Friday, February 8, 2019
As 5G networking inches closer to reality, one of the more stubborn problems also will be one of the smallest. Several issues have yet to be cracked with beamforming and massive MIMO antennas, which will make millimeter wave (mmWave) spectrum—a key ingredient in 5G networks—work on multiple devices and base-station locations.
Millimeter wave is problematic yet promising. Between bands 30 Ghz and 300 Ghz, mmWave promises high-bandwidth point-to-point communications at speeds up to 10 Gbps. But the signals are easily blocked by rain or absorbed by oxygen, which is one reason why it only works at short ranges. Beamforming is a way to harness the mmWave spectrum by directly targeting a beam at a device that is in line of sight of a base-station. But that means antennas in devices, and base-stations on network infrastructure, have to be designed to handle the complexity of aiming a beam at a target in a crowded cellular environment with plenty of obstructions.
It’s increasingly clear that much of the responsibility for making the high-speed, low-latency 5G networks work will fall to these overactive antennas. But their role is so complicated that few designers or evaluators are confident they know what the antennas should do, let alone how to verify the results. In fact, few designers are confident they know where to place or how to design the antennas, and even fewer understand how to tell whether one is working as it should.
“A 5G MIMO antenna is hard to test because it never stops changing,” said Larry Williams, director of product management at ANSYS. “They can’t be passive like traditional antennas that broadcast in every direction, so other things can connect with them. The 5G version doesn’t broadcast. It points a beacon at one object and makes an individual connection to 3, 8, 10—a thousand objects simultaneously.”
That makes testing all the more difficult because everything is in motion. “You can’t put them in a shielded room and test the connection from a distance because it realigns itself constantly,” Williams said. “If an object was 10 feet away at an angle of 0 degrees, and it’s now at 30 degrees, performance would drop off at zero and improve at 30. The way it connects to everything is completely different from 4G. It constantly changes, so you have to test it in a completely new way, too.”
That’s just one piece of this complex puzzle. “With 5G phones you have MIMO (multiple-input, multiple-output), carrier aggregation, essentially bandwidth-on-demand,” said Michael Thompson, a senior solutions architect at Cadence. “And because high-frequency signals can be blocked if you put your hand in the wrong place—probably two or three patches in different places on the phone—you have to look at which arrays are going to interfere with each other, or which beams. Plus, you might have transceivers at the bottom end of the phone, antennas at the top, and your signal that has to move across this flexboard. And you have to try to keep the power steady and get the signal as close as possible to the antenna to transmit it, even though impedance can change wildly across the flexboard. Carriers also want to sell this same unit in a lot of places that use different networks and frequencies. But if you had 30 output filters on a 4G phone to deal with different frequencies, in 5G you’re going to have 100. So even if nothing else changed, the number of channels went up 3X, which means test times go up dramatically.”
Sizing the problem
The range of 5G connections is so short that design teams need to make radio-testing rooms longer so they can find out how far those mmWave signals actually can travel. In addition, ultra-short distance networking creates changes that require physical changes in chip or system design, not just different approaches to validation, said Susheel Tadikonda, vice president of engineering at Synopsys’ Verification Group.
The high-frequency/short-distance problem, which could limit a 5G cell to a width no greater than the distance between poles on a city street, puts pressure on a system to track and maintain all those connections, Tadikonda said.
In addition, 5G puts a very high priority on almost-zero latency, which implies the need to rely more on silicon for complex processing and less on software.
“That could mean you have to move processing of beamforming and MIMOs and all that stuff out of software and put it into the PHY layer,” Tadikonda said. “Before, you could afford to do that kind of thing in software and send the results up the chain. But here we have a front-end architecture with a tightly coupled requirement for very low latency. You have to put a lot more intelligence up front to handle the beamforming and all that stuff quickly and make sure the antenna and the digital RF portion are also tightly coupled. In 5G you’re going to have to deal with having more than one antenna and figuring out where to put it, but there really is a need here that is going to make the PHY layer more complicated.”
The almost unlimited variety of devices likely to become part of private or public 5G networks will cause a variety of design issues, which in turn will be exacerbated by the wide variety of frequencies likely to be considered 5G to one degree or another.
“Another problem is that 5G basically makes every phone in the world obsolete, so a LOT of people want one,” Tadikonda said. “Demand is high, and pretty soon you’re wondering if there is enough test equipment in the world to make enough parallel lines to test everything and get it out the door in time because Christmas is coming and providers want to sell 100 million phones, so don’t waste time.”
This becomes even more imperative as the amount of data generated by ubiquitous arrays of sensors increases. Some of that data will require significant processing, and not all of it can be processed in place. “The key is how to process data and how to aggregate data,” said Mike Fitton, senior director of strategic planning at Achronix. “If you think about a 12-hour flight on an aircraft, that’s going to generate 844 terabytes of data. Some of that data can be used to reduce fuel consumption and maintenance, but where do you process that data?”
At least some of it has to be processed in the cloud, and that requires a very fast connection. “5G can do a whole bunch of different stuff, whether it’s peer-to-peer for V2X in autonomous driving, V2Cloud, V2V, V2Infrastrcuture. You certainly want to do as much processing as close to the source as possible, but you’re always wrestling with power. Where is the equilibrium point? It may be an aggregation point rather than the cloud, but you need power to process all of that data.”
This is why 5G is getting so much attention, and in this scheme the antenna is one of the major challenges. Frequencies don’t present as wide a variety of challenges as the question of where on a device to put the antenna, how many are needed, or how to keep them from interfering with each other. And everything must be tuned to integrate smoothly with 5G hardware, which could involve several carriers, many manufacturers, and could reflect pre-standard, early standard or final standard 5G interoperability standards.
“There may be so many conditions to test that you can’t test them all, which is worrying,” Thompson said. “It’s not going to take long to get to something they’ll call 5G in some things, or to the point that you’ll have to figure out how to move something through to manufacturing. There may be an order of magnitude greater complexity here than in the move from 3G, but people I talk to in design and manufacturing think there are techniques and approaches to get products out the door, so the 5G products will get there before too long.”
Lots of notice
Not all of this is entirely surprising, however. Millimeter-wave networks are not a novelty for which brand-new solutions have to be developed, argues Mike Demler, senior analyst at The Linley Group. Descriptions of point-to-point microwave relay networks provide plenty of insight into benefits or problems with wireless networking at the high end of the spectrum, he said.
The U.S. Federal Communications Commission (FCC) already has held an auction for licenses on the “low” 5G 60MHz spectrum, and it completed another round on Jan. 25 for 5G mmWave spectrum at 28GHz. The agency won’t announce the winners until after the 24GHz mmWave auction that starts in March, but it will auction licenses for three other 5G frequencies later in 2019—at 37GHz, 39GHz and 49GHz.
“The successful conclusion of our nation’s first high-band 5G spectrum auction is a significant step toward maintaining American leadership in 5G,” FCC Chairman Ajit Pai, stated last month. “The FCC will continue to aggressively push more spectrum into the commercial marketplace.”
The FCC allowed unlicensed use of some parts of the spectrum, especially below 60GHz, and issued licenses at low cost for microwave-network connections with proper registration, in the 71 to 95GHz bands, where users reported data rates as high as 10Gbit/sec, according to a 2008 study by Communication Infrastructure Corp. That study found microwave relays with data rates as high as 10Gbit/sec. It also reported degradation of those connections due to ambient oxygen—which absorbs electromagnetic radiation in the 60GHz range—but that rain was more of a problem, making 60GHz links unreliable in any area with substantial rainfall.
Copyright © 2019 CST, Inc. All Rights Reserved