I used to think it was hard to make money off something that doesn’t exist, until I spoke with Eric Starkloff, National Instruments’ executive vice president of global sales and marketing. We had set up time to talk about its annual Trend Watch, a view of what NI sees as the most important trends in test and technology.
We touched on lots of areas, from big analog data, to modular, software-centric and platform-based test. However, it was 5G, or more specifically the vagaries of it and NI’s ability to make a business out of that “nothingness,” that stood out. This business magic came into start relief in the lead up to, and the aftermath of last week’s MWC 2016 in Barcelona.
5G is in a similar spot to where the Internet of Things (IoT) was until recently, where, “when all’s said and done, there has been more said than done.” The difference being, of course, that 5G—at a fundamental technological level—is vastly more complex and as yet undefined. That, ironically, is why National Instruments has been able to cash in.
By adding communications IP to its LabView platform, NI set itself up as a virtual proving ground for 5G in order to show if the mathematics holds true for the many proposals as to what 5G should comprise. These proposals include massive MIMO 128-antenna arrays, new and more efficient waveforms, and useful operation in the millimeter bands at 60 to 70 GHz. At that region, there are wide swaths of spectrum to be had to increase data rates, but of course there is the problem of RF propagation and penetration dropping off dramatically for a given power output.
The uncertainty is such that a lot of prototyping is needed, and that’s NI’s calling card. “NI is unique from a business point of view,” says Starkloff, “in that we have a position were we are, frankly, selling and making money on this prototyping part of 5G, well before anyone is making any money on test systems—there's not even a market yet for 5G test systems.”
Of course, many companies are working on 5G, “but there aren't any test systems yet, so my peers in the industry, their message for 5G has been ‘when you need test equipment, we're going to be there.’ Our message is we'll do that, too, but we're there now, with systems that will help prove out 5G.”
Starkloff is confident that “every research vector in 5G right now is using NI’s technology to prove it out.”
It’s a big claim, but it’s in proportion to Starkloff’s Trend Watch report where NI’s technology approach is enabling everything from industrial IoT, 5G and big data, to leading the march toward platform-based test.
Here are some points from our conversation that will help put Trend Watch 2016 in perspective:
Massive change: Hot off a global customer tour, Starkloff said the No. 1 concern is the pace of change in devices that are being built. No shock there, but he made a good point in that the trick is to find a test strategy that can evolve at the speed of your devices. That means having to become more software-centric and automating earlier in the design flow. To help with this, NI is pushing forward on desktop testing and software-defined instruments for the engineering workbench.
Big Analog Data and Edge Analytics: We’ve discussed this at length in previous articles, but the points are still valid: Get more out of the data you get from testing of devices.
“Measurement data is really a goldmine and that there are incredible insights that can be gained from V&V [verification and validation] test to improve a design, as well as production testing from a quality management point of view,” he says.
Semiconductor test is ripe for analytics and big data techniques. For example, it’s already been shown that by using data analysis to identify ICs that are well within their design limits, a premium can be charged for those ICs, as they’re likely to be more reliable.
Edge analytics refers to the management of large data sets, such as the gigabytes of data acquired during system test. By putting analytics close to the measurement, or the sensors, you can avoid having to transmit all the data upstream and to the cloud.
Platform-Based Test: It’s inefficient to develop a test platform for each device or each generation of devices, so a platform approach is critical such that only software or minor hardware changes are required. This requires the definition of a common-denominator framework. Not so easy, as when speeds increase and there are faster rise times, the hardware becomes the limiting factor, but it’s a good goal.
Modularity: This has been NI’s calling card for many years, but as Starkloff said, it’s becoming more attractive as core technical capabilities and features become more common. The test philosophy needs to be more universal and like a platform-bases strategy, it shouldn’t be tied to a single device under test (DUT).
“There's things you can do architecturally in how you structure your test executive and test code modules that exploit that modularity… and still have the best possible ability to be extended to unanticipated changes in the designs,” says Starkloff. NI provides best-practice consultation on how to do this.
In production test, this modularity is particularly important, given impact the IoT has on design cycles. With more universal testers for hardware and software, you get a more flexible ramp up/ramp down of production, providing a level of flexibility which can make—or break—a manufacturing and production deal.