A Blog by Jonathan Low

 

Dec 18, 2019

5G Will Supercharge the Internet of Things. Not Everyone Thinks That's Great News

The system is not ready to handle the density and speeds - and experts are not sure how or when they will be.

Which adds potentially frustrating efficacy issues on top of the still (and perhaps never) unresolved privacy concerns. JL


Rob Pegoraro reports in ars technica:

5G promises the same sort of breakthrough in reliability and performance that desktop operating systems delivered when they moved from cooperative multitasking to preemptive multitasking (the operating system keeps them in their own lane). 5G's version of this is called "network slicing," in which a wireless carrier can divide its bandwidth into discrete channels, each with its own defined specifications for speed, capacity and latency. (But) the long-term for 5G IoT promises to support a density of devices far beyond what current-generation LTE can deliver—up to a million "things" per square kilometer, versus almost 61,000 under today's 4G.
It's true that inorganic users don't yell at customer-service reps or trash-talk companies on Twitter. But connected devices can also benefit from some less-obvious upgrades that 5G should deliver—and we, their organic overlords, could profit in the long run.
You may have heard about 5G's Internet-of-Things potential yourself in such gauzy statements as "5G will make every industry and every part of our lives better" (spoken by Meredith Attwell Baker, president of the wireless trade group CTIA, at the MWC Americas trade show in 2017) and "It's a wholly new technology ushering in a new era of transformation" (from Ronan Dunne, executive vice president and CEO of Verizon's consumer group, at 2019's Web Summit conference).But as with 5G in the smartphone and home-broadband contexts, the ripple effects alluded to in statements are potentially huge—and they will take years to land on our shores. Yes, you've heard this before: the news is big, but it's still early days.

Massively multiplayer mobile bandwidth

The long-term map for 5G IoT promises to support a density of devices far beyond what current-generation LTE can deliver—up to a million "things" per square kilometer, versus almost 61,000 under today's 4G. That density will open up possibilities that today would require a horrendous amount of wired connectivity.
For example, precision-controlled factories could take advantage of the space in the airwaves to implement extremely granular monitoring, and 5G IoT promises to do that job for less. "You can put tons of environmental sensors everywhere," said Recon Analytics founder Roger Entner. "You can put a tag on every piece of equipment."
An automated robot production line at SIASUN Robot & Automation Co., Ltd. High-density IoT deployments could put monitoring tags on everything in this picture.
Enlarge / An automated robot production line at SIASUN Robot & Automation Co., Ltd. High-density IoT deployments could put monitoring tags on everything in this picture.
China News Service / Getty Images
"Either I upgrade this to fiber to connect the machines, or I use millimeter-wave 5G in the factory," echoes Rüdiger Schicht, a senior partner with the Boston Consulting Group. "Everything we hear on reliability and manageability of that infrastructure indicates that 5G is superior."
Millimeter-wave 5G runs on bands of frequencies starting at 24GHz, far above the frequencies employed for LTE. The enormous amounts of free spectrum up there allow for gigabit speeds—at the cost of range, which would be limited to a thousand feet or so. That still exceeds Wi-Fi's reach, though.
Low-band 5G on the same frequencies today used for 4G doesn't allow for a massive speed boost but should at least cover far more ground, while mid-band 5G should offer a good mix of speed and coverage—at least, once carriers have more free spectrum on which to provide that coverage. (If you'd like a quick refresher on the various flavors of 5G, our story from a couple of weeks ago has you covered!)In the United States, fixing those spectrum issues hinges on the Federal Communications Commission's recently-announced plan to auction off 280MHz of so-called C-band spectrum, between 3.2GHz and 3.98GHz, on a sped-up timetable that could see those bands in service in two to three years.
And that means there's some time to figure things out. Companies aren't lighting up connected devices by the millions just yet.
The current 5G standard—formally speaking, 3GPP Release 15—does not include support for the enormous device density we're talking about. That will have to wait until Release 16, now in its final stages of approval, although Entner warns that we won't see compatible hardware for at least another year or two.

