5G is coming, but not everywhere, not all at once, and not the fastest version of this technology right away. In fact, the probable scenario is that 5G will be rolled out first in densely populated urban areas, starting in 2020 or 2021, with increasingly widespread adoption over the next decade after that.
But 5G is unlikely to ever completely replace 4G LTE, just as a smart phone today rolls from 4G LTE to 3G and 2G as reception decreases. Backward compatibility is an essential ingredient in all of these standards. 5G signals are very high frequency. The technology can scale to 300 GHz, versus 2.6 GHz for LTE. While that allows signals to carry significantly more data—basically scaling bandwidth to increase data density—the higher frequency also makes the signals more susceptible to interference from objects such as trees, buildings, and people. Even your own body can block millimeter-wave signals.
“5G challenges include the backhaul, siting, and spectrum,” said James Faucette, executive director of Morgan Stanley, in a recent speech. “With 5G, you need hundreds of times more base stations. 5G operates at a much higher frequency [than previous wireless standards], and when you get to millimeter waves they will barely cover a room. Signal unpredictability and how far they can go are big problems.”
None of this has stopped 5G development, but it certainly has affected the technology rollout schedule. The first implementations are likely to be fixed wireless, which are basically line-of-site. Millimeter-wave signals will not go through windows, so they will require an antenna. And there will have to be so many repeaters and small cells that site leasing will become a financial issue for carriers, said Faucette. “Companies will have to pay rent to a lot more people.”
Getting ready for 5G
This year’s Winter Olympics in Pyeonchang, South Korea, gave a hint of what’s ahead for this technology. The increase in bandwidth was on display for everything from virtual reality to 8K video, which doesn’t require special 3D glasses. Samsung even provided SmartSuits for skaters, which use sensors to map body position and send vibrating signals to wearable devices.
But it’s autonomous vehicles that are expected to really drive up demand for this technology.
“5G represents the fundamental technology needed for the autonomous driving experience,” said Steven Liu, vice president of UMC‘s Corporate Marketing Division. “The rising adoption of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I, V2X) means a continued increase in the number of vehicular radar systems. Technologies needed for these systems include the car’s anti-collision radars and global positioning systems, as well as sensors that will be needed to interact with stoplights and vehicle dispatchers. These will work in conjunction with existing systems, such as passenger comfort and infotainment control and engine monitoring subsystems that regulate temperature, tire pressure, and gas. Trucks for long-distance transportation will require systems for load leveling, load shifting, curve and wind shear, all working together to ensure that cargo is not damaged during transport and that the truck container is stable throughout its journey. All of these 5G communication applications will be essential for the system’s ability to perform its respective operations.”
In fact, 5G could be so critical to assisted and autonomous vehicles that it could alter the design of the electronics used within these vehicles. But that will likely depend at least partly on which is ready first.
“With the onset of electric vehicles and the onset of ADAS, connectivity in vehicles could become a 4G/5G story,” said Mike Rosa, director of strategy and technical marketing for Applied Materials‘ 200mm Equipment Product Group. “This is largely based on a timeline for each. As 5G comes online, the amount of silicon content could decrease because you have more memory in the cloud. That won’t be used for everything, but a large chunk of processing may be serviced across a 5G pipe.”
A tale of two technologies
5G comes in two flavors. One utilizes the sub-6 GHz band, which offers modest improvements over 4G LTE. The other utilizes spectrum above 24 GHz, ultimately heading to millimeter-wave technology. As a general rule, as the frequency goes up, so does the speed and the ability to carry more data more quickly. On the other hand, as the frequency increases, the distance that signals can travel goes down. The result is that many more repeaters and base stations will be required. That’s good news for the semiconductor industry, but it also means that rollouts are likely to take longer than previous wireless technologies because the amount of infrastructure required to make it all work will increase significantly.
“5G is very high frequency and lower noise, and it will enable new applications,” said Jamie Schaeffer, 22FDX program director at GlobalFoundries. “For base stations, it will require a digital front-end with data converters. And with 5G handsets, those will be low-power handsets where you will need to integrate a front-end module. But for things like facial recognition, 5G at 24 to 40 GHz will be the best solution out there.”
5G devices will be able to cobble together partial signals, too, using techniques known as beamforming and beam tracking, as well as Massive MIMO (multiple input, multiple output).
All of this comes with tradeoffs. As the frequency goes up, the thickness of films used for RF filtering goes down, and that creates another problem.
“At 2 to 2.5 GHz, the front-end RF film—which is generally an aluminum nitride-based film—is typically about 1 micron thick,” said Applied Materials’ Rosa. “As the frequency goes higher, that gets thinner and thinner. The process is harder to control for stress uniformity on an 8-inch and maybe a 12-inch wafer. So you increase the Scandium doping, but that only goes so far. Eventually you get to the point where you need to look at how you develop those films, which are sputtered today. In the near-term, this isn’t a big deal, but over time we will need to look at alternative ways to deposit those films.”
Even the materials used in those films will likely change. For example, lithium niobate is being suggested as a possible replacement for aluminum nitride because it can double the efficiency of electromechanical coupling. And while much of the switching today is done using silicon germanium, that could be replaced by gallium nitride in base stations, which will have to balance the need for increased power to drive more signals to more repeaters, and the rising power costs of the power itself.
Other technology issues
Nor is this confined to base stations. 5G handsets could run out of range quickly, leaving them searching for signals. That will deplete batteries faster than in strong-reception areas.
