By Nigel Wright and Mike McKernan of Spirent Communications
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. These new techniques impose significant challenges on the design and development of wireless devices, greatly complicating the associated RF testing.
With multiple antennas used to differentiate multiple incoming faded signals, it is necessary to test actual reception over the air (OTA). Field OTA testing can record results at a specific time and place, but cannot produce the required statistically meaningful data sets.
A better approach is to identify real-world RF characteristics and re-create them in the laboratory using RF channel emulators and a chamber to provide a realistic, controllable, and repeatable test at reasonable cost. More has to be learnt about the OTA behaviour of MIMO systems before it will be possible to create virtual OTA models, so current laboratory testing is performed physically in two forms: the one using a large anechoic chamber and the other using a smaller and less costly reverberation chamber.
This article will discuss some basics of MIMO as well as the need for (and challenges of) MIMO-OTA testing and how successful it can be in emulating real-world propagation conditions.
Requirements for MIMO testing
MIMO operation is based on the idea that receiving antennas can distinguish between separate data streams transmitted in the same frequency band as long as there is some spatial (directional) difference in the routes taken from transmitting antennas to receiving antennas.
This is a tall order in a pocket-sized smartphone, but it is nonetheless a fundamental part of LTE. An MxN MIMO system – ie with M transmitting and N receiving antennas can at best increase maximum data rates by min{M,N} times that available from a Single-Input Single-output (SISO) system. For example, a 4x2 MIMO would double the data rates while a 4x4 MIMO system would quadruple them given ideal conditions. Real-world conditions, however, reduce that advantage.
SISO device testing typically uses ‘conducted’ signals – connected by cable and bypassing the antenna. SISO antenna performance characterisation is less complex than in MIMO, so it is tested separately in an anechoic chamber to measure Total Radiated Power (TRP) and Total Receiver Sensitivity (TRS). For SISO devices, TRP and TRS are adequate measures, but MIMO performance is a function of so many factors ¬– including propagation environment, antenna design/orientation and baseband algorithms – that it needs a radically different approach to testing. The figure of merit most commonly used to differentiate between a “good” and “poor” MIMO device is data throughput under realistic operating conditions.
Such “realistic operating conditions” involve two factors: realistic faded signals, combined with realistic reception at the antennas. While channel emulators such as Spirent’s SR5500 or its new VR5 HD can create the appropriate signals, conducted-signal testing would not address MIMO’s dependence on the physical antenna design. This is why, with our increased reliance on MIMO techniques, radiated OTA testing has become critical and will become ever more critical as the number of antennas increase in future devices.
MIMO OTA testing needs to replicate the real world propagation environment. In the context of a relatively wide-bandwidth technology like LTE, it is important to emulate the spatial aspects of the wireless channel. MIMO OTA testing uses the 3GPP’s Spatial Channel Model (SCM) and Spatial Channel Model Extension (SCME) channel models for this purpose.
The models are defined with six RF paths, each representing the signal received after reflecting from a cluster of “scatterers” located near the device. Both the Angle of Arrival (AoA) at which the signals arrive at the receiver, and the angle spread (AS) must be modelled, in addition to the Angle of Departure (AoD) from the transmitter that also influences the channel throughput.
Instead of being received from uniformly distributed directions, each multipath signal component 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 the “Power Azimuth Spectrum” (PAS).
When the user is on the move, the signal path’s unique AS and AoA will be affected by Doppler shift, producing a unique Doppler spectrum. 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.
All of these variables need to be accounted for in order to create a valid MIMO OTA test.
Anechoic versus reverberation chambers
MIMO-OTA testing requires a signal source, one or more channel emulators and a number of transmitting antennas installed within a shielded chamber that can be either:
1. An anechoic chamber – typically a large chamber with RF absorbent walls to eliminate reflection. Although this can be a costly solution, it allows great control because the RF signals provided by the wireless channel emulators come directly from the antennas in the chamber, which can be precisely and repeatably situated.
2. A reverberation chamber – typically a smaller chamber that allows reflections from the chamber walls. Mechanically controlled reflective “stirrers” or “paddles” can be used to model dynamic conditions.
The two types of chamber offer different benefits. Although some useful, cost-effective MIMO-OTA testing can be performed in a reverberation chamber, more detailed testing requires an anechoic chamber that allows the generated field to be precisely controlled for better spatial fine-tuning in a controlled and repeatable manner.
Anechoic chamber testing
Taking the 3GPP’s Spatial Channel Models (SCM) as an example: modelling the SCM over the air requires properly matched statistical properties, notably the spatial correlation properties, defined in terms of 20 spectral components. One obvious approach would be to reproduce these components using 20 separate antennas, but this is not practical considering that the SCM actually contains many reflecting clusters, each consisting of 20 specular components. What’s more, the model must accurately represent the AS and AoA at the device.
After extensive research, Spirent has determined that it is possible to fully match these spatial correlation properties with as few as three antennas per reflecting cluster. Precise control of transmission parameters can accurately reproduce all the correlation properties, AS and AoA of the SCM-modelled signals. Additionally, Spirent’s research has shown that each transmit antenna in the chamber can be used to contribute signal content for the multiple reflecting clusters that make up the SCM model.
Although this does dramatically reduce the number of antennas required, it complicates matters in other ways that need to be addressed. For example, the time-domain characteristics of the signal need to match the SCM, and the signal received by the mobile must have the same level-crossing rate or “fading rate” as the SCM model. The limited number of antennas make this something of a challenge, but channel emulators can be used to create accurately controlled faded signals into each transmit antenna in such a way as to match both the correlation properties and the time-domain statistics of the SCM. The combination of generated signals must also be shown to have a Doppler spectrum resembling the target model.
