5G reinvented: The longer, rougher road toward ubiquity
Last Updated on by Segun Ayo
The analogy that product managers liked to invoke in the 1980s and 1990s, whenever their companies were upgrading a popular line of software, was that it was like replacing the engine on an airplane in-flight.
Assuming everything ends up working out, thetransition should be the equivalent of a part-for-part reconstitution, reassembly, and refueling of an entire fleet while it’s up in the air and performing maneuvers. But this fleet happens to have flown into a massive thunderstorm in the midst of an earthquake. The pandemic has sent shock waves through manufacturers’ typical supply chains, and the first of what could be several extended lockdown periods resulted in a measurable demand shock economic event.
Meanwhile, the United States Government has officially thrown down the gauntlet, forcing manufacturers here to exclude China’s Huawei from their 5G supply chains. Extending a November 2019 order prohibiting US companies from using outlays from the country’s Universal Service Fund to purchase equipment from companies whose host countries pose a threat to national security, on June 30, the Federal Communications Commission designated Huawei as one of those companies. As the FCC order reads:
The Commission also initially designated two companies, including Huawei, as covered companies for the purposes of this rule, and directed the [Public Safety and Homeland Security] Bureau to determine whether to issue final designations of those companies. Based on the totality of evidence before us, the Bureau hereby issues this final designation of Huawei, as well as its parents, affiliates, and subsidiaries, as a covered company for purposes of this rule.
The “other” country alluded to in that paragraph is device maker ZTE.
The Universal Service Fund is the modern culmination of a long-standing agreement between telcos and the government, assuring them of federal assistance in providing communications service to rural and outlying areas. The new order came just two weeks after the Commerce Dept. issued a special exception for US companies working with Huawei expressly for the purposes of contributing to the global 5G standard. It’s reasonable for one to ask, what would be the incentive for any company, anywhere in the world, to participate in architecting a market where it’s prohibited from participating as an equal player?
Wednesday’s US move prompted the UK to consider following suit, potentially reversing its earlier decision to give Huawei a limited role in building out the perimeter of that country’s 5G network, though not its core. In a statement to the press Tuesday, UK Digital Secretary Oliver Dowden was quoted as saying, “Given that those sanctions are targeted at 5G… it is likely to have an impact on the viability of Huawei as a provider for the 5G network.”
Huawei is not the company responsible for originating the 5G small cell tower initiative; China Mobile is. Still, it could be China that retaliates on Huawei’s behalf. That country’s trade war strategy in recent months has been to strangle the US’ allies, particularly Australia, first by suspending beef imports and then raising tariffs on barley exports. (China may be relying on a dated playbook where US administrators care about the well-being of US allies.)
It’s the small companies that could be most impacted by the pinch of supply chain constraints — firms like Boulder, Colorado-based Comptek Technologies. Last February, the City of Los Angeles awarded Comptek the contract for its integrated smart street poles, like the model shown here.
Financial analysts are revising their long-term projections for 5G-related revenue downward, on account of both the pandemic and the trade war. Particularly affected, says analysis firm Research and Markets, will be the progress of small cell deployment.
The process for getting cities and townships nationwide to grant permits for small cell deployment continues. In Fishers, Indiana, for instance, work began last April to grant Verizon permits to deploy small cells, mostly on lampposts, throughout commercial districts and neighborhoods. Regulators nationwide have been working to streamline and fast-track permit approvals for fixtures that are smaller than a certain height, usually 50 feet.
But circumstances beyond the control of any municipality have conspired to slow this process down anyway. Meanwhile, device manufacturers including Apple are weighing the prospects of postponing rollouts of 5G devices until next year, when supply chains may perhaps be stabilized, and consumer demand may have normalized.
If 5G is ever to live up to its promise of restoring sustainability to the world’s communications networks, it needs to either slash costs in one department, or build an entirely new department — a revenue stream, preferably as part of an already established, competitive market.
5G, plan B
There are two 5Gs, and that is by design. The architecture that purges the network of all radio and communications components and methods from the past, while maintaining compatibility with older devices (user equipment, or UE) is called 5G Stand-Alone (5G SA). Release 16 of the 3GPP engineers’ architecture for global wireless communications, is being formally ratified and finalized on July 3. It was delayed on account of the pandemic, but only by a handful of months. 3GPP R16 is the second round of 5G technologies, in a series that has at least one more round devoted to 5G, most likely two.
