Industry Updates

'SAMENA Daily' - News

Comcast taps edge compute and AI to detect and localize power outages

Comcast is pairing AI with its distributed access architecture to detect and localize power outages. 'This is a step-function in ultimate reliability,' says Chief Network Officer Elad Nafshi.

Symmetrical multigigabit speeds, low latency and enhanced security are among the pillars of the cable industry's so-called "10G" initiative.

Another key component of 10G that tends to get less billing is network and service reliability. And it's one that Comcast is starting to thrust into the spotlight as the operator rolls out a distributed access architecture (DAA), virtual cable modem termination system (vCMTS) and an emerging lineup of Full Duplex (FDX) amplifiers.

Also playing a big, future role will be a new family of DOCSIS 4.0 chips for nodes, cable modems and amplifiers embedded with artificial intelligence (AI) resulting from a collaboration of Broadcom and Comcast. Those new, unified chips will support both flavors of D4.0 – FDX and Extended Spectrum DOCSIS (ESD).

Comcast is starting to tap into that edge compute platform, along with the real-time telemetry data it generates, to help the company detect and pinpoint power outages and then quickly take appropriate actions to keep the network online when commercial power goes out.

Striking 'GOLD' at the edge

That initiative, internally referred to as Global Outage Localization & Detection (GOLD), is starting to be deployed and will be expanded as Comcast pushes ahead with DAA and other elements of its network upgrades, according to company EVP and Chief Network Officer Elad Nafshi.

"This is something in-field right now and we're working very hard to make this broadly available," Nafshi said of GOLD.

GOLD, he says, is enabled by an edge compute platform that's largely powered by Comcast's vCMTS and DAA platform and will be enhanced by the company's future deployment of FDX amplifiers.

"The vCMTS is effectively an edge compute cluster that's running the CMTSs and applications. And it comes with the real-time telemetry, real-time visibility and real-time detection that you can start to collect," Nafshi said.

As Comcast starts to introduce FDX amplifiers, the AI capabilities of that edge compute platform will be extended to within feet of customer homes, providing Comcast with a detailed and localized set of AI pattern detection and actionable data.

And that's where GOLD is starting to play a role. The telemetry data alerts Comcast when a digital node flips over to a battery backup because access to commercial power has been cut. In turn, the AI in the system can also correlate and determine whether some homes in the area are equipped with backup generators.

When all of that data is pulled together and analyzed, Comcast can make sense of the alarms it's receiving and quickly determine that it's facing a commercial power outage and take appropriate action. And rather than dispatching technicians to resolve a commercial power outage that they can't fix, those technicians can instead be deployed to keep close watch on battery drainage and connect them to power supplies to ensure the network stays online even during prolonged power outages.

Related:Comcast's Nafshi says DAA is critical step in DOCSIS 4.0

And that data can also be applied toward the ongoing training of the underlying AI system.

"That's really where edge compute enables you to tie all of this together like never before," Nafshi said. "This is a step-function in ultimate reliability. It enables us to optimize the customer experience to what I consider outside of our control."

Competitive advantages

It's a way for Comcast to leverage the investments it has made in edge compute, but it also has other benefits, Nafshi says.

"This is our advantage against some of our competitors by having this amount of actionable data being collected and being actioned on," he said. "For us, all of the opaqueness is gone. It's right there in front of us … It really redefines customer reliability and what customer reliability is going forward."

Nafshi believes the coming rollout of self-configurable and self-adjusting FDX amplifiers will be "game changers" for Comcast's network reliability efforts. "This is a whole reimagining of what cable networks can be, should be."

He acknowledges that FDX amplifiers won't make noise disappear, but they will be able to "mitigate around noise and then pinpoint a decision."

And that, he says, will fit with other capabilities that Comcast now has in-hand, including "XMF," a relatively new monitoring platform that enables the operator to quickly zero in on fiber cuts and other types of fiber impairments in the access network. XFM, the company has stated, can reduce that detection and location time to about 90 seconds compared to a prior typical time window of about two hours.

Rolled up, Comcast now has the ability to boost reliability across fiber, coax and – more recently – power, Nafshi said. "This is where the benefits of edge compute really come together."

Plugging into the power of the cable network

Others within the industry are taking advantage of cable network telemetry data to closely monitor the status of commercial power.

Gridmetrics, a subsidiary of CableLabs, is tapping into hundreds of thousands of node-based sensors to create a unique, neighborhood-level view of the status of the power grid.

Gridmetrics has turned that data into a set of products to help entities such as public safety organizations, banks, insurance companies and assisted living facilities receive alerts on power outages so they can take rapid action.



Source: https://www.lightreading.com/ai-machine-learning/comcast-taps-edge-compute-and-ai-to-detect-and-localize-power-outages

ATTENTION