Tech news, like political news, often tries to cover all the bases, so it would be reasonable for me to be cynical about Light Reading’s recent story on 5G “standalone”. 5G’s “next big flop” sure sounds negative, but in truth LR has shown a healthy level of 5G skepticism in the past. The reason this particular tale is important is that 5G Standalone is where most of the technical features of 5G reside. 5G Standalone is 5G New Radio (NR) plus 5G Core, and if anything other than total capacity and per-user bandwidth matter, it’s the 5G Core that provides it.
The article starts by telling us the truth about 5G, which is that improvements in the user experience that were promised widely are hype. It’s very likely that most users with 5G devices won’t even get 5G in many of their areas, and when 5G is available it’s unlikely to offer any different experience. I have a 4G phone, my wife has a 5G model, and there is no difference in our experience, anywhere we’ve used our phones. Except for the fact that her wallpaper is different, I probably would have a hard time telling which phone I was using.
The hype around 5G speeds is likely due to two truths. First, phone users don’t care much about mobile technology, only their own experience. Other than being faster, what claims can any mobile or fixed broadband service offer? Second, mobile operators and phone vendors wanted to promote 5G quickly, which induced them to focus on the non-stand-alone (NSA) model, that builds 5G radio on top of 4G’s Evolved Packet Core for mobility management. NSA is cheaper, and 5G Core standards are just now evolving to a reasonably complete state.
I can’t agree with the article that NSA is a “4G crutch”. It’s a necessary evolutionary step; you don’t get from one wireless generation to another by fork-lifting everything in your infrastructure and having everyone throw away their old phones. Evolution, preparation, is a good thing. It’s just a shame that 5G proponents didn’t think a bit more about it, in the area of applications.
There’s an underlying truth about 5G Core: users do not now, nor are they likely to ever, care about it. 5G Core doesn’t include applications, which is what users do care about. What it purports to do is facilitate them by first slicing 5G service into independent pieces to separate applications with different needs, and then by promoting a software-hosted-feature model that would facilitate the development of those applications. I say “purports” here because it’s really those two points on which 5G has flopped.
The obvious problem with network slicing is that it either supports something like MVNO relationships (which first are already supported, and second don’t have much meaning to phone users, even those who use MVNO services), or it supports new applications that we can’t yet identify. Logically, you’d need to have the applications first before you started slicing networks to separate them.
Obviously, we don’t have them, and that’s more of a problem than it might seem. Why would we believe that there was this great groundswell of 5G application opportunity that for some reason could not be realized at all on 4G? Wouldn’t we have plenty of examples of those applications limping along on “primitive” 4G networks (that, by the way, seem to deliver exactly the same experience as 5G)? When there’s a technology constraint acting on a market, things cluster at the boundary point where that constraint starts to bite.
What we have instead is ironic. 5G proponents think that “applications” are going to pull 5G through. The proponents of those very applications think 5G is going to pull them through. In point of fact, there’s not much of relationship between most of the applications touted as 5G drivers and 5G itself. The article points out that self-drive cars and remote surgery as 5G applications are for those who have a business or personal death wish.
That’s probably true, but the real story is that all our 5G-application-think is based on that same illusion that drives much of 5G publicity and application hype today—the need for speed, or latency, or some simple connection-network property. The article makes that clear, but it also seems to think that “standalone” 5G is somehow the missing piece. It laments the lack of 5G Core support (standalone) in handsets, when I’m not sure that handsets really have to support the specific Core features. In any event, we’re still not thinking about a vast and totally new application opportunity.
5G, whether Standalone or NSA, is not going to create applications. Even our semi-realistic thoughts (like augmented reality or my own “contextual computing”) are postulating the creation of an ecosystem of technologies that we’ve not even really started developing. In fact, as I said in my recent blog series on edge computing, we really don’t even have a model for an edge-specific application architecture, and if there is any 5G application opportunity it’s hard to see how it wouldn’t be latency-sensitive and require edge computing as well as 5G.
That, to me, is the key point. If latency matters, then it’s not just 5G we need to look at, it’s hosting close to the edge, it’s creating highly agile and scalable application architectures, and it’s building applications where latency matters, which is where applications couple the world of IT and the real world closely together.
Every significant wave of enterprise IT spending since the dawn of computing has been launched by a technology shift that brought computer technology closer to the worker. We could argue the same for consumer technology spending. It could be that 5G’s lower latency, combined with edge computing’s lower latency, is the next logical step in bringing computers, experiences, closer to people. But what are those experiences? We either have to define and offer them in conjunction with 5G, or we at least have to create an ecosystem, an architecture, within which competitive innovation can create them.
It’s fair to ask here why all this hasn’t been obvious from the first, and why we’ve not moved mountains, market-wise, to resolve the 5G application issues. I offer three reasons.
First, the telcos, handset vendors, and mobile network infrastructure players all know that 5G is going to deploy, period. It’s an evolution of wireless networking that’s essential for continued good mobile services. Trying to sell users on 5G is probably more a media priority than a market necessity. We don’t need to justify the deployment, but we can’t write about it and generate clicks without making it interesting. What we see about markets depends on media priorities, not reality. “News” means “novelty” not “truth”.
Second, none of the stakeholders in traditional mobile networking have any experience with “applications”. The truth is that voice, text, and Internet access are the “applications” they’ve needed, and they still see the future in those terms. From that rather stodgy position, they see their responsibility to the market fulfilled when the tools are available to exploit. Exploiting them is a problem for others to solve.
Third, networks don’t create things people really value. We don’t want the stuff the network does, we want the stuff it delivers. When the Internet came along, it was a geeky invention of science and technology gurus. What changed the Internet wasn’t a change in its technology, it was the creation of the hypertext transfer protocol and markup language (created in 1990 by Berners-Lee). The Internet explosion was due to the “worldwide web” application architecture. Future latency and edge explosions will be based on some kind of application-layer shift, too.
The closing of the article says “’5G standalone will allow us to launch new services not even contemplated today,’ said Howard Watson, the chief technology officer of the UK’s BT, during a recent press conference. The time for contemplation is long overdue.” That’s true, but finding the right players to do the contemplating is perhaps the biggest challenge of all.