AT&T avoided more vexing challenges this year as traffic shifted and spiked during the global pandemic due in large part to its ongoing effort to virtualize its network, according to one of the operator’s network leaders.

“Our journey to virtualization, which started five to six years ago, really helped us to have the tools and capabilities to spin up and add capacity really quickly,” Igal Elbaz, SVP of wireless technology at AT&T, said at the Oppenheimer 5G Summit earlier this week. 

While virtualization can help operators add capacity quickly, AT&T doesn’t use it much for that effort, he said. “We typically have the ability to understand the growth and we’re building accordingly, but in times like this this is where we have the opportunity to bring those investments that we’ve made in virtualization to work.”

AT&T recently hit its long-standing goal to virtualize 75% of its network functions with SDN either three months early or nine months late, based on a series of confusing and previously contradictory executive statements. 

AT&T’s 5G mobility core and many elements of routing and voice services are now virtualized, Elbaz said, calling it “an essential capability” during the COVID-19 crisis.

“The fact that we had the infrastructure and the knowledge of how to use virtualization helped us to bring capacity quickly to serve customers. Clearly there’s virtualization efforts in all parts of the architecture on the wireless network,” including the network core and more recently in the radio access network (RAN), he said. 

“There is virtualization on the wireline side of things and, because everything is moving to disaggregated software and hardware, you can start building common infrastructure that can run both functions and services on the same infrastructure,” Elbaz explained. 

In March, when the World Health Organization declared COVID-19 a global pandemic, there was some concern and doubt that AT&T wouldn’t be able to meet its 5G buildout goals, but it overcame challenges and met its targets, he said.

AT&T is closing out 2020 with a low-band 5G network that covers 230 million points of presence (PoPs), and its millimeter-wave (mmWave) 5G network is now live in parts of 37 U.S. cities. “We’re building millimeter-wave where we think it matters at this point in time, where you can actually build those experiences” in entertainment districts, high-density areas, airports, venues, and corporate campuses.

The operator’s claim to “nationwide” coverage on its low-band network, much like Verizon, relies heavily on dynamic spectrum sharing (DSS), software that allows operators to use the same spectrum for both 4G LTE and 5G.

Elbaz addressed the “noise” about the performance challenges of DSS, but argued it’s an important capability that operators need to deploy 5G at scale. “Like any other part of the network, it’s going to take some time to evolve and get to the level it needs, but it’s an absolutely important one,” he said. 

Edge computing is one of the most exciting parts of the 5G story, and it plays a significant role in why AT&T and other operators claim 5G is “not just a network evolution, it’s a real paradigm shift,” Elbaz said. “The spectrum of improvement is so wide in so many things.”

“For the first time, this is a network that is built in the cloud. It’s cloud native, which will allow us to be much more efficient in how we build and operate our network,” he said. Other technology advancements that help operators expose capabilities to network APIs or improve the positioning capabilities of the network also allows AT&T to operate and manage 5G as a “real-time network” and that’s where edge computing starts to play a role, he added. 

Elbaz defines the edge as a distributed cloud and architecture that allows AT&T to run network functions, applications, intelligence, machine learning, and data services, closer to its customers. “That convergence of network, compute, and storage is really unique and really what can unleash a lot of the immersive use cases,” he said.

AT&T is pursuing two flavors of edge: the AT&T Network Edge, where it partners with Microsoft Azure and Google Cloud to allow developers to build and support applications at the edge of its network; and multi-access edge compute, which is designed for private enterprise applications.

“Now, for the first time, because of that architecture I described, this distributed architecture, we can extend our wireless network into the enterprise premise. You don’t have to go and build a complete network, you can just extend some of the elements that allow customers to now have 5G services on prem,” Elbaz said.

He also argued this is where 5G most critically outperforms WiFi for enterprises. “If you need a controller for robotics in a manufacturing site, the WiFi network is not good enough. You need that wireless network there,” he said, adding that it also allows organizations to keep their data and applications stored locally for greater privacy and security. 

Manufacturing robotics that require latency under five milliseconds, sometimes under one millisecond, can’t be supported by WiFi, Elbaz argued. “You need the management and the licensed spectrum in order to be able to provide capabilities that allow for those use cases.”

Many enterprises will have both WiFi and 5G for different uses and they will coexist and be managed in a more converged way, he said. 

Finally, Elbaz described open RAN as a huge opportunity, adding that AT&T will deploy and implement the technology over time. “We have every intention not to just participate, but also to contribute and lead that effort,” he said.

Open RAN is important, he argued, because traditional wireless systems are closed, tightly integrated, and served by a very small group of vendors. “The fact that we can now work on open specifications and open interfaces, and interoperability between different parts of the wireless networks allows us to have improved [total cost of ownership]” and foster innovation, Elbaz said.

“I don’t think we’re very unique. We have more than one vendor in our network, and each one of those vendors comes with its own management system,” he said, adding that a common model deployed on open interfaces frees operators from relying on dedicated management systems.

“Our ability on a closed system to fine tune the network to better serve and customize the network to the benefits of our customers is limited. And the fact that we’re moving into an open network allows us to apply intelligence on top of the network,” Elbaz said, adding that he disagrees with the notion that open RAN is only for greenfield operators.

“I really believe that in AT&T we have the expertise and we have the knowledge to adopt open RAN,” he said. “We will deploy and implement open RAN. … It’s not going to happen overnight. It’s going to take some time, but we are absolutely interested and we’ll be participating in that effort.”