AI/ML in Broadband Networks: the Role of Standards – EnterpriseAI

Earlier this year a new initiative to create standards for artificial intelligence (AI) and machine learning (ML) in the cable telecommunications industry was launched. The working group, which draws members from both inside and outside of cable including giants like IBM, is exploring how AI and ML can be leveraged to make the network more efficient. Success of this group will have significant implications for businesses across the country by moving the industry toward 10G (the broadband technology platform of the future with residential speeds up to 10 times faster than todays networks) with greater speed and by supporting scalability of new technology deployments across the network.

The new initiative is part of the SCTEISBE Standards program, the only ANSI-accredited platform for developing technical standards supporting cable broadband networks. Standards for these networks impact the more than 66 million people across the U.S. who rely on broadband access. Currently at least 20 expert members are working together to drive telecom standards and operational practices for AI and ML. The resulting standards will improve network efficiency, move the industry towards faster adoption of 10G, and allow products to be interchangeable and interoperable, thus accelerating the deployment of products and technologies in an ever-changing broadband landscape.

Still in its early stages, the AI/ML working group is analyzing current and projected projects utilizing AI and ML among member companies to determine what standards are needed. As cable network operators increasingly embrace the advantages of using AL and ML to run their networks, the initial focus of the group is on internal uses of the technology to improve the network. Primarily the AI and ML computing algorithms are complementing human efforts to optimize network operations and correct network impairments before customers even notice an issue. Three initial applied examples of AI and ML have risen to the surface in the first few months of work.

One application being explored for the creation of an industry standard is the use of ML on HFC node splits. Network operators commonly use node splits to provide greater bandwidth and capacity to a given geographic area. Because node splits require significant labor and capital investment, cable operators typically must prioritize where to invest. This prioritization process has required an intensive manual effort in the past. Using machine learning to solve this challenge, an algorithm considers multiple variables including service load and cost to provide an actionable and prioritized report for the cable operator to act on. By applying ML to automate node splits, the network will run more efficiently and customers will continue to receive their high-speed services without interruption as the network grows.

The working group is also looking at creating standards to control video piracy by applying artificial intelligence on the network that detects signatures of bad actors. Initial findings indicate that video piracy can be significantly diminished with this use of AI. The development of a standard for this application would provide incredible benefits for content creators, streaming services, and film production companies, among others. Benefits like these emphasize the importance of having technology experts from outside of the cable industry collaborating with cable experts on these working groups.

Machine learning is also being applied to spectral impairment detection across the access network which allows for an automated diagnostic report and mitigation activity. Spectral impairments, from a variety of sources such as external signal interference, account for a significant portion of infrastructure issues. The rapid identification of impairments allows for fewer disruptions to the network and to the user. A standard for this application would help all cable operators and improve network service for everyone.

These examples are only the start of how AI and ML will help in building and operating more complex and robust networks and services. With standards averaging six months to two years to develop, the group expects to start publishing standards in 2021 related to AI/ML. And, over the next few years, the uses of machine learning to optimize network operations is expected to grow significantly which will lead to new demands for standards.

The AI/ML working group is one of seven working groups that make up the Explorer initiative. Each group represents industries, technologies, or practices that will place significant demands on telecommunications infrastructure, including telehealth and telemedicine, aging in place, autonomous transport, smart cities, and more. As the cable industry pushes towards 10G, opportunities for the advancement of emerging and yet-to-be imagined technologies continue to grow. It is crucial to ensure these advancements are met with industry standards to usher in a new era of connectivity and allow businesses to ensure that their products and services are optimized to reach customers across the broadband network.

About the Author

Chris Bastian is senior vice president and CTIO at SCTEISBE, the not-for-profit member organization for cable telecommunications. Bastian heads SCTEISBEs ANSI-accredited, award-winning Standards program. Prior to joining SCTEISBE, he spent 15 years in leadership roles at Comcast. For more information, visit scte.org/explorer.

Related

Read the rest here:
AI/ML in Broadband Networks: the Role of Standards - EnterpriseAI