Verizon gets ‘far out’ when it comes to the edge, virtualization

During an interview with Fierce last week, Adam Koeppe, SVP for Technology Strategy and Planning at Verizon, discussed Verizon’s path to virtualization.

The company has declared itself the leader in virtualized Radio Access Network (vRAN) innovation, deploying more than 8,000 virtualized cell sites with a goal of deploying over 20,000 by the end of 2025.

“I believe it’s the largest virtual infrastructure deployment in telecom,” when you factor in the core, edge and now the “far edge” of the network with cell site functions and RAN running on Verizon’s cloud platform, he said.

AT&T grabbed a lot of headlines over the years when it professed its goal of virtualizing 75% of its network, which it achieved in 2020.

Verizon hasn’t talked about a specific percentage, and Koeppe said it’s not something Verizon has bothered to quantify because “no matter what anybody says, you spend hours trying to translate it.”

The virtualization of core functions started five or six years ago, and the approach, from both the wireless and wireline network, was to look at directional network functions and platforms and move those to a virtual architecture, he said. It didn’t mean virtualizing everything because plenty of legacy stuff doesn’t require virtualization.  

Simply put, the vast majority of Verizon’s 4G LTE network traffic and all of its 5G network is served by virtual infrastructure, he said. 

With Verizon’s approach to operating its cloud platform for network functions, “our goal is to never relegate the customer-facing parts of the network to somebody else,” he said. “We want to ensure that our network functions are in our control, operated by our people,” and those network functions are on a cloud within Verizon’s control and network, he said.

Other operators have chosen to take parts of their network and move them to a public cloud, similar to what AT&T is doing in part with its partnership with Microsoft, he said. Dish Network also has been very public about running network functions on Amazon Web Services (AWS).

“But we are not doing that and have no plans to do that,” Koeppe said.

That applies to the core, to the edge and to the far edge of the network, he said, with the “far edge” referring to equipment closest to the customer, like small cells, macro cells and repeaters. The edge refers to things like fiber rings and switches, while the core is the backend where orchestration and operational control functions take place.

Side note: Last year, AT&T announced it was moving its 5G core to the Microsoft Azure cloud, which sparked a lot of speculation about AT&T ceding control. During a media tour of AT&T Labs in Redmond, Washington, earlier this month, Igal Elbaz, SVP of Technology and Network Services at AT&T, discussed at length AT&T’s relationship with Microsoft, saying AT&T “didn’t outsource anything.” 

In fact, “we still operate this. We still own this. The only thing that’s changed is we do not develop that cloud operating system layer any longer internally, and we are taking advantage of companies that have the scale and innovation in that area,” Elbaz said.

Catching Amazon’s waves

Back to Verizon, Koeppe said an important delineation is Verizon’s partnership with cloud service providers when it comes to public and private Mobile Edge Compute (MEC)-like functions.

With Amazon Web Services (AWS), its Wavelength compute platform is paired with Verizon’s wireless and wireline network to provide a low latency service to their development community.

The difference there is AWS’ Wavelength compute platform, which provides the low latency for those applications that require it, is running inside Verizon’s network. “So while it’s Amazon’s platform, it runs in our physical network,” and actually sits in Verizon facilities in what he described as a unique partnership.

Of course, Verizon partners with all three cloud providers, but AWS was the first to physically put infrastructure together with Verizon’s network. Since then, it’s been working closely with Microsoft to find a way to integrate the Azure solution. Similar work is underway with Google, but “a little bit further out,” he said.

Size and scope of C-band deployment

Verizon announced last week that it had successfully completed lab trials using 200 MHz of C-band spectrum. Verizon started its C-band deployment by rolling out 60 MHz of spectrum and then last month announced it’s beginning to deploy 5G over 100 MHz of the spectrum. 

Every market has at least 140 MHz of C-band dedicated to 5G (with the exception of Alaska and Hawaii, which were excluded from the auction) and where it won the full 200 MHz, that covers roughly 40 million POPs in rural and suburban parts of the country. That speaks to the opportunity for fixed wireless access (FWA) in that footprint, he said.

The C-band deployment is believed to be the fastest of any deployment of new spectrum at Verizon, he said. It’s also a complicated endeavor in the sense that satellite players need to move out before wireless carriers can move in. Plus, carriers were struck at the 11th hour or so with major aviation concerns from the FAA and commercial airlines.

Verizon started with 60 MHz and negotiated with satellite companies for early clearance to make 100 MHz available in 30 additional markets, which is being deployed now. The rest of the spectrum clears by the end of 2023, which is where it gets the full up to 200 MHz of C-band bandwidth.

Some airports remain where the carriers continue to work closely with the FAA, but by and large, they’ve worked through the FAA’s concerns, according to Koeppe. There are specific guidelines around how they deploy near airports but can offer service near airports in almost all cases. Ongoing work continues with the airline industry to replace older altimeters that don’t have the appropriate filtering capabilities.

What about 5G standalone?

The question du jour is when the 5G standalone (SA) part of the 5G picture is going to commercially launch. That’s where more advanced 5G comes into play.

The goal at Verizon is to marry SA core capabilities with SA-capable devices, Koeppe said. Considering the road maps of handset OEMs and chipset providers like Qualcomm, the timing for SA functions at scale is 2023, but Verizon is already doing testing and moved some friendly commercial traffic onto the SA core, he said.

“It’s important to match those SA capabilities with your SA device penetration,” he said, and they see that happening at scale next year.

It’s still early in the technical development in terms of voice services for 5G, aka Voice over New Radio (VoNR), and Verizon isn’t sharing a timeline for that right now, he said. Voice isn’t nearly as forgiving, so to speak, as data, so service providers need to make sure it’s fully baked before they launch it.

T-Mobile was the first to launch SA and it’s got VoNR live in parts of two markets – Portland, Oregon, and Salt Lake City, Utah, but it also acknowledged that it’s no slam dunk when it comes to voice services and a new generation of wireless technology.