Fig 1. In this Azimuth Systems Field-to-Lab configuration, multiple basestations/sectors are mapped through the channel emulator to recreate real-world conditions.
Fig 2. Azimuth’s AzMapper software teams with its ACE MX channel emulator to provide RF environment playback that mimics the time-varying nature of the real world.
Fig 3. The octoBox, a small form-factor anechoic chamber, goes a long way toward facilitating OTA testing of conventional and MIMO wireless devices. Devices under test are situated in the center chamber, and the test antennas are at bottom. The upper portion is a master chamber housing test equipment or a partner device transmitting to and receiving from the devices under test.
Fig 4. Anritsu’s MT8820C is an integrated one-box tester that performs calibration and RF parametric and functional test for 2G, 3G, and LTE devices.
The advent of 4G/Long-Term Evolution (LTE) technology and multiple-input multiple-output (MIMO) antenna schemes in the wireless realm has brought significant changes in the way wireless devices and networks are tested.
In the past, simpler antenna configurations and lower-speed data protocols could be tested in a so-called “tabled” environment. Manufacturers simply plugged a cable into a port on the device and piped in an emulation of the wireless channel. It took the antennas completely out of the equation and went directly to the heart of the matter, which was receiver and transmitter performance as well as checking algorithms for cellular handoffs.
But alas, things are no longer so simple, and over-the-air (OTA) testing is now the order of the day. OTA test means that someone has to get into a vehicle with the prototype handset and physically evaluate it in field conditions where subscribers will actually use it. This is a rather expensive proposition to do once, never mind every time a design change is made to a handset or basestation.
This is just one of the changes in test methodologies that has been brought to bear by the move to 4G/LTE. In this article, we’ll look at this and other ways in which wireless test is changing and why.
Test Challenges Abound
One of the biggest issues in RF design for 4G/LTE is the ongoing spectrum fragmentation and the multiplicity of radios within handsets. Triband devices are required in Europe unless you’re staying put. The same goes for North America, only it’s three different bands. Asia Pacific? Can you say seven bands? This means significant design and test challenges for RF engineers in ensuring that performance is consistent across those multiple bands.
Also near the top of the list in terms of challenges is interoperability. In the United States, 4G is new enough that coverage is still a matter of “islands” centered on large metropolitan areas. Thus, handsets must be able to hand off seamlessly between existing 3G networks and the newer 4G/LTE networks. They must do so without dropping data connections or calls, and optimization is required on both the network and device sides of the link.
The issue of dealing with voice calls in smart phones also poses test challenges. The era of ubiquitous voice-over-LTE is at least a year or more out. Meanwhile, service providers are falling back on intermediate measures.
For example, Verizon uses its 1X network for voice while data is moved on the 3G and/or 4G/LTE networks. This two-radio approach can have a significant effect on battery life if design criteria aren’t exacting.
Data performance is itself a critical test area. With channel bandwidth of 20 MHz, data rates should theoretically reach 50 Mbytes/s. But what does that translate to in terms of real-world speeds? Many factors come into play here, including MIMO implementations and network issues.
Closely related is how handsets handle data retry for 4G/LTE. When the handset is trying to establish a data link to a server, does it relentlessly hammer the signaling channel until it connects, or does it bow out gracefully and retry after some interval?
The biggest change in the way mobile devices are tested is due to the arrival of MIMO technology. MIMO has brought a proliferation of test schemes that involve intensive field testing of devices.
The issue is how well you can translate conditions captured in the field to the testbench. This is critically important to network operators, infrastructure vendors such as Nokia and Siemens, and chipset makers such as Qualcomm.
So-called “drive testing” with mobiles reveals what most of us already know—depending on where you might be relative to the nearest basestation sites, calls may drop or data throughput may fall off. But recreating this environment in the test lab, replete with multipath reflections and Doppler shift, is exceedingly difficult.
For example, on a drive through midtown Manhattan, your handset might be within range of a basestation a block or two away at one moment and suddenly have that signal blocked by large buildings. If the algorithms in the handset do not embody sufficient intelligence, it can result in a dropped call. “These types of scenarios can’t be modeled with statistical channels,” says Erik Org, Azimuth’s senior marketing manager.
One solution to this conundrum is a product such as Azimuth Systems’ ACE MX MIMO channel emulator (Fig. 1). As part of Azimuth’s “Field-to-Lab” methodology, the ACE MX channel emulator can be used to prepare and replay drive-test RF data (see “MIMO Channel Emulator Selected For Testing 4G Wireless Broadband” at www.mobiledevdesign.com).
MIMO Paradigm Shift
“Going from 2G/3G to MIMO was a paradigm shift in test,” says Ajay Patel, executive director of product management at Azimuth Systems. Test of 2G/3G devices was somewhat simpler. You might still use test gear such as the ACE MX channel emulator to simulate the outdoor environment, but only in development of the radio chipset. After that initial silicon development phase, 2G/3G devices could be tested with a simple RF cable and attenuator.
The test scenario becomes more complex in a MIMO environment. “Typically, it’s done in two stages,” says Mike Barrick, business development manager at Anritsu. The first stage is to take measurements in a cabled environment to establish a performance baseline for the receiver front end and baseband processing. “The idea is to measure not IP throughput, but rather RF throughput,” says Barrick.
Then, most test labs add external signal fading to get a feel for how throughput varies under changing signal conditions, albeit still in a cabled environment. “I’ve seen it suggested that all LTE MIMO devices need external RF connectors on them for this purpose,” says Barrick. “Some phones actually require test engineers to drill holes in them to gain access to the signals.”
The second stage repeats the RF measurements, but now they’re done in an anechoic (non-echoing) chamber, duplicating the environment you had in a cabled environment. Having established that performance baseline in the cabled testing, the effects of the antenna are considered in this stage.