The other 5G architecture is the one in use today in the United States: 5G Non-Stand Alone (5G NSA). It relies on the underlying foundation and existing base station structure of 4G LTE. By building 5G services and service levels literally into crowns that reside above or below the 4G buildouts (a “crown castle,” which also happens to be the name of one of North America’s largest owners of telco tower real estate), 4G has been giving 5G a leg up. Once it’s found its footing, the idea is that 4G can begin winding down.
But the 5G SA phase of the R16 specification has taken its sweet time to mature, admits Aeneas Dodd-Noble, principal 5G architect with Cisco.
“Here we are in 2020, but we actually have the specification from June 2019 as the one we want to rely upon,” Dodd-Noble told attendees of the recent Cisco Live virtual conference. “So it’s only been available for just over a year, and typically, when that happens, the technology itself actually takes a while to catch up with the standards. This is a time when operators are thinking about how to actually deploy the 5G Stand-Alone Core, [and] which use cases to pursue.”
R16 puts forth a variety of new options for revenue channels. In the absence of a network of small cells reaping benefits for telcos at an operational expense level, there may be ways for operators — and even other classes of technology vendor interested in becoming operators, even at a small scale — to carve out new markets for connectivity. For anyone who recalls the Oklahoma Land Run of 1889, there’s still ample time and promise for “boomers” seeking to stake their claim to untamed territory.
Yet in the midst of the human devastation caused by the coronavirus, the resistance by governments to their peoples’ calls for social change and racial justice, the tectonic separation of the US and China from the same planet as one another, and the awesome specter of climate change epitomized by the melting Arctic permafrost, the blazing Siberian forests, and chunks of the Sahara Desert taking to the sky and landing on Mississippi, what remains just happens to be a ticking time bomb. The technological advancements that 5G NSA makes to 4G LTE towers omit the single element that made 5G urgent in the first place: Its power generator.
It was a phenomenon noticed by power specialists just a few months after R15 was ratified. 4G towers are massively inefficient. As one of the world’s premier experts on power transformer technology, Schneider Electric’s Steven Carlini, explained in a company blog post last November, 4G towers draw 50 percent more electricity than they use. To borrow a phrase from data center facilities managers, that’s a Power Utilization Efficiency (PUE) ratio of 1.5. By comparison, Google claims the PUE for its cloud data centers averages about 1.11.
By itself, according to Carlini’s calculations, a 4G tower draws about 6 kilowatts (kW) of power during regular operation and up to 10 kW at peak capacity. Adding 5G NSA to that tower raises that draw to 10 kW at normal periods, and 13.7 kW at peak.
If all of the US’ estimated 106,527 telco towers, as of 2019, were equally upgraded with 5G NSA, using electricity industry-standard figures, just those towers alone would be producing an extra 82.7 metric tons of carbon dioxide emissions for every hour of standard operation.
Move to 5G SA, however — which would wean the telco industry off of 4G LTE completely — and not only would each old tower’s electricity consumption plunge to 4 kW on average and 6 kW at peak. By moving the control components to the cloud, the PUE for those components that were formerly housed in the tower would plummet to about 1.1.
This is the jackpot for telcos, the big payoff for all those years of investment, and for writing off the “long-term” part of 4G’s “Long-Term Evolution.” What’s more, the deployment of air-cooled small cells to blanket the continent, supplementing the existing large-tower network, would reduce their overall energy consumption even further.
The user plane sets sail
The key 5G portfolio technology for bringing this best-case scenario about, as Carlini correctly pointed out, is Multi-access Edge Computing (MEC). In a geographically distributed network, the “edge” is the furthest point closest to the telco’s customer: the nearest attainable gateway to a data center, of whatever size, that can reach the customer with computing services at minimum latency.
“The beauty of the 5G network,” explained HPE CTO for Communications and Media Solutions Jeff Edlund, “is that it is designed to put the compute, the storage, and the processing as close to the service delivery point as possible. For instance, if you want to run a very low-latency application such as autonomous driving, you need to push that processing all the way out to the edge, so that you get very quick response times.”
