Introduction

Multiple-Input-Multiple-Output (MIMO) antenna techniques are a key factor in achieving the high data rates promised by next-generation wireless technologies such as LTE and LTE-Advanced. In addition to the challenges these new techniques impose on the design and execution phases of wireless products, MIMO adds several new wrinkles to RF testing. One area that experiences a renewed sense of importance with MIMO is the topic of Over-the-Air (OTA) device testing.

This article will discuss some basics of MIMO as well as the need for (and challenges of) MIMO-OTA testing. It will address the anechoic chamber-based MIMO OTA method and will briefly examine how successful this method is in emulating real-world propagation conditions.

Figure 1 - MIMO-OTA test configuration: Spirent mapping software, VR5 HD Spatial Channel Emulator and anechoic chamber. (Chamber photo courtesy of ETS-Lindgren)

MIMO Overview

The Need for MIMO-OTA Testing

Under ideal conditions an MxN MIMO system (one using M transmitting antenna elements and N receiving antenna elements) can increase maximum data rates by a factor of min{M,N}times those available from a Single-Input Single-output (SISO) system operating in the same bandwidth. In other words, a 4x2 MIMO system can (under ideal conditions) double the data rates available in a SISO implementation, while a 4x4 MIMO system might quadruple those rates. However, the real-world environment is not only far from ideal, it is far from static.

Most traditional SISO device testing is performed with a conducted signal. Test equipment is literally cabled to the transceiver of the device under test (DUT) by means of a temporary antenna connector, which effectively bypasses the device antennas. Supplementary SISO OTA tests are run in an anechoic chamber to characterize the antenna performance. These tests use two figures of merit to quantify antenna performance: Total Radiated Power (TRP) and Total Receiver Sensitivity (TRS, also known as Total Isotropic Sensitivity [TIS] by the CTIA). However, the results of OTA testing have been considered of secondary importance compared to conducted testing results.

However, this approach is insufficient when working with MIMO systems. MIMO device performance is especially sensitive to a variety of factors including propagation environment, antenna design/orientation and baseband algorithms. Modern channel emulators such as Spirent’s SR5500 or its new VR5 can deliver realistic conducted signals to MIMO devices, but the very act of cabling to the antennas impacts the effects of some of these factors.

MIMO systems are designed to take advantage of spatial diversity available in the propagation environment. The spatial diversity is quantified by the correlation between antennas, a function of both the propagation environment and the antenna patterns. Since antenna design and orientation are critical in MIMO systems, MIMO device evaluation is incomplete without the inclusion of antenna effects under realistic propagation conditions. This requires an OTA test method.

Challenges in MIMO OTA

While TRP and TRS are useful figures of merit when evaluating SISO devices, MIMO performance is a function of so many disparate factors that it must be evaluated in a slightly different way. The figure of merit most commonly used to differentiate between a “good” and “poor” MIMO device is data throughput, measured under realistic environment conditions.

A useful MIMO OTA testing method must accurately emulate the propagation environment seen in real-world network deployment. In the context of a relatively wide-bandwidth technology like LTE, it is important to emulate the spatial aspects of the wireless channel. The 3GPP standards for MIMO OTA testing call out the Spatial Channel Model Extension (SCME) channel models for this purpose.

The SCME models define a model of six RF paths, each representing the signal that might be received after reflecting from a cluster of “scatterers” located near the DUT. For a MIMO receiver, both the angle at which the signals arrive at the DUT (known as Angle of Arrival and abbreviated to AoA) and the angle spread (AS) are significant and must be modeled by the system. It is also important to model the Angle of Departure (AoD) of the transmitted signals, since this also influences the throughput of the MIMO channel.

A multipath component does not arrive at the DUT from uniformly distributed directions. Instead, each is spatially concentrated, resulting in a particular angle spread and a unique angle of arrival. The directional distribution of power per component is quantified as a parameter called Power Azimuth Spectrum (PAS).

Since each path can have a unique AS and a unique AoA, the mobility (direction and speed of travel) of the user produces a unique Doppler spectrum for each path. While the composite-environment Doppler spread may resemble the U-shaped spread seen in narrow-band channels, the per-path Doppler spread will retain their wideband characteristics. A further discussion of these characteristics may be found in a relevant article found in the Microwave Journal Technical Library2. All of the effects of these antenna pattern parameters dictate the correlation between device antennas, and all must be accounted for in the MIMO OTA method.

