The authoritative news source for communications regulation
AI Spotlighted

Networks Becoming More Energy Efficient, but Work Remains, Experts Say

Carriers are making strides toward cutting energy consumption and becoming greener, but companies need to improve how they work together, speakers said Wednesday at a TelecomTV green network summit. Questions remain, including how AI can help make networks more efficient, while the use of AI technology by itself consumes a growing amount of power, experts said.

Start A Trial

The telecom industry “accelerated” the move to greener networks out of necessity, said Mark Gilmour, chief technology officer at data transport network provider ConnectiviTree. “It’s the right thing to do” but more important is the “impact to the bottom line,” he said.

Current high energy prices are forecast to decline, Gilmour said. The challenge will be to sustain momentum when that happens, he said. The model for the telecom industry needs to fundamentally change, he said: “It’s not sustainable economically, financially or climatically to just keep building duplicate networks. … That does mean we have to change the way we work with each other and collaborate with each other.”

Providers want growth, which is “why we exist -- to meet that demand,” said Ahmed Hafez, Deutsche Telekom vice president-technology strategy. Traffic is doubling every 12-18 months, he said. “We can deliver more for less,” he said. When operators build infrastructure they need to put energy efficiency “front and center” as part of the design, he said.

Hafez sees too much focus on hardware and chipsets. “The software has a lot to do,” he said. “Talking about AI … and new technologies like blockchain and others, they consume a lot of energy,” he said. Developers and software companies also need to optimize the software, he said. “Everybody, collectively, should work on” energy consumption “targets,” he said: “We need to work together much more.”

I see an industry looking for transformation,” said Jan Berglund, director-products and solutions at China’s Comba Telecom Network Systems. “It’s hard to do it, but we still need to do it. … At this point we’re mostly talking.” Berglund said it’s also increasingly clear industry needs to focus on the energy demands posed by AI technology, as well as potential benefits.

Collaboration “is happening," but “there’s more that we need to do,” said Manish Singh, CTO of the Telecom Systems Business at Dell Technologies.

We have to collaborate and then work upwards on the stack,” Singh said. The work of the ORAN Alliance on open radio access networks is also important, he said. “The way that the architecture is getting decoupled, interfaces are getting opened … provides a whole new opportunity” to use AI and machine learning to reduce energy consumption, he said: When network traffic falls below peaks “can we turn off bands, can we turn off cells, can we shape beams in the directions we need to?”

The challenge is how to manage peak usage, which is a problem that electric utilities have faced for “for ages,” said Luis Jorge Romero, European Telecommunications Standards Institute director general. When demand is lower, how does a carrier switch off servers so that they don’t consume power, while all the servers are still needed to handle peak usage? he asked.

Peak demand will grow, and we need to deliver more with, ideally, less resources and a lower energy consumption,” said Anita Dohler, Next Generation Mobile Networks Alliance CEO. Most energy consumption comes from the RAN, and industry is doing its “homework,” she said. Providers are optimizing architecture “using relatively simple measures like shortening cables, by going to bus-based [omnibus-based] architectures … which already is delivering quite a good improvement in the efficiency,” she said.

It’s possible to keep network energy consumption flat even as providers roll out 5G, said Francesca Serravalle, Vodafone UK head-infrastructure and energy. “An innovation mindset is key,” she said. Companies need to do short-term experiments to see if a specific technology works and “if not, let’s move on,” she said. Network modernization is also important, she said.

AI could help make networks become more efficient, and moving to cloud-native architecture means more efficiency, Serravalle said. Networks also need to become more “customer-centric,” she said. “We can deliver green networks without impacting service innovation,” she said.