The authoritative news source for communications regulation
Intelligent Networking

Carriers See Bright Future for AI but Say Challenges Remain

Carriers are using AI and machine learning (ML) in potentially transformative ways, experts said Tuesday during day two of Fierce Network’s Cloud Native 5G Summit. But speakers also reminded attendees AI is in its early stages and has a ways to go before providers fully embrace it.

Start A Trial

Ross McWalter, head-data, AI/ML and generative AI at Amazon Web Services, described how AI can help detect problems and make networks smarter. Service disruptions and network outages can have significant effects on revenue, he said. “We leverage cutting-edge AI and ML technologies to minimize service downtime and ensure seamless operation,” he said.

AI can help providers prevent network disruptions before they occur as well as spot issues rapidly, he said. For example, AI can create an end-to-end representation of a network, McWalter said. This twin network approach offers a “comprehensive view” of network elements so that providers can identify faulty components “in real time,” which “enables rapid troubleshooting,” he said. Generative AI can recommend mitigation and recovery strategies when problems occur, he said.

AI makes wireless networks more efficient, McWalter said, citing AWS’ next-cell prediction technology. “By predicting which mobile cells users will connect to, operators can optimize network performance and manage cell-to-cell handover,” identifying congested cells and minimizing energy use, he said.

Telecom operations historically were static and siloed, said Brandon Larson senior vice president-cloud and AI at Mavenir. Automation can be used to install software, do upgrades, change configurations and scale the network, he said. These are things staff always did and can now be done “with less human intervention,” he said.

AI will go beyond what is done today through automation alone, Larson said. “We’re using AI to learn from the network,” he said: “It’s able to elevate network insights [and] problem-solving, and it’s also able to make more intelligent decisions” including on things like energy and spectrum efficiency, he said. “We’re able to use AI to forecast things that will happen in the network,” he said. AI is “complementary” to automation, he said.

AI is part of the conversation” for any new product the carrier offers, said Beth Cohen, Verizon advanced networking and security product strategist. But some people “mix up” AI and automation, she said. For decades, telcos have been automated, she said. “What’s new is bringing the insights that AI can bring.”

AI is more than large language models -- learning models that are pretrained using vast amounts of data, Cohen said. “I’m not 100% [sure] that the large language models are ready for prime time,” she said. AI “has a role to play” in operations, in capacity planning and in other areas, she said. At this point, industry is focused on business support and operations support backend systems, she said. There’s a lot of interest in bringing AI into network slicing and in other wireless technologies, she said.

Cohen said she prefers to call AI intelligent networking. “We still have a ways to go,” she said. “I’m not 100% convinced there’s a business case” for AI “now, but I think the business case will follow within a year or two,” she said. Verizon is using AI in operations, “but it’s not for the network per se, it’s kind of more operational support,” she said.

Carrier business models traditionally focused on being a “pipe” and keeping the network running, said Arun Bhamidimarri, architect-telco solutions at VMware by Broadcom. Monetization is about transforming telcos into tech companies “with agile business models” and a focus on automation, he said. “They can refocus on innovation and getting new services to the market” like network application programmable interfaces, edge services and network slicing, he said.

Carriers are also focused on cutting costs, Bhamidimarri said. “Business is a margin game,” he said. Making the change to cloud-native networks is critical, he said. The cloud provides “elasticity, resonancy, faster deployment and scale,” he said: “Look at hyperscalers. That’s the proof in the pudding. They have been able to churn out so many monetizable services by keeping their costs down.”