Historically, the first edge ever to have culminated in the drive for edge computing was the content delivery network (CDN). Such a service acts as a cache for high-bandwidth content, stationed a fewer number of “hops” away from its end-user, than the server from which it originates. 5G and edge engineers, including at HPE and Cisco, have taken to calling the CDN’s zone of operations the near edge.
The far edge is what Edlund’s referring to: The moving target when the end-user is highly mobile or highly remote. At such a variable locale, it takes more than an IP address to be able to locate the user. One has to leverage the mobile device network, at the very least, to pinpoint the user’s location.
But since every user is essentially mobile anyway, or at least is capable of becoming mobile at any time, as an HPE engineer proposed during his company’s recent Discover 2020 virtual conference, perhaps every application should be treated as a payload in the 5G network.
“Applications require more and more low latency, they require special security or privacy solutions, or they require bandwidth optimization,” explained Rolf Eberhardt, head of orchestration solutions for HPE. “All these kinds of features are not possible if you’re running on a centralized data center, hundreds of miles away from your central location.”
Ostensibly, Eberhardt is the advocate for a management platform that brings together applications and services that are highly distributed across multiple computing assets. These assets would include near-edge facilities such as micro data centers (µDC) stationed inside branch offices, business locations, and corporate campuses. (Long-time readers will note that LAN servers made from PCs used to perform this function in the 1980s and 1990s, but as my grandmother used to say, what goes around.)
But they would also, intriguingly, incorporate far-edge facilities that are a little distance away from the user, but not too far. These may be mini data centers (MDC) stationed in refurbished packing containers on company premises, or smaller, Tier-1 data centers. (It’s confusing, admittedly: The largest class of data center, in data center parlance, is Tier-4, where Tier-1 is the smallest. The class of Internet Service Provider with access to the backbone and the largest reach, in ISP lingo, is Tier-1, with Tier-3 being the smallest.)
In his presentation, Dodd-Noble cited a recent Cisco survey of some 1,350 enterprise professionals worldwide. Given a list of use cases and asked to rate their relative levels of appeal and interest, a full 98 percent of respondents said they would find a private cellular network, either on 4G LTE or 5G, “appealing.”
The ability to determine which class of facility is most suitable, and then to push to that point-of-presence (PoP) not only the workload but also the policies with which that workload is then managed remotely, are the key functions of MEC, and perhaps the crown jewel of 5G Release 16.
With MEC, said Cisco’s Dodd-Noble, “we can create a private network using publicly available spectrum.” As you may recall from our last installment, 5G NR-U is the standard’s provision for utilizing unlicensed spectrum — radio frequencies that are deemed public property by the government. Wi-Fi has always done this, but on a much smaller scale; NR-U could be deployed on a huge, 1.2 GHz slice of mid-band spectrum in the US, once the Federal Communications Commission’s plans are finally launched. As he continued:
We can both bring in the service provider as well as the enterprise network, and combine them. And this way we can actually have MEC local presence in the premises of the enterprise. We can therefore have wireless as well as 5G enabled in warehouses as well as automated systems that today may use Ethernet cables themselves — cut those cables, and provide those as a wireless service. As we have gotten used to cutting the cord on our entertainment experiences, here we’re cutting the cord in the factory, not so that we can actually have a variety of connections, but actually to make the changes to the connections, the rewiring that happens when you move robots around, much easier because we have no wires.
MEC, Dodd-Noble explained, brings three features to enterprise networking:
- Latency reduction, bringing compute power closer to where it’s generally consumed. Public clouds have availability zones like “US-East” and “US-West.” A MEC network can have a geography, like “Sacramento,” “Corpus Christi,” or “Blaine County, Idaho.”
- Edge offload, which reduces the costs incurred in keeping databases current, by keeping data close to where it originated. If you’re a California state agency, a Gulf of Mexico climatologist, or a logger, you may now be able to run simulations using data that’s already very close to you, without having to transfer it all into cloud-based buckets first.
- Bandwidth reduction, reducing or even eliminating much of the communications exchanges between local facilities and cloud data centers — exchanges which often do involve telco fiber, but end up being more of a burden for telcos than they’re worth.