Slicing up networks

Another upcoming 5G feature promises the same sort of breakthrough in reliability and performance that desktop operating systems delivered when they moved from cooperative multitasking (programs have to be polite enough to stay out of each other's way) to preemptive multitasking (the operating system keeps them in their own lane).
5G's version of this is called "network slicing," in which a wireless carrier can divide its bandwidth into discrete channels, each with its own defined specifications for speed, capacity and latency.
Take the example of a 5G-connected car. "An operator could provide one network slice for lower-bandwidth but critical communications and another slice for consumer entertainment, each with their particular quality-of-service requirements," explained Rysavy Research President Peter Rysavy in an email exchange with Ars.
For wireless carriers, meanwhile, network slicing offers the promise of transcending their dumb-pipe status by offering a defined service they couldn't guarantee before.
"Suddenly you can actually have SLAs [service-level agreements] that are worth the paper they're written on," Entner said. "The carrier can actually dedicate resources to you, the client, and guarantee that you actually get that."
Schicht was not as sold on the potential for network slicing, commenting that "there's a lot of fantasy" about it. But he did suggest that it could work on a nationwide level for such wide-area tasks as ensuring public-safety communication.
A firefighter battling the Mendocino Complex fire on August 7, 2018, near Lodoga, California.
Enlarge / A firefighter battling the Mendocino Complex fire on August 7, 2018, near Lodoga, California.
If that sounds to you like the sort of paid prioritization deals that net-neutrality advocates decry, you may not be wrong.
"The steps between network slicing and paid prioritization are not very many," warned Stan Adams, open Internet counsel at the Center for Democracy & Technology, at a January briefing hosted by that Washington think tank.
All of this, however, lies even further in 5G's future than a massive increase in the scale of Internet-of-Things devices.
"Full network-slicing capability will occur in Release 17, so deployments will be in the 2023 to 2024 timeframe," Rysavy tells us. "It's a bit early now to determine what type of markets operators will be addressing with slicing."
That timing should also allow more time for carriers to amass the mid-band spectrum they need to avoid getting stuck between mmWave and low-band, with their forced compromises of range or speed.

Cloud computing gets edgy

The third big IoT possibility in 5G hinges on the technology's low latency, which promises to erase (or at least minimize to near-irrelevance) the difference between on-device processing and remote computation by eliminating most of the network hops between a device and a data center.
"We cannot afford to send the data all the way to the data center and back," explained Cristina Rodriguez, vice president of Intel's Data Center Group, at an October event in Washington, DC, hosted by Georgetown University's business school. "In other words, we need distributed data centers."
In this edge-computing model, computation would happen closer to individual devices: not so much cloud as fog. Much like network slicing, it offers wireless carriers a chance to make their connections less of a commodity product.
Both edge computing and network slicing also benefit from 5G's improved authentication and encryption. "There's a security by design," Schicht said. "You get this all embedded in how you build and design the network."
“The steps between network slicing and paid prioritization are not very many.”
Peter Brown, chief solutions architect at Wind River, said in an email sent by a publicist that 5G-served edge computing builds on the ongoing move in services from full-stack virtual machines to containerized applications. Wind River is an embedded-software firm that is developing a platform to push Kubernetes and Docker containers out to 5G-connected nodes.
Cloud vendors have been inking their deals with wireless carriers to ensure they remain part of this future, Rysavy noted. For example, Microsoft is working with AT&T to deploy its Azure stack on the edges of AT&T's network. Amazon's Wavelength, meanwhile, offers the ability to embed AWS processing and storage in the data outposts of such carriers as Verizon, Vodafone, and KDDI.
"These edge systems will support both 4G and 5G," he added. "5G, however, has an advantage, because it facilitates breakout of application data at the edge."
But here, too, Rysavy advised against expecting a quick revolution: "It's early days for edge computing, so I expect edge computing to evolve and mature throughout the 2020s."
jump to endpage 2 of 4
A Waymo self-driving car doing its thing in May 2019.
Enlarge / A Waymo self-driving car doing its thing in May 2019.
Picture Alliance / Getty Images