Finally, in order to generate an accurate radiated SCM for devices with good polarisation diversity, both the vertical and horizontal components must be modelled and matched to the SCM reference model. Accounting for multiple vertically polarized antennas is relatively straightforward, but combining horizontally polarized signals is much more challenging.
All this would add up to a considerable exercise in mathematical calculation, and increased error risk, were it not taken care of by Spirent’s MIMO-OTA Environment Builder software. Used in combination with the Spirent VR5 HD Spatial Channel Emulator or SR5500 Wireless Channel Emulators, setting up the environment requires just three steps:
- Setting up the channel model parameters. This can be as simple as selecting a standard channel model from a drop-down box, then entering mobility and other parameters – otherwise customized models can also be created.
- Describing the in-chamber transmitting hardware. Probe-specific parameters are selected, and even greater accuracy is achieved by a one-time chamber calibration process that tells the software how to account for radiated path loss, etc.
- In-test editing. During test execution, on-screen controls allow dynamic adjusting of parameters in order to quickly see the effects of realistic environmental variations, such as editing angles of departure and arrival, as well as the total signal power received. On-screen graphics offer immediate feedback to minimize user error.
The MIMO-OTA Environment Builder provides comprehensive graphical summaries of the effects of the modelled channel as shown in the following two tab views.
The left side of the figure shows per-probe values of power distribution related to both the vertical and horizontal polarizations. It also contains a polar summary of the power angle profile for the given model overlaid on a representation of the actual antenna probes as configured in the chamber. The graphic to the right of this polar plot is map of the generated channel correlation as compared to the ideal narrow-band model. The two plots on the upper right-hand side depict the ideal and generated Doppler spectra, while the two on the bottom show the power azimuth spectra at both the base station and the device under test.
Spatial Channel Emulation with Spirent VR5
Emulating a MIMO channel requires MxN separate emulated radio channels, where M is the number of antenna elements at the transmitter and N is the number of elements at the receiver. If bi-directional or handover testing is required, the number of required links immediately doubles. Deployments using 4x2 MIMO are currently under development and will be followed by 4x4, 8x2 and 8x4 schemes. The number of channels required for testing already numbers in the dozens, but can be provided in a single 6U high hardware unit by the latest Spirent VR5 HD Spatial Channel Emulator.
Passive components such as splitters, combiners and duplexers are integrated within the unit, while both the system’s output power and dynamic range are designed to eliminate the need for outboard amplifiers.
The VR5 renews some of the hardware features that made its sibling, Spirent’s SR5500 Wireless Channel Emulator, an industry favourite: internal power meters at each input and output help ensure accuracy and repeatability, and feed real-time updates on the front-panel display. Controlled additive interference (AWGN) generated by the unit is also measured in real time and those measurements are displayed to the user.
New test equipment targeted to the wireless market should always be “future-proofed” by implementing an order-of-magnitude increase in the quality of RF specifications. With its considerable performance headroom – radical RF handling capabilities, including cutting-edge output power range, noise floor, and overall channel quality – the VR5 will easily handle test requirements for several years to come.
Particular attention has been paid to the user interface. As the technology gets ever more intricate, testers find themselves having to do more with fewer resources, usually in step with very aggressive product schedules.
The VR5 design team built an interface specifically designed to simplify control over a complex MIMO environment. The front-panel touchscreen offers sophisticated control with just a few swipes of the finger, all while minimizing the opportunities for user error.
During setup and configuration, the front panel presents a step-by-step process offering combinations of test cases, environment scenarios and operator parameters. Graphical configuration information is presented at each step to help the operator quickly recognize and correct setup errors. While most test cases can be configured strictly through this high-level control, the user still has the option to set lower-level parameters for customized testing. Most test cases, even complex high-antenna-count scenarios, can be set up in less than a minute.
During test execution the graphical feedback is sustained. The Channel Player shows real-time updates of the power and delay associated with individual fading paths. The VR5 adds the Temporal Player view, which provides real-time updates of selectable measured parameters such as C/N, input power, or output power. The Temporal Player is best used with the Dynamic Environment Emulation (DEE) feature, allowing the user to map out per-path fading parameters in a spreadsheet, varying over time. The emulator reads the spreadsheet file and physically creates the defined RF channel.
DEE can also be used to minimize the cost and time spent drive-testing mobile devices. Drive-route data can be captured and stored using a commercial cellular scanner. Optional Spirent software converts the captured data into a DEE file for playback in the lab. Aside from the convenience and cost savings of re-using data captured from a single drive route, this “virtual drive test” method adds a level of repeatability that is not possible by repeated physical drive-testing.
For special cases where statistically anomalous fast fading is required, a second fading engine is built into the unit. The fine-time engine, called Fading Lab, lets advanced users create RF environments based on sample-by-sample RF data parameters. For example, RF researchers can generate sample data by using commercial mathematical software (e.g. MATLAB), ray-tracing software or custom software. Channel-sounder data is another potential source of this fine-time information. The Fading Lab engine processes this data and creates the corresponding physical RF environment.
Ever-increasing levels of performance
Mobile users are demanding ever-increasing levels of performance that can only be met by costly investment in 4G technology. A key component of LTE and LTE-Advanced platforms is MIMO multiple antennas, and realistic testing of MIMO devices presents an enormous challenge.
Whether testing is carried out in an anechoic or a reverberation chamber, the VR5 HD Spatial Channel Emulator, with software packages for building the environment and emulating dynamic conditions, allow even inexperienced operators to run sophisticated tests without error. Continual graphical feedback during execution is further insurance against misleading results.
Although the final test will always be in the mobile user’s hands, Spirent’s test expertise provides the best insurance that the test will be passed with flying colours.
Nigel Wright, Mike McKernan Spirent Communications