MEC treats each telco network, with all its facilities distributed throughout the globe, like its own hybrid cloud. It builds a private customer network around the workload that network will host. And it then pushes all the virtualized components of that network toward the customer, at the near edge or the far edge.
This is possible by leveraging the key principle of software-defined networking (SDN) as a catapult for remote computing. First and foremost, it separates the control plane (the part of traffic devoted to maintenance and logistics) from the user plane (the part that runs applications.) MEC hurls the entire user plane towards the network edge, leaving the control plane local, inside the telco’s local facilities. So the geographical location of the network slice that the user sees is theoretically anywhere within 5G’s reach.
The concept is called Control/User Plane Separation (CUPS).
“Once you start looking at CUPS architecture,” explained Ravi Guntupalli, Cisco’s director of technology for next-generation mobility and 5G, “the user plane itself might be deployed in a completely different domain. And that is new for service providers. Traditionally, it’s been in a controlled environment. They need to start worrying about, where does my user plane go tomorrow? It might be in an enterprise location, or in a public cloud.”
The closest analogue to this type of functionality in a living organism, if there were such a thing, would be a being whose central nervous system could be separated from its brain by thousands of miles of fiber optic cable. SDN has always separated the control plane from the user plane; it’s not only how the system functions but why. The fabric linking the two could be any length, theoretically. But wireless communications have always been capable of tracking a mobile user when it’s relocated and re-routing the connection to that user seamlessly. 5G CUPS extends that capability to an application that can be located, or occasionally relocated, on a processor that is mapped somewhere in the telco provider’s network.
Critical to the capability of 5G networking equipment to pull this off will be a new generation of processors and internal components. With the world’s supply chains yanking themselves, or being yanked, off of China, this is where Qualcomm is working to fill the gap.
“Being able to deploy 5G in private networks — not just public operator networks, but private networks for industrial IoT, for example, is really important,” stated Danny Tseng, Qualcomm’s director of technical marketing, in a recent press conference. “You can make sure you have control over your resources, like spectrum. You can control your own deployment. You’re able to maintain the important, sensitive data on your own network, and not have to send to the operator. Release 16 enables that.”
Qualcomm is seeking to add a fourth major addition to enterprise networking [at the far right of the chart above]: Time-sensitive networking (TSN). “It’s allowing a network to deliver packets of data in a very deterministic time manner,” said Tseng. For example, a factory floor robot may be expecting a data packet from a central controller at a given interval, such as 5 ms. That’s usually only feasible if packets are adequately cached in advance. But if a robot is working in a non-repeatable pattern — not manufacturing something, let’s say, but taking something apart — that cached data may not have been collected. In a real-time operating scenario, all the processing power and all the data must not only be nearby but a fixed interval of time away.
From this perspective, you begin to get a picture of Cisco, Qualcomm, and other 5G stakeholders such as HPE and Intel, carving as many niches for prospective edge, or edge cloud, customers as they can. That’s their strategy; here are their present tactics:
- Peeling the user plane away from the public cloud, leveraging using telco fiber to relocate it someplace the telco customer would find measurably more convenient;
- Taking the left-behind control plane from the cloud and relocating it to the 5G Core. This way the telco, having effectively replaced the cloud service provider, can manage remote data centers from its central offices, while just the data plane can reside on the edge.
It’s a value proposition that CSPs can’t make just yet. While agreements such as the one edge data center provider Vapor IO has made with colocation provider Digital Realty do make progress in pushing data centers towards the edge, it can only happen right now on a city-by-city basis, or even site-by-site. The reason is that such deals require fiber connectivity to link the core to the edge. And telcos already have it all.
When 5G began, its “hail-Mary” pass was supposed to be a gamble on small cells reducing costs and paying off. But with time running out on the clock, telcos now have to make a few initial shuffle plays, and perhaps a quarterback sneak, to put themselves in a better position to score on that pass. The gamble now is that 5G can break off a chunk of the public cloud — one of the few truly successful new industries created in North America thus far this century. For this to work, service providers such as Verizon and AT&T in the US, and Vodafone and Telefonica in Europe, must make themselves the equals of AWS, Azure, and Google Cloud in the global computing economy.
Put another way, for 5G to change the world as planned, the world has to be changed first. Not the easiest of propositions.