Self-driving cars not self-justifying 5G use cases

Having all of these capabilities hanging from the nearest 5G cell site does not mean that customers will or should want to take advantage of them. Consider one of the most frequently invoked scenarios for 5G: connecting the self-driving automobile.
"Latency becomes the difference between stopping in centimeters and meters," said Samsung senior vice president and tech-strategy head June Hee Lee at the MWC trade show in Barcelona this February. "This is why 5G is the key to autonomous driving."
But companies actually building autonomous vehicles recoil from the notion of having a car await braking or steering commands from the cloud.
"You still have to rely on onboard computation for anything safety-critical," said Waymo chief technology officer Dmitri Dolgov at an event at that Google offshoot's headquarters in May.
GM's Cruise Automation subsidiary is taking a similar approach, emphasizing self-contained operation with a minimal attack surface.
"If you don't need something, take it out," said Cruise Automation's principal autonomous vehicle security architect Chris Valasek in a talk at the 2018 Black Hat security conference in Las Vegas. "We will just design the system where there are no inbound connections—the car only makes outbound connections."
Qualcomm, the firm behind the chipsets in most US smartphones, has spent years building a case for connected cars powered by its own platform. But Qualcomm isn't suggesting that a self-driving car will need connectivity to steer or brake.
An ambulance could send a give-way signal to autonomous vehicles ahead.
"The car is basically a locally controlled implementation," senior vice president of product management Nakul Duggal said. "It is not going to the cloud to take permissions."
Instead, Qualcomm sees 5G's role as a backstop to the car's own sensors, transmitting information to and from other vehicles and nearby infrastructure—"V2X," short for "vehicle to everything," in industry jargon.
For example, Duggal suggested that an ambulance could send a give-way signal to autonomous vehicles ahead that might be out of the line of sight.
Wind River's Brown says that sensors along a highway could respond to a crash by sending slow-down commands to vehicles well in advance; during free-flowing traffic, they could supplement the data collected by sensors on vehicles to augment the collective intelligence of the highway.
But, again, 5G won't be in the driver's seat.
"I see 5G building blocks that help enable and make a self-driving car work better, but this is only one building block in a much bigger story," said Boston Consulting Group's Schicht.
jump to endpage 3 of 4

Smart factories, but maybe not smart cities

The idea of using 5G to breathe intelligence into an industrial setting or an entire city tops many what-if scenarios around this technology. Rysavy called smart cities and factory automation two of "the most promising and realistic IoT applications." But the more confined and defined space of a factory may make for a better home for 5G innovation than a city's traffic-filled streets.
"I'm a little bit of a skeptic with smart cities, because who's going to pay for this all?" Entner said. "Cities are notoriously short on money."
He added that the Federal Communications Commission's move to cap how much cities and counties can charge wireless carriers for new cell-site deployments will likely discourage smart-city experimentation.
Many smart-city functions, meanwhile, don't necessarily need 5G so much as just more bandwidth, period. At a Business Insider event in Washington in December, the question of whether smart cities need 5G drew noncommittal responses from panelists.
"I don't think it's the only thing," said Tony Encinias, Dell's chief strategy and innovation officer for state and local government (and before that, CIO of the state of Pennsylvania).
A smart lamppost stands in the Kai Tak area of Hong Kong, China. Hong Kong will install about 400 multi-functional smart lampposts in a pilot scheme to enhance the collection of real-time city data and support 5G mobile network implementation, according to the government's website.
Enlarge / A smart lamppost stands in the Kai Tak area of Hong Kong, China. Hong Kong will install about 400 multi-functional smart lampposts in a pilot scheme to enhance the collection of real-time city data and support 5G mobile network implementation, according to the government's website.
Paul Yeung / Bloomberg via Getty Images
"They need connectivity," said Kevin Garlan, North American innovation head for Citi, adding that he wouldn't get hung up over 4G versus 5G.
Factories, however, have greater needs in a smaller space. Ericsson touts how wireless IoT has enabled it to improve the efficiencies of factories in Estonia, and China by precisely tracking tools and components—although the white paper Ericsson released in February notes that this "Industry 4.0" project includes 4G as well as 5G technology.
Still, the result is that Ericsson says it will need fewer humans at work in these facilities.
Industrial settings can bring other constraints. Schicht warns that many industrial firms are locked into relationships with entrenched equipment providers, making it "non-trivial" for competitors to pitch new control products to factory operators. He suggests that track-and-trace functions—using 5G to monitor the whereabouts of important components or tools, as in Ericsson's implementation—would represent the most common use case.
Wireless carriers themselves might find that 5G's network slicing and edge-computing capabilities can help them run their own businesses better. That would represent a sequel of sorts to "everything as a service" tools that bring cloud smarts to managing cloud services.
Still, by looking three to four years down a road that's already taken many unexpected turns, we necessarily risk losing sight of what might await us just over the next hill. Or, more plainly, history suggests that throwing bandwidth at a problem can lead to unexpected solutions.
Qualcomm's Duggal makes that point: "Whenever you have a big technology transition that happens, it becomes difficult for people to imagine what the world might look like if that technology existed."

0 comments:

Post a Comment