- Google is aiming for Level 5 autonomy in its network. But operators aren't there yet
- Telcos struggle to get the right data in the right place to serve as the foundation for autonomous AI
- But autonomous operation is a priority, and operators are pushing hard
GOOGLE CLOUD NEXT, LAS VEGAS – Google let slip in an interview with Fierce Network that it’s pushing to roll out Level 5 autonomy on its fiber backbone network by the end of this year. However, telecom operators, which have strong incentives to have AI run their own sprawling networks, are not expected to do the same anytime soon.
Angelo Libertucci, Google Cloud’s Global Head of Industry for Telecom and former Juniper Networks exec, told us in an exclusive interview why not. He said, there's a good reason: telcos aren’t running the race from the same starting line. And a cloud company naturally has the advantage applying cloud-based AI to automate its network.
The problem for telcos, according to Libertucci, primarily centers around network data. More specifically, operators are still working out how to get the right network data, with the right enrichments, to the right place.
“We can’t just jump into AI. It actually starts with a foundation of data,” Libertucci said.
“We need to get the network data in a place where it’s actually usable,” he continued. “Then you could start talking about immediate analysis, real-time root cause analysis, autonomous remediation, recommendations so they can get a bit more proactive and then eventually allow an agent to make a change.”
Baby steps toward autonomy
Google is tackling this problem with wireless operators to start with, in part because 3GPP standards mean that all operators have at least some similarities and the data formats are more easily parsed by AI. There has been some visible work on this front. For example, Deutsche Telecom worked with Google Cloud to develop a RAN Guardian AI agent, which is capable of autonomous performance monitoring, issue classification, rerouting and network optimization.
Libertucci said this approach — tackling a subsection of the network and its related data rather than the whole kit and kaboodle — not only avoids overcomplicating the data work that needs to be done but also allows operators to prove the value of making such a move in the first place. Furthermore, it creates a template for work that can be replicated across the business.
Libertucci, by the way, wasn’t the only one to say that dealing with data is an issue for telcos.
Ramnish Singh, SVP of Engineering at SaaS provider Onix (which has worked with the likes of T-Mobile, Verizon and Telus), told Fierce that telcos are still facing an uphill battle getting the right data into the cloud in the first place. Why? Because there are so many legacy systems and “very little documentation available about what data is in what source.”
“That’s problem number one,” he said.
Sights set on autonomous networks
That said, telcos are apparently more determined than ever to push through the data doldrums so they can reap the benefits of AI and autonomy.
“Autonomous network operation is now the most engaged piece of work that we have with all of our customers,” Libertucci said. “This is a priority for our customers.”
If you had asked him a decade ago whether self-driving networks would see the light of day, Libertucci said he would have doubted it would happen in his lifetime. Now, though, it’s a different story.
“What I’ve seen happen just in the last year, I’m much more optimistic,” he said. “From my perspective, I think within five years you will start to see fully autonomous networks.”
Want more? Check out all our Google Cloud Next coverage here.