Jason Inskeep is director of the 5G Center of Excellence at AT&T Business. He was interviewed by Fierce Wireless’ Executive Editor Monica Alleven as part of the November 15, 2021, Fierce E5G show. Inskeep talked about getting the right mix of licensed, unlicensed and shared spectrum to customers, as well as the role of Mobile Edge Compute (MEC).
This transcript was lightly edited for brevity and clarity.
Fierce: It seems like there's been a lot of talk about private networking when it comes to LTE and CBRS. Can you bring us up to speed? What does AT&T offer by way of private wireless and LTE? And what did we do before CBRS?
Inskeep: Interesting topic. It's one of those things where you're often like, well, did 5G bring about CBRS in private and LTE or was it always here and it's always been here, right? You've seen companies like utilities, places like that, that have big, big areas to cover, obviously big capital accounts to go out and procure spectrum to create their own private cellular network. So it's not a new principle per se. CBRS as well as software defined network have opened up the aperture a little bit, so other players can participate in it. From our vantage point, spectrum being one of the key, if not the most important part of any cellular network, we sought to find ways to not only use the new spectrum of CBRS, but also take advantage of our licensed spectrum and different technologies and architectures to help our customers really create a suite of services that started with what we call MEC. At AT&T, MEC is basically localizing our licensed spectrum.
Secondarily, obviously, CBRS comes along. That's another place for us to go into, depending on the use case and site type. And then, as we kind of keep going forward, we're seeing more hybrid and kind of merged architectures that bring together both spectrum types as we begin to look forward into the future.
Fierce: I want to talk about spectrum for a little bit. One of the questions is the role of unlicensed spectrum and these private wireless networks. You have that to some extent with CBRS, with the PALs (Priority Access Licenses) and GAA (General Authorized Access). I'm interested in what you're doing with that. There’s always been an issue between Wi-Fi and cellular, and LTE comes into the mix more and more with CBRS. The 6 gigahertz is coming online. How do you juggle those things and how do you know what's best for an enterprise?
You mentioned unlicensed. Unlicensed would be your 2.4, 5 Gigahertz, 6 Gigahertz – the Wi-Fi spectrums that have been used. Shared is where CBRS sits, which has some characteristics of both unlicensed, which anybody can use at any time, as well as characteristics of licensed spectrum. And, you know, from that perspective there may be some power restrictions. You’ve got to think about maybe some availability things, you’ve got to think about across that entire paradigm, as you start looking at spectrum, and how you use it and how you operate it.
Then you kind of fall back into what does the core look like, right? Where does the core sit in conjunction with the spectrum options I have available? Those two pieces typically are the biggest, I don't want to say gaps, but learning areas that we have to work with our customers on because getting that right mix between licensed, unlicensed and shared will ultimately get customers what they need. I don't think it's going to be one or the other in terms of what happens as we go forward. I think as time progresses, given the way wireless networks are growing, you're going to need all of it, inclusive of Wi-Fi, inclusive of private, inclusive of public. Mixed together in terms of the different licensing and structure models.
Fierce: Another question has to do with your FirstNet network and that spectrum. Are you able to use that for private networking? Or are there restrictions to that? It seems like maybe you could use that.
Inskeep: The FirstNet network itself is kind of its own, what's called quasi-private network for first responders. It has a dedicated spectrum band inside of that with Band 14. There may be some opportunities as we continue to forward… think about hospital environments, things like that, where obviously the first responder comes into play, and kind of merging that opportunity with what we already do.
To your point, there's definitely some more considerations we have to think about in terms of how that packet is encrypted and what we do on top of normal encryption with a cellular network that we have to take into consideration. But that's just one. As we think about these hub and spoke type architectures in a way we call MEC, and how we could blend that together, for that different user type, so there’s certainly an opportunity there.
Fierce: Speaking of MEC, can you sort of explain how that works, like how it works in a private network scenario and how does that play with 5G? It’s mostly a 5G thing, isn't it?
Inskeep: No, it actually happened late LTE, right? So after the network became software enabled, 5G being the first, or late LTE actually being the first software enabled network, the control and user plane split. When you look at MEC in terms of the ETSI standard, you see kind of a combination of compute and network running across a similar hypervisor.
What we try to look at, because we're a networking company, is kind of getting back to some hypothesis. How do I give the best of Wi-Fi and the best of cellular on a premise, right? And one of the ways to do that is obviously I want to get them control of licensed spectrum and looking at how our version of MEC operates, which is really a big focus on the network layer.