“In doing OTA testing, you can tell if your MIMO antenna implementation is affecting performance positively or negatively,” says Barrick. Additionally, adaptive multiple antenna techniques, including TX and RX diversity, spatial multiplexing, and beamforming, involve sophisticated closed loop algorithms that must be tested under a range of controlled (emulated) channel conditions.
There are also uses for the opposite scenario from the anechoic chamber, which is a reverberation chamber. In such cases, mechanical “stirrers” (metal bars) are waved around to simulate channel fading. In either case, external channel emulators are added to provide higher-power delay profiles, faster Doppler shifts, and multipath-fading correlation.
What drive testing—and capturing of the RF environment with equipment from vendors like Azimuth—accomplishes is mapping of the resulting data to create static tests that are completely controllable. Part of Azimuth’s Field to Lab solution, the AzMapper analyzes field data and generates a playback file that the ACE MX channel emulator can replay in OTA fashion (Fig. 2).
The ACE MX channel emulator provides not only the ability to repeatedly recreate the RF environment captured in the drive test, but also to contrive scenarios to really stress the device under test (DUT). “You can’t build statistical channel models for some things,” says Azimuth Systems’ Ajay Patel. For example, drive testing could be conducted in an extreme RF environment such as Manhattan and then doctored in the playback to look much worse than in real life. “Then you’d know that your handset can really survive New York,” says Patel.
Channel emulation techniques are applicable in various areas of research, development, and design. It is used by chipset makers such as Intel, Broadcomm, and Qualcomm and by makers of hardware on the infrastructure side, such as Motorola.
Why MIMO Poses Challenges
As wireless protocols advance, data throughputs rise and designers will turn to higher-order MIMO implementations. These days, the most common setups are 4x2 and 2x4, yet designers are venturing up as far as 8x4 in some implementations.
But before long, there will be a need to test an 8x8 MIMO configuration and emulate an environment in which multiple handsets are connected to multiple basestations at the same time. In such a scenario, there could be multiple uplink and downlink data streams simultaneously. All must be modeled at the right bandwidth and have the horsepower in the channel emulator to model and control them.
Azimuth Systems is about to release a software upgrade for the ACE MX emulator to 50 MHz for support of 8x4 MIMO. Where does a 50-MHz requirement come from? Everything in wireless is moving toward wider bandwidths.
For LTE-Advanced, standards committees are looking at 100-MHz channel bandwidth. This is achieved by aggregating up to five 20-MHz carriers. For wireless local area networks (LANs), it’s even worse, as bandwidths of 160 MHz are being considered.
“Thus, test equipment has to have even wider bandwidths to be able to look at signals outside the channel, due to interference, mixing, and intermodulation,” says Jan Whitacre, LTE marketing program manager at Agilent Technologies.
Higher-order MIMO implementations such as 8x8 pose particularly thorny test challenges. Anything beyond 8x4 falls into the realm of LTE-Advanced, which is likely to be three to four years away in terms of commercial deployment.
But that’s not stopping some test companies from trying to prepare for 8x8 MIMO. At the recent NIWeek conference and exhibit, National Instruments demonstrated a prototype implementation of an 8x8 MIMO LTE-Advanced downlink (see “Getting Ready for 8x8 MIMO”). With this prototype, National is clearly positioning itself for bleeding-edge research into LTE-Advanced implementations.
Integration Comes To LTE Test
For design teams navigating the waters of 4G/LTE test, pulling together all the elements of a test platform can be daunting. Fortunately, a good amount of integration has already taken place in the test vendors’ offerings to ease the task somewhat.
A recent example is the octoBox, a small form-factor OTA test platform from octoScope (Fig. 3). The octoBox facilitates OTA testing of conventional and MIMO wireless devices in a customizable refrigerator-sized anechoic enclosure that does the same job as large and very costly walk-in anechoic chambers and facilitates test setup.
The octoBox comprises a miniaturized OTA test setup in a compact self-contained unit. The system provides stable and controlled measurements that reduce development and test time while increasing test coverage and product reliability. For fast production, the octoBox supports simultaneous testing of up to eight small devices, like smart phones, and simultaneous testing of multiple radios in a single device.
In addition to measuring a DUT’s 3D antenna pattern, the octoBox can integrate test equipment, including a MIMO channel emulator, interference generators, RF sensors, data monitors, and other instruments. It supports a wide frequency range from 700 MHz to 6 GHz and can be used for testing multi-radio smart phones, with radios that include cellular, 3G/4G, Wi-Fi, Bluetooth, and GPS, all operating in different regions of the spectrum.
Another example of integration in test equipment is Anritsu’s MT8820C one-box tester, a multi-format instrument that performs testing for 2G, 3G, and LTE devices (Fig. 4). Among its capabilities are calibration, RF parametric testing, and functional testing, including call processing or no-call-based testing. Parameter setups and pass/fail limits for tests defined in 3GPP 6.521-1 are preprogrammed into the instrument, including automatic setup of uplink and downlink bandwidth allocations.
Anritsu has further enhanced the utility of the MT8820C by teaming with two vendors. In one case, ETS-Lindgren has integrated support for the MT8820C in its EMQuest EMQ-108 MIMO OTA test package. ETS-Lindgren also has released an instrument driver for use with its EMQuest EMQ-100 antenna measurement software. This software not only controls the instrument, but also can position the measurement antenna relative to the DUT to measure 3D radiation patterns.
In another instance, SATIMO offers support for the MT8820C tester with its MV-Scan multi-probe systems, which are used to emulate multipath propagation and to perform MIMO tests in a controlled and realistic environment.