New EXFO platform leverages AI, ML and “little data” for service assurance
To hear EXFO CEO Philippe Morin tell it, the virtualization of service provider networks is degrading network visibility and the ability to troubleshoot rapidly, while data lakes are more like overwhelming data swamps— and 5G is only going to make things more challenging.
Morin said that as EXFO’s customers have moved toward network virtualization since 2016, they have seen a significant rise in outages: an increase of 46% in the past three years, with 65% of those originating in virtualized domains, according to EXFO. Network operator teams are constantly putting out fires: they spend 67% of their time fixing network and service issues, and it typically takes a dozen people from three different teams more than three hours to identify each outage’s root cause, EXFO said.
Despite the best efforts of operators to gain insight into impacts on customer experience, they are almost completely blind to subscriber-impacting events, by EXFO’s reckoning. The company says that carriers are 98.7% blind to such events. And the big data strategy of dumping enormous amounts of network information into a data lake and then trying to sift through for answers while balancing storage and computing needs, just isn’t cutting it when it comes to rapid problem detection and troubleshooting.
Current tools provide only “performance averages and aggregates, rather than per-subscriber problems with unique service experiences. It is only an approximate view of the health of their networks and how well they are performing,” according to EXFO. Instead, Morin said, a “little data” approach is called for, of lightweight, right-time and relevant information that can inform an automated solution which learns from that data.
So EXFO is leaning into artificial intelligence and machine learning to help network operators cope with the complexity of their newly virtualized, flexible networks. The company launched a new offering this week to meet those needs: Nova Adaptive Service Assurance, or Nova A|SA, which it says is the first intelligent automation platform enabling mobile network operators to deliver high-quality service experiences in a 5G world by integrating information from relevant network layers, domains, and data sources—both from EXFO’s own solution portfolio and from third parties.
“The combination of more users, more connections, more apps and more convoluted networks has created a perfect storm of complexity for operators,” said Morin. “By delivering only the right data at the right time, Nova A|SA is a unique intelligent automation platform to provide operators with 100% visibility into user experience and network performance. We’re talking about operations teams being able to resolve issues in minutes rather than days—or preventing them entirely.”
Morin added that while Nova can handle 5G, “it’s really not just targeted for 5G, because we think there are already challenges in terms of better visibility in today’s networks.”
But 5G does present some uniquely challenging demands for service assurance, both in meeting human customers’ high expectations and in handling the possibly even higher demands of machine-to-machine communications for massive IoT.
When a human subscriber deals with an outage or service degradation and experiences a call drop, they may shrug it off—or they may call their service provider about it, Morin said, which will prompt analysis of the issue. To a degree, humans will likely tolerate some service degradation and just try the call (or loading the app or website) again later.
“Losing a connection originating from an M2M connection, the machines won’t be as tolerant,” Morin said. “So there are all of these drivers for us to come down to saying that we need a different way to do service assurance, and we call it a new breed of service assurance that has AI-powered machine learning—so that when we introduce machine-to-machine, there is better visibility for the service providers through automation, and they can act.”
EXFO described its Nova SenaAI, which leverages machine learning, as the “central nervous system” for the platform, which the company said includes the capability to identify and even predict user-impacting events that were previously invisible. The company said that platform can “[pinpoint]who was impacted, where and for how long, as well as [diagnose]the root cause of issues for rapid resolution.”
While machine learning-based tools always have a time period in which they have to ingest data and begin learning the baseline behavior and normal quirks of a given implementation, Morin said that EXFO’s solution can be up and running with a matter of a few hours—not weeks or months.
EXFO developed the new platform in partnership with its carrier customers, it said, “to prepare for a 5G world in which connected devices, rather than cellphone subscriptions, are expected to drive 95% of new revenues for operators.” In addition, the company pointed out, most human customers base their opinion of a brand on network performance, and “there is enormous commercial opportunity for operators who successfully deliver on the promise of 5G and win customer loyalty.”
“If we want to be successful to get 5G deployed in a higher volume, I think it’s going to be really important that service assurance is provided in an automated way,” Morin said. “And that’s what our customers have been looking for, because they realize the challenges that they have with 4G networks that are starting to virtualize.”