “TGn channel models,” Tech. Rep. IEEE P802.11, Wireless LANs, Garden Grove, CA. USA, 2004, http://www.802wirelessworld.com/8802
 D.S. Baum, J. Hansen, G. Del Galdo, M. Milojevic, J. Salo, and P. Kyosti, “An interim channel model for beyond 3G system: extending the 3GPP spatial channel model (SCM),” in Proceedings of the 61st IEEE Vehicular Technology Conference (VTC ’05), vol.5, pp.3132-3136, Stockholm, Sweden, May-June 2005.
 IST-WINNER II Deliverable 1.1.2 v.1.2, “WINNER II Channel Models”, IST-WINNER2, Tech. Rep., 2008 (http://www.ist-winner.org/deliverables.html).
 Miia Nurkkala, et al., “Test Plan for OTA Throughput Measurements in a Fading Channel Environment”, COST2100 TD(09)964, Wien Austria
 3GPP TS 34.121-1 v8.8.0 (2009-09), 3 rd Generation Partnership Project (3GPP), Technical Specification Group Radio Access Network, User Equipment (UE) conformance specifications, Radio transmission and reception (FDD), Part 1 Conformance specification (Release 8).
 TR 37.976 “Measurement of radiated performance for MIMO and multiantenna reception for HSPA and LTE terminals (Release 10)”, version 1.1.0