The Anechoic Chamber and MIMO-OTA Testing

Some very useful, cost-effective MIMO-OTA testing can be performed with a combination of channel emulators and a reverberation chamber, which allows reflections to propagate within the chamber. More detailed testing can be performed using channel emulators and an anechoic chamber, which allows the generated field to be completely controlled by the channel emulator, an arrangement which enables better spatial fine-tuning.

The anechoic-based MIMO OTA method provides a means of accurately emulating all the spatial aspects of a wireless channel in a controlled and repeatable manner. In order to create a Rayleigh-faded multipath spatial channel environment inside the anechoic chamber, a channel emulator and relevant mapping software (such as Spirent’s MIMO OTA Environment Builder) are used to distribute signals across both the horizontal and vertical elements of each transmitting antenna in the chamber. A detailed discussion of the anechoic chamber-based method is available Microwave Journal’s Technical Library.

Each chamber is equipped with a number (usually 4, 8 or 16) of cross-polarized antenna pairs, all of which are fed signals via the channel emulator. Figure 2 illustrates the distribution of power across 8 probes of 6 multipath delays. Each probe has a vertical (top) and horizontal (bottom) element.

Figure 2 – Power distribution across probes and multipath components in a typical MIMO-OTA test

Questions naturally arise around validating that the fields generated within the chamber are faithful to theory across a wide variety of relevant parameters. Proof points regarding the accuracy of emulation come from a software GUI that provides graphical displays of both theoretical and generated fields. Feedback from Spirent’s MIMO OTA Environment Builder software (used in a session with the new VR5 HD Spatial Channel Emulator) can illustrate how effectively the emulated channel models the theoretical idea.

Figure 3 plots both ideal and modeled correlation coefficients as functions of AoA.


Figure 3 - |Correlation coefficient| vs. AoA, ideal and modeled

In Figure 4, the six color-coded paths of the model as seen by the DUT are depicted in a pair of power delay profiles. The ideal model is shown in the left. The actual generated field (right) includes a Direction of Travel component (shown as a dotted red arrow) and the locations of eight probes (numbered blue circles).

Figure 4 - Power Angle Profiles of ideal model (left) and generated field (right)

As one last data point, the ideal (left) and generated (right) Doppler spectra are depicted in Figure 5. The six paths are once again color-coded, showing the rough U-shape of the composite field against the individual per-path Doppler spectra.

(5a)

(5b)

Figure 5 a and b - Ideal (top) and generated (bottom) Doppler spectra of a channel model used in MIMO-OTA testing

Conclusion

This article looked at the challenges involved in the end-to-end testing of MIMO devices.

Key discussion points and conclusions:
• The test cases and figures of merit on which the industry has relied for SISO device testing no longer provide sufficient predictable qualification as to the performance of a MIMO-capable device in real-world environments.
• The methods available in anechoic chamber-based testing lend themselves to control of fine details required to create important aspects of the real-world RF spatial field in the chamber.
• Graphical feedback from MIMO-OTA software can be used to validate the accuracy of environment emulation in the chamber.

Authors

Michael McKernan is a product-marketing manager for Spirent Communications’ Performance Analysis – Wireless business. Prior to joining Spirent in 2000, Mike spent many years in telecom and communications engineering management. Mike has a BSEE from NJIT and an MBA from Rutgers University.

Madhusudhan Gurumurthy is a Senior Applications Specialist at Spirent Communications. He joined Spirent in 2006 and has contributed in roles in Product Development, Product Marketing and Standards Strategy. He also represents Spirent at 3GPP RAN meetings. Madhu earned a BTech-EE from Pondicherry Engineering College and an MSEE from the University of Cincinnati, where he developed a MIMO channel-estimation algorithm as part of his thesis.

References
1. 3GPP TR 37.976: "Measurement of radiated performance for MIMO and multi-antenna reception for HSPA and LTE terminals".
2. Doug Reed, MIMO Over-the-Air Testing, http://www.mwjournal.com/BGDownload/Spirent_MIMO_OTA_Testing.pdf (March, 2010)