Telecom vendors may still want 5G discussions to revolve around hardware — towers, cell sites, and base stations — but for operators like AT&T, the biggest differentiator between 4G and 5G is not the hardware, it’s the software.

And while those hardware elements do play a crucial role in 5G deployments, software is where operators can really streamline their costs and manage their network. “We believe software is an essential element to 5G and should get as much love as hardware when discussing 5G,” said Melissa Arnoldi, president of technology and operations at AT&T.

Arnoldi, who gave a keynote address at the Brooklyn 5G Summit 2018 today, said that the company has embraced the disaggregation of hardware and software because it is necessary to do so to get the most functionality out of 5G. “Our architecture is disaggregation. Not only disaggregation on how we manage our features but disaggregation down to the white box. Software gives us speed, reduces our costs, and gives us the ability to do network slicing,” she noted.

Arnoldi reiterated AT&T’s plans to install more than 60,000 open source, software-powered white boxes across its network over the next several years to support 5G. In March, AT&T announced that these white box routers are part of a “radical realignment” of its network architecture.

AT&T has said it will be the first to launch mobile 5G service based upon the 3GPP non-standalone 5G NR standard. It plans to launch 5G in 12 markets this year. So far, it has revealed three of those markets — Atlanta, Dallas and Waco, Texas.

And as part of those deployment plans, the carrier has also said it will enhance its current cell sites and build new sites using small cells. These sites will use white box routers powered by the company’s disaggregated network operating system (dNOS), which is now being hosted by the Linux Foundation and in March was renamed DANOS.

Arnoldi also provided some updates on the company’s 5G trials that have taken place across the country from Austin, Texas, to Kalamazoo, Michigan. In the Michigan trial the company focused primarily on small businesses and found that there was very little impact to the millimeter wave (mmWave) signal from elements like snow, rain, or even foliage. “We learned that mmWave signals can penetrate foliage and glass and walls better than anticipated,” she said, adding that the company could deliver up to 1 gigabit of speed at 900 feet from the antenna.

In the South Bend, Indiana, 5G trial AT&T tested a full 5G architecture with a mobile core and delivered data with very low latency. “Latency — that is a differentiator,” she said.
In the company’s Waco, Texas, 5G trial it used Flexware, its virtual network functions (VNF) platform to handle the switching and routing functions for the trial. In that trial, AT&T delivered speeds of up to 1.2 gigabits with a latency of less than 10 milliseconds. And it was able to deliver those speeds while simultaneously connecting hundreds of users