LIVE FROM MWL UNWRAPPED: THE 5G EVOLUTION: Mauro Goncalves Filho, SVP for AI RAN with SoftBank Telecom America Corp, highlighted a real-world use case for AI RAN and provided a timeline for when the technology would reach commercialisation.

Speaking to Mobile World Live on the first day of Unwrapped, Goncalves Filho touted progress the AI-RAN Alliance has made since its inception two years ago, which includes a membership of more than 100 operators and vendors.

“The focus has been on creating blueprints so that all the elements of AI RAN can work together,” he said. “The next big step is to move towards commercialisation and that’s where we’re going.”

He added SoftBank and its partners anticipate a phased approach with “some implementations between now and 6G”.

“It’s probably another couple of years,” Goncalves Filho stated. “It goes with the maturation cycle of the capex that’s already in there”.

“When it comes time to refresh the equipment, I believe we will see a lot of it being substituted with AI-RAN equipment.”

AI RAN in action
The operator already has several AI RAN use cases in the fold, including the deployment of multi-user (MU) MIMO in a live network.

“Within a lab environment that’s clean and controlled, it works very well. We wanted to see it in real life,” he explained. “We went outdoors in a very busy radio environment in Silicon Valley at the Nvidia campus and we tested it outside. It worked pretty well.”

In the deployment, SoftBank confirmed stable operation of MU-MIMO using its AI-RAN product in an outdoor setting.

“It’s working on 16 layers, which, as far as we can tell, is a world first,” he said. “What’s fun about it, it’s all the processing that’s happening on the DU, normal radio interfaces and so on. So, it’s a fully virtualised network”.

“The throughput is three times what it is when we turn off the capability. That means better service for the users and better utilisation of spectrum.”

He said the 16-layer MU-MIMO is particularly useful across high-density environments in places like Tokyo, central London and the Manhattan borough of New York City where dynamic adaptation is critical for maintaining service quality.

“The network is changing all the time,” Goncalves Filho noted. “Those reconfigurations need to happen very quickly. We can only do this because now we can do all these calculations very fast, using parallel compute, very close to where the user is”.

“[It’s] a classic case of AI RAN being deployed to benefit the network”.