The role of multiple input multiple output (MIMO) in delivering the higher data rates promised by LTE is not widely understood outside the hard-core technology community. MIMO makes use of multiple antennas on both the basestation and device to help achieve the higher data rates of LTE over current 3G UMTS technologies, but it also adds a lot of additional complexity. Most initial LTE deployments make use of 2x2 MIMO (2 antennas on the basestation and 2 on the device), but they also require additional complex processing of signals to deliver real performance gains.
The performance of MIMO is very sensitive to the implementation of the antennas, the environment in which the device is being used, and even the orientation of the device itself. This means that a small change in the position of the device when in use can result in significant change in data rate. For example, the user experience of a streaming video could go from excellent to marginal or poor just because the device was moved slightly during use.
Although many early LTE deployments are making use of 2x2 MIMO, future deployments may use 4x2 or even 4x4. Some field trials of TD-LTE are already using 8x4! While it is relatively easy to add additional antennas to a base station, it is much more challenging to add them into cramped smartphone designs. For MIMO to provide maximum gain, the antennas should in theory be spaced at least a half wavelength apart. Take the example of the 700 MHz LTE spectrum band used by Verizon Wireless in the US: half a wavelength separation turns out to be 200mm – much larger than most smartphones! As a result, creative approaches have to be used to implement MIMO antennas in smartphones, data dongles and other mobile devices where small size is a consideration. This can lead to design compromises that affect performance.
And the number of antennas on base stations and devices are only likely to increase with the evolution in technology. LTE-Advanced, which is only two or three years away from deployment, will use much more complex MIMO implementations to help deliver maximum downlink data rates exceeding 1Gbps. So the industry has been ramping up its testing of MIMO, first on base stations, more recently on devices. In order to precisely model the signals arriving at the MIMO antennas, fading emulators or channel emulators are used, which accurately emulate the characteristics of the radio channel between the base station transmitter and the device’s receiver antennas, or vice versa.
For MIMO to be effectively tested in the lab, the trends described above place more and more demands on the channel emulators used in testing. They must provide the most realistic possible emulation of the signals arriving at the MIMO antennas and also need to emulate more and more signal paths between the transmitting and receiving antennas as the number of antennas employed continues to increase. MIMO is a very complex technology, so setting up test conditions with lab-based equipment can be a very complicated and time-consuming task.
For one thing, MIMO performance depends on the spatial and directional characteristics of the radio links between transmitting and receiving antennas. Realistic replication of these spatial effects was not necessary with legacy 2G and 3G Single Input Single Output (SISO) designs, but it is critical in MIMO. This leads to an increased dependence on the physical design of antennas embedded in a mobile device. Since the effect of antenna design was significantly less important in SISO technologies, realistic testing of testing MIMO devices requires an approach called MIMO Over-the-Air, or MIMO-OTA. This technique involves an RF-shielded chamber that houses multiple transmitting antennas.
Control of the signals fed to these antennas is managed by a channel emulator such as Spirent’s new VR5 HD Spatial Channel Emulator. The calculations required to set up MIMO-OTA tests can be extremely complex and are best facilitated by specific antenna-mapping software, such as Spirent’s MIMO-OTA Environment Builder. The accuracy of the radiated signal is a function of the specific test methodology, which in turn is a function of the type of chamber available for use. MIMO-OTA has led to intense discussion within industry standards bodies, but in general, the industry has settled on a couple of approaches, using either anechoic or reverberation chambers.
MIMO’s complexity also places increased importance on other specific testing requirements: for many years the “holy grail” for RF engineers was a concept called “Virtual Drive Testing”, or “VDT”. In this technique, an RF environment is captured in a vehicle that is driven through the area of interest. The collected data is processed and used to re-create the environment in the lab, decreasing costs and liabilities while providing a new level of repeatability.
While this has been on the industry’s “wish list” for some time, MIMO’s dependence on so many new factors provided an increased level of importance on this technique, and the industry responded. New software tools are making virtual drive testing a proven reality in labs around the world.
About the Author
Nigel Wright is responsible for Spirent’s corporate marketing activities, as well as for expanding industry engagement of Spirent’s wireless product line. Since joining Spirent in 1998, he has also served as the vice president of client services, director of professional services for the wireless business segment, and general manager for the GPS testing business. These roles and Wright’s current position have provided him with unique insight into the needs of Spirent’s telecom operator, equipment and device manufacturer customers, as well as a first-hand look into market trends and issues. Wright is a telecom industry veteran who is frequently quoted in leading industry publications and has also spoken at a variety of high-profile events around the world.
Just to be clear, MIMO does not only apply to LTE. MIMO can also be used, for example, with any of the 3G WCDMA schemes.
I agree entirely with the point about antenna and device orientation, though. For MIMO to be effective, the propagation paths between transmitter and receiver, all of which share the same frequency channel, must be as uncorrelated as possible. Once you start seeing a dominant path emerging, the spectral efficiency of MIMO drops. It works best in pure Rayleigh channels.
Blog Doing Math in FPGAs Tom Burke 13 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...