It allows us to use the control plane that's intact, which is basically the identification authentication of the SIM card and let the customer have a little bit of control of that. So they can kind of peel off the data layers that they need. And some of those attributes that they would look for could be everything from the destination IP, where that traffic is supposed to end up down to different SIM, SIM and credentials and authentication. So now the customer has control essentially of the routing and they get the advantage of the license spectrum, right?
You're also seeing this move further into the ETSI standard, that collaboration and convergence of network and application. So now you have that networking layer that I just talked about moving the user plane or the cellular network down combined with and sharing attributes with the compute layer, right? So I'm merging with the Microsoft – good big partner of ours. And we start working on how the network informs the application ... which gets us to kind of that next generation of application experience that we're all kind of hoping and planning for as we move to 5G.
Fierce: Can you talk about some of the industry sectors that are particularly interested in not just private enterprise, but also the MEC construct as well?
Inskeep: Yeah, it's interesting as you think about it, our definition aside, when I look at how MEC, we look at MEC as another part of our private cellular suite and we kind of started our model from an inside out perspective, right? Which is, ‘Hey, if I can do it inside of a smaller or larger campus, I can do that in a real world from an architectural perspective.' Where you start running into your hurdles as outside of the technical fail fast, but more on the operational fail fast, i.e., inside a manufacturing facility, the manufacturer creates the data and they own the data. So they don't have to think about Joe Public and Jane Public and consumer traffic, meaning they can build a network that makes the most sense … build it the right way for the performance they need.
As you kind of move across that ownership model, let's say going from a manufacturing site to maybe a retail environment, which could include a warehouse – very industrial, but a store, even a big store, you're going to have a mixed population coming in and out, right? So they have to make a decision … what's gonna make the most sense for me? Is it Wi-Fi, all Wi-Fi? Is it a mix? That could be based on the site type and user type that keeps pushing you across to the real challenging areas. Maybe it's an airport, maybe it's a city where you've got all these different mixtures of data and businesses and enterprises and potentially device owners.
Technically it can look the same, but the mixing of that data creates a real challenge in terms of how do we all work together to get the optimal performance for the end user? So for us, we started heavy on that industrial side, and I think you've seen a lot of trend of that. We noticed the trend is really based on the data ownership. Whereas the data owner and who is the data creator, and from there, you can kind of work up the stack and again, from a technical architecture perspective, we think it's pretty flat, but then it's how do we share that information amongst the constituents in the ecosystem?
Fierce: OK. I'm curious too about something we used to talk about a lot more – distributed antenna systems, DAS. Are they still being used or does this change any of that, with the MEC? Is there any relation there?
Inskeep: It’s another way to get RF coverage out there, and you're going to still see that in different environments. You're going to see obviously more smart, small cells. You're seeing some interest or at least some evaluation of combined architectures or combined radios. And a lot of the things you start to see there is how do I put one radio out and do it all, but that plays into challenges, right?
When you think about licensed spectrum has the ability to broadcast at a higher power ratio than CBRS or Wi-Fi. If you build it for Wi-Fi, you're probably overbuilding it for cellular and vice versa. So that comes into consideration. You've got all these different types of things. DAS, I don't think's going anywhere, because some venues, you take a stadium as an example … they want to give as much capacity as possible to that area. Other facilities may be better off with small cells. It'll depend again on really the site type and the space that you're dealing with.
Fierce: I remember talking to some folks about the eyeball antennas, the big antennas for the big enterprise, concerts outdoors and stuff like that. Are those still used?
Inskeep: Yeah. There's different versions of that that will come up for extra capacity and extra space. And you'll see more of the temporary type pop-ups happening now. We did something, it’s been a couple years now, for aesthetics at a golf course. So instead of having to pull fiber to do things, you can have these little pop-up networks that attach to the bigger network. Some of them were called COWs [cells on wheels] but now they're getting even more advanced to the point where you're starting to think about a compute network all in one, one kind of little machine that can roll out for them. What you talk about is there, as well as new advances that kind of, you know, encapsulate this converged architecture.
Fierce: What is the role of cloud providers in all of this? We haven't talked about them too much yet.
Inskeep: Because of software defined networking, because of new spectrum options, they’re a new player in the cellular game, in terms of how we can work together with them. Obviously if we're thinking about convergence, they've got to move a little bit to the network side and we've got to move a little bit more to the application side, to ultimately get the experience we want to our customers.
So in a way, it's very important that we partner more because if we don't, I don't think we've reached those experiences, really that are being talked about with 5G going forward. Obviously, we've got a really tight relationship in terms of the things we're doing with cloud providers, in particular Microsoft. And so as that continues to evolve, you'll see more use cases, more scale and more opportunities come, not only from a premise perspective or private cellular perspective, but also how that bridges out into the macro world, when you think about smart campus, smart city types of environments.