As usage spiked during the early parts of the pandemic, Comcast has accelerated the pace of capacity upgrades and sped up the use of an AI system called 'Octave' to boost the efficiency of its access networks.

Jeff Baumgartner, Senior Editor

June 29, 2020

5 Min Read
Comcast beefs up network capacity, uses AI to keep data demands in check

To help it stay ahead of data demands driven by the early days of the pandemic, Comcast has accelerated capacity upgrades on its core and regional networks and has fast-tracked a relatively new AI-powered system called "Octave" to boost the performance of its access network.

As network usage spiked in March and April as millions stayed at home, the operator more than tripled the number of network enhancements compared to the same timeframe a year earlier, Comcast EVP and Chief Network Officer Jan Hofmeyr explained in this blog post.

Of note, Comcast added 1,700 new 100-gigabit links during this time, primarily between the backbone and its regional networks. Comcast has also taken advantage of the almost 700,000 diagnostic speed tests it conducts daily to determine which areas were seeing the greatest demand due to COVID-19. Per Comcast's latest stats, the operator has seen a 32% increase in upstream traffic and a 11% jump in downstream traffic during the pandemic.

"We basically added in three months the capacity we were planning to add for the entire year ... We believe we have the capacity we need for the rest of this year," Hofmeyr said in an interview.

Despite the accelerated addition of capacity in recent months, he also expects Comcast to return to a regular cadence under which it typically doubles the capacity of its network every 2.5 years. "We've always designed the network to be ahead of the traffic trends," he said. "More importantly, we design it to easily expand if we need to."

Hofmeyr said Comcast, which has deployed more than 33,000 route miles of new fiber in its core network since 2017, has been able to deploy that added capacity in the near term without disrupting service. Although Comcast needed people on the ground at the facilities to install new line cards and other physical equipment, it steered traffic around those areas during the upgrade process. Comcast made those capacity updates in an accelerated, compressed timeline, but was able to replicate that recipe throughout its network, Hofmeyr said.

With traffic now stabilizing, plateauing and even coming down in some instances, Hofmeyr said his team's attention is turning toward automation and the enhanced use of AI technologies in Comcast's core networks.

Adding AI smarts to the access network
Meanwhile, Comcast has been ramping up its use of AI technologies in its access networks. That's centered on the rollout of a relatively new platform, called Octave, that enables the MSO to rapidly pinpoint network anomalies and noise-activated issues by polling and analyzing anonymized data from gateways and modems. In addition to utilizing that data to locate and fix network impairments, Comcast is using it to boost the efficiency and performance of the access network.

With respect to data performance on the access network, Octave enables Comcast to dynamically set the maximum modulation profiles of individual cable modems based on current network conditions. That's a big change from having to drop all of the modems to a lowest common denominator modulation profile if just one modem in a given area is not capable of supporting a higher modulation due to noise or other plant conditions.

The feedback from Octave, for example, helps to ensure that modems or gateways that are capable of humming along at 4096-QAM can utilize that higher and more data-efficient modulation profile even if another modem in the group has to dip down to, say, 256-QAM because of LTE interference or some other type of network impairment that can cause noise to bleed back into the HFC network.

To enable this in real time with Octave, Comcast feeds anonymous polling data from tens of millions of devices on its access network to an AI decision engine. This engine analyzes the information and then sends commands to the cable modem termination system to dynamically set the highest supported modulation profiles of individual modems on the network.

"We want to be able to optimize the delivery, based on the actual edge reality," Elad Nafshi, SVP of next generation access networks at Comcast, explained. "Now I can open up a whole bunch more capacity that is optimized by the time of day and optimized to the customer that I can act on in real time and ... fluctuate that delivery to ensure [customers] are getting the maximum efficiency down to the house."

Tony Werner, Comcast's president of technology, product and Xperience, told Fast Company that a version of Octave deployed prior to the pandemic increased download capacity across its network by 36%. When the pandemic hit, Comcast fast-tracked a version of Octave centered on the upstream. That resulted in a 20% increase in upstream capacity, the report said.

Octave is "not just automating what smart engineers can do. It's going to places where they just couldn't process that amount of information and come up with solutions quick enough to do what [Octave] does," Werner told the publication.

Nafshi said Octave has helped Comcast manage and optimize the four carriers it's running in the spectrum allocated to its DOCSIS 3.0-based upstream. (Comcast has not yet launched OFDMA-based DOCSIS 3.1 upstream channels.) Additionally, the AI-powered system has put Comcast in position to eventually add a fifth and possibly a sixth channel in the lower part of the 5MHz-42MHz spectrum band that's been set aside for the upstream.

"Not only are we making better use of the four upstreams we had before, we're able to add additional upstream channels and, therefore, add additional upstream capacity," Nafshi said.

Related posts:

— Jeff Baumgartner, Senior Editor, Light Reading

Read more about:

AI

About the Author(s)

Jeff Baumgartner

Senior Editor, Light Reading

Jeff Baumgartner is a Senior Editor for Light Reading and is responsible for the day-to-day news coverage and analysis of the cable and video sectors. Follow him on X and LinkedIn.

Baumgartner also served as Site Editor for Light Reading Cable from 2007-2013. In between his two stints at Light Reading, he led tech coverage for Multichannel News and was a regular contributor to Broadcasting + Cable. Baumgartner was named to the 2018 class of the Cable TV Pioneers.

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like