5G technology has a long way to go, but power consumption and cost issues alone are enough to hinder it

5G is getting closer, but it won’t be everywhere, it won’t arrive all of a sudden, and it won’t be the fastest version of the technology that’ll arrive first. What’s actually likely to happen is that 5G will debut first in densely populated metropolitan areas, starting in 2020 or 2021, and then spread gradually over the next decade.

Just as today's smartphones span three generations of wireless standards, 2G, 3G, and 4G LTE, 5G is unlikely to completely replace 4G LTE. Backward compatibility is an important consideration for all of these standards. 5G signals have very high frequencies, up to 300GHz, while 4G LTE is in the 2.6GHz band. Increasing bandwidth can increase data density, so the increase in frequency allows 5G signals to carry more data, but higher frequencies also mean that the signals are more susceptible to interference from trees, buildings, and human bodies. Our own bodies can also block millimeter wave signals.

[[223684]]

"5G technology has several major challenges to overcome: backhaul, addressing and spectrum," James Faucette, executive director at Morgan Stanley, said in a recent speech. "To use 5G, you have to deploy hundreds of times more base stations. 5G operates at much higher frequencies than previous wireless standards, and when you get to millimeter wave frequencies, 5G signals can barely cover a room. The unpredictability of the signal and how far it can travel are big issues."

Figure 1 Spectrum range

None of these difficulties have stopped the development of 5G, but they have certainly affected the promotion plan of 5G technology. The first implementation form of 5G terminals may be fixed devices in wireless form, basically wireless transmission. Millimeter wave signals cannot pass through windows, so an antenna is needed. So many repeaters and small base stations must be deployed, and the installation sites of these equipment need to pay rent, which will put considerable financial pressure on operators.

Figure 2 5G technology acceptance curve

Get ready for 5G

This year’s Winter Olympics in Pyeongchang, South Korea, revealed some of the promise of 5G technology. Everything from virtual reality to 8K video that doesn’t require special 3D glasses demonstrated the increased bandwidth of pre-5G technology. Samsung even provided SmartSuits for skateboarders, who used sensors to map body position and send vibration signals to wearable devices.

However, the application that will really drive the demand for 5G technology is self-driving cars.

"5G represents the foundational technology needed for the autonomous driving experience," said Steven Liu, vice president of marketing at UMC. "Autonomous driving technology requires more vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I, V2X) communications, which means that the number of radar systems deployed in cars will continue to increase, including technologies such as collision avoidance radar, GPS, and sensors that recognize stop signs and traffic police dispatch vehicle gestures. These systems will be combined with existing systems, including comfort control systems, infotainment systems, and engine monitoring subsystems that monitor temperature, tire pressure, and regulate gas. Trucks used for long-distance transportation require load balancing, load transfer, and curve shear systems that must work together to ensure that cargo is not damaged during transportation and that containers remain stable throughout the entire driving journey. 5G communications are critical for these systems to perform their respective operations correctly."

In fact, 5G is so important to assisted and autonomous driving that it could change the design of the electronics used in these cars. But how automotive electronics ultimately evolve will depend in part on which of the 5G technology and component suppliers are ready first.

“With the advent of electric vehicles and ADAS, 4G/5G will likely become the mainstream standard for automotive communications,” said Mike Rosa, director of strategy and technical marketing for Applied Materials’ 200mm equipment products group. “With the use of 5G technology, the electronics in the car may be reduced because there is more storage in the cloud. Of course, the cloud will not handle everything, but a lot of processing can be done in the cloud and then provided to the car through the 5G link.”

Two technologies

There are two options for 5G technology. One is the sub-6GHz frequency band, which is a slight improvement on 4GLTE, and the other uses frequencies above 24GHz, and the final form is millimeter wave technology. Generally speaking, as the frequency increases, the data transmission speed and the ability to transmit more data at a faster speed will also increase. On the other hand, as the frequency increases, the signal propagation distance is also reduced, and as a result, more repeaters and base stations need to be deployed. This is certainly good news for the semiconductor industry, but it also means that the launch time of 5G technology will be longer than previous generations of mobile communication technology, because more time is needed to deploy more infrastructure required for 5G communication.

"5G has very high frequencies, lower noise, and can enable new applications," said Jamie Schaeffer, 22FDX program director at GlobalFoundries. "From a base station perspective, a digital front end with data converters is required. 5G mobile phones need to integrate front-end modules and achieve low power consumption. For applications such as facial recognition, 5G technology in the 24GHz-40GHz frequency band is the best solution."

5G devices can use beamforming and beam tracking techniques, as well as massive MIMO (multiple-input, multiple-output) technology, to piece together multiple sets of split signals.

There are technical trade-offs behind all solutions. As the frequency increases, the thickness of the films used in RF filtering becomes smaller, which creates another problem.

“At 2-2.5 GHz, the thickness of the front-end RF films (usually aluminum nitride-based films) is typically about 1 micron,” said Applied’s Rosa. “As the frequency goes higher, the films get thinner and thinner. It’s difficult to control stress uniformity on 8-inch and 12-inch wafers with the current process. So the scandium doping step was added, but it has its limits. Eventually you find that you need to go back to the way you develop these films, and now they are grown by sputtering. In the short term, this may not seem like a big deal, but over time, we need to find some alternatives to deposit thinner films.”

Even thin-film materials may change. For example, it is now widely believed that lithium niobate may replace aluminum nitride because it can double the efficiency of electromechanical coupling. Most RF switching devices are now implemented using silicon germanium, but in base stations, on the one hand, you need to increase power to drive more signals to more repeaters, and on the other hand, you need to deal with the power itself. The electricity cost, so gallium nitride may replace silicon germanium.

Other technical issues

It's not just base stations that need to address power consumption issues. When 5G phones are searching for a signal after being disconnected, they will drain their batteries faster than if they were in an effective service area.