“With user equipment, there is a lot of work to be done on antennas so that when you hold the device, it will work,” said Sarah Yost, senior product marketing manager for software-defined radio at National Instruments. “There also are efforts underway to create efficient beam patterns for all those antennas. If you have 8 to as many as 64 inputs on a phone, the beam patterns will be very big. You may have 12 transmit and 12 receive patterns, and all of those might be different amplitude.”
That makes it time-consuming to test 5G chips using today’s equipment and methodologies. “Today, the test time is milliseconds,” Yost said. “If you add in all of these beam patterns and more capacity, you could see test times that are as much as 2,500 times longer. You still have to test those chips, but now you need a different approach. One that we’re looking at is over-the-air testing.”
The advantage of this approach is that it provides continuous testing for devices to optimize signals, but it’s a new concept in testing. “The advantage is that you can upgrade this as a modular platform so you can keep up with changes in the standard,” said Yost. “It allows the platform to act as part of a real network, which lets you make calls earlier in the design process.”
This may be combined with some version of external system-level test to speed up the process, as well as built-in self-test.
Planning for changes
Things don’t get much clearer on the design side, either. Unknowns make it difficult to optimize chip designs. As a result, flexibility needs to be added, both at the architectural level with a flexible fabric and layout, and at the logic level with programmability.
“People are wondering whether they will need bigger control systems,” said Sundari Mitra, CEO of NetSpeed Systems. “This requires a fundamental change in the architecture. There is more dynamic compute required, and that means a new level of sophistication in these designs. You can’t take the traditional fabric and mold it into 5G because it will require heterogeneous computing. It’s not just a single processor that needs access to memory.”
5G, viewed from any angle, is a disruptive technology by itself. But when it is paired with other disruptive technologies, particularly autonomous driving, the number of unknowns increases significantly.
“A car will need 5G connectivity all the time if it is going to be autonomous,” said Ty Garibay, CTO at ArterisIP. “These cars will be generating terabytes per hour of data. Some of this will be processed by edge chips. But 5G will be key to even relaying data post-processing. The challenge will be bringing together different types of processing and I/O together. It’s difficult for any human to hook it together well enough.”
And unlike previous generations of technology, 5G adoption likely will be a mix of technologies that will evolve over a long period of time. So while the rollout of was relatively quick, 5G handsets and base station coverage outside of cities could take decades. In fact, it’s not clear if this technology will ever be universal.
“If you look at all the 5G systems that are out there, they’re prototypes,” said Mike Fitton, senior director of strategic planning and business development at Achronix. “That’s why they’re all on programmable logic. The standards are changing and there are different applications emerging. So you need to build some programmability into ASICs. It you look at 3G and 4G, the early market for these technologies was almost exclusively based on FPGAs, which were then often replaced by ASICs to drive down the cost and power. We’re seeing the same in 5G. But 5G will take longer. The first part of this market will be UHF, then millimeter wave. So you’ll have a sea of 4G, then 5G will start showing up in islands, and it will move out from there.”
When that happens isn’t clear, though. Geoff Tate, CEO of Flex Logix, said that interest in eFPGAs has gone up significantly because of all of this uncertainty. “There is increasing interest in embedded FPGAs with 5G than before 5G,” he said. “Within a base station, there are power constraints. If you can get rid of the SerDes, that gives you big power savings. That’s important because the performance requirements are increasing. At the same time, there will be more base stations. Right now there is roughly one every half-mile, but with 5G the density of those stations will need to be greater. You’ll see one mother station and then daughter stations, which will have greater constraints on power.”
This bodes well for embedded programmable devices because they offer more flexibility than ASICs, and they are smaller and lower power than discrete FPGAs.
Achronix’s Fitton noted that the next rev of the 5G spec, release 16, will add some additional capabilities, as well. “You’ll start seeing this for IoT-types of applications, where you’ve got ultra-reliable low latency, and new user models will start to emerge.”
Technology in transition
Put in perspective, there are a lot of moving parts in the 5G ecosystem, beginning with the initial 5G rollout and continuing to future iterations of the technology. In effect, this is like managing a 3D matrix over time, where the pieces within the matrix are in various stages of research, development and even definition.
“With 5G, you have analog, digital and RF coming together,” said Ranjit Adhikary, vice president of marketing at ClioSoft. “Once you start using the technology, you get bugs that migrate as part of IP, and after a while nobody can say why something was done. People move around to different companies and a lot of the knowledge disappears.”
Adhikary said this has an impact on the IP development, optimization and characterization, which evolve alongside the technology. “With 5G, we don’t have much third-party IP right now. But we need to make sure as this rolls out that it’s all captured well from the system-level perspective, including scripts and flows. This whole trend started with cable modems, where the protocols and specs changed so fast that even over the course of a couple months it was difficult to keep track of the changes. And now you have more companies all over the world, so you have to keep track of which part you’re using, which version of the spec it was developed for, and which version of the third-party IP. If there is a new version of the spec, how do you link all the dots for whoever is using that IP?”
Uncertainty across multiple markets and technologies is becoming intertwined, raising more questions about how 5G technology will be used in the future, when it will commercially available, and ultimately how much it will cost in terms of dollars and other resources.
“All of these have additional requirements including endurance and cost, which still pose significant system design and manufacturing challenges when trying to achieve automatic electrical control in the future,” said UMC’s Liu.
5G is an important puzzle piece in these technology shifts, but exactly how big and when remains to be seen.