“A lot of work needs to happen on the UE antenna side so that it works when you pick it up,” said Sarah Yost, senior product marketing manager for software-defined radio at National Instruments. “Right now, the industry is still working on how to create efficient beam patterns for all those antennas. If you have eight to as many as 64 inputs on a phone, the number of beam patterns is very large. You might have 12 transmit modes and 12 receive modes, and all of those modes might have different amplitudes from each other.”

That makes testing 5G chips extremely time-consuming using today’s equipment and methods. “Today, testing is done in milliseconds,” Yost said. “If you need to test all these beam patterns and more capabilities, it takes 2,500 times longer. Testing these chips is essential, but you need to do it differently today, and we’re experimenting with over-the-air testing methods right now.”

[[223685]]

Figure 3: 128-antenna massive MIMO testbed developed by the University of Bristol and Lund University

The advantage of this approach is that it continuously tests the chip, thereby optimizing the signal, but in the test world, this is a new concept. “The advantage of this approach is that you can upgrade it to a modular platform to track changes in standards,” Yost said. “It makes the platform part of a real network, which allows you to make calls early in the design process.”

This approach can be combined with some external system-level tests to speed up the testing and built-in self-test process.

Plan for change

The current design level is not very clear. Too many unknown factors make it difficult to optimize chip design. Therefore, both in terms of architecture and structural layout, flexibility needs to be increased, and the logic level of programmable capabilities must also be flexible.

“People are wondering if they need larger control systems,” said Sundari Mitra, CEO of NetSpeed ​​Systems. “That requires fundamental changes at the architectural level, requiring more dynamic computing, which means that the complexity of these designs will increase. You can’t take a traditional system architecture and piece it together into a 5G system because 5G requires heterogeneous computing, and it’s not just a single processor that needs to access memory.”

5G is a disruptive technology in its own right from any perspective. However, when it is combined with other disruptive technologies such as autonomous driving, the unknowns increase significantly.

“When a car goes into autonomous mode, it will always need 5G connectivity,” said Ty Garibay, CTO of Arteris IP. “These cars will generate terabytes of data every hour. Some of this data will be processed on the terminal chip, and some of it will need to be forwarded to the cloud for processing. 5G will be key to forwarding this data. The challenge is how to bring together different types of processing and I/O, which is a big problem for anyone.”

Unlike previous generations of mobile communications technology, the adoption of 5G is likely to be a long-term evolution of multiple generations of technology. Therefore, although 5G will be launched in the near future, it may take decades to achieve 5G mobile phone and base station coverage outside of cities. In fact, there is currently controversy over whether this technology is universal.

"If you look at the so-called 5G systems today, they are just prototypes," said Mike Fitton, senior director of strategic planning and business development at Achronix. "That's why they all use programmable logic. Standards are changing, and some different application scenarios are emerging. So you need to build some programmability into the ASIC. The same is true for 3G and 4G. The early markets for these technologies were almost entirely based on FPGAs. After the market matured, they were replaced by ASICs to reduce costs and power consumption. The same is true for 5G, but it will take longer. The first stage of its evolution will be UHF, and then millimeter wave. So the market performance will be that 5G will just start as an island in the vast ocean of 4G, and then slowly expand its sphere of influence."

However, it is not clear when 5G will break out. Geoff Tate, CEO of FlexLogix, said that due to this uncertainty, the industry's interest in eFPGAs has increased greatly. He said: "With the emergence of 5G, people's interest in embedded FPGAs has continued to increase. Base stations have power limitations. If SerDes can be eliminated, a lot of power can be saved. Power saving is important because performance requirements will become higher and higher. In addition, 5G requires the deployment of more base stations. Now there is about one base station every half a mile, but the deployment density of 5G will be greater. A large base station will be followed by many small base stations, which will put greater demands on power."

This bodes well for embedded programmable devices, as they are more flexible than ASICs and smaller and consume less power than discrete FPGAs.

Achronix's Fitton noted that the next revision of the 5G specification, Revision 16, will add some additional capabilities. "You'll find that these new features are very suitable for IoT-type applications, where ultra-reliable low-latency characteristics are required, and 5G will enable some new use cases."

Technological evolution

At a high level, there are many moving parts to the 5G ecosystem, from the initial launch of 5G to future iterations of the technology. In effect, it’s like managing a 3D matrix that’s constantly changing over time, with various parts of the matrix at different stages of research, development, and even definition.

“In 5G, you have analog, digital and RF systems coming together,” said Ranjit Adhikary, vice president of marketing at ClioSoft. “Once you start using the technology, some bugs migrate around in the IP, and after a while, no one can figure out why something was done. People move from one company to another, and a lot of IP knowledge just disappears.”

Adhikary said that as 5G technology evolves, the development, optimization and description of IP will also be affected. "There are not many third-party IPs for 5G today, but we have to make sure that when a new version of 5G technology is released, all elements must be well tracked at the system level, including scripts and processes. The evolution of 5G technology will start with cable modems, because its protocols and specifications change so quickly that it is difficult to track its changes after a few months. In addition, there are so many companies around the world, and you have to track which version the IP you are using was developed for, which version of this third-party IP. If there is a new version of the specification, how do you ensure that everyone using this IP is using the correct IP?"

in conclusion

Uncertainties across multiple markets and technologies are intertwined, raising more questions about how 5G technology will be used in the future, when 5G will be commercially available, and how much money and other resources will ultimately be spent.

“All of these markets have additional requirements for range, cost, and they will still pose significant challenges to system design and manufacturing when implementing autonomous electrical control in the future,” said UMC’s Liu.

Among these technological transitions, 5G is one of the important challenges, but how difficult it is and when it can be commercialized remains to be seen.