How Globe Telecom uses AI for mobile network energy efficiency

The radio access network (RAN) is notoriously the most power-hungry part of telecom networks. Base stations account for more than 70% of an operator’s total energy consumption, according to GSMA Intelligence. Under pressure to cut energy costs and carbon emissions, operators are increasingly turning to artificial intelligence (AI) to make mobile networks more efficient.

One of the latest is Globe Telecom in the Philippines, which announced this week that it will use Nokia’s AVA Energy Efficiency software to reduce power consumption and energy costs. The Software as a Service (SaaS) solution uses AI and machine learning algorithms to automatically switch off idle equipment when not in use.

In a proof of concept with Globe, the software contributed to the equivalent of annual power savings of 3% to 6%.

The solution is complementary to the operator’s existing energy saving capabilities in the RAN, explained Gerard Ortines, Head of Network Solutions and CAPEX Management at Globe Telecom.

“It uses AI/ML to learn the energy utilization pattern of each network element, like the base station. In turn, it reduces energy cost without compromising the network performance and is easy to implement… [The solution] enables dynamic energy consumption through advanced algorithms based on real-time network traffic while maintaining premium mobile user experience,” he said.

The AVA software has been applied to Globe’s 4G network and is currently integrated in Nokia equipment located in the country’s southern regions of Visayas and Mindanao. Ortines noted that it is possible for AVA to work with other suppliers’ RAN equipment, but that the operator needs further testing with its equipment ecosystem.

AI is one part of the answer

For Globe, AI is an important complement to other energy saving programs. “AI/ML plays a vital role in energy efficiency and contributes in the realization of power savings while providing better customer experience,” said Ortines.

Automated power saving is one of Globe’s initiatives for deploying more energy efficient equipment and “green network solutions.” These include rolling out wideband remote radio units (which use less power than single-band RRUs by supporting multiple bands from one antenna); upgrading microwave links to higher capacity fiber; and decommissioning end-of-life equipment.

According to its 2022 Annual Report, Globe has deployed more than 9,000 green network solutions since 2014. This and other efforts have resulted in a 4.42% reduction in Scope 1 and Scope 2 emissions for last year. The operator aims for group-wide net zero emissions by 2050.

Richard Webb, Senior Analyst at TM Forum, said that for the industry’s climate ambitions, “AI on its own is not enough, though it is a powerful tool.”

“For CSPs to achieve net zero targets they have to deploy a wide and deep range of available technologies and techniques. Whether it is 3G network turn-off, investment in renewables, generating excess power through solar at cell sites and many other things, they all need to be considered. Achieving net zero against the rising tide of data usage is a challenge that AI alone cannot solve,” said Webb.

Moving beyond PoCs

More telcos are embracing AI-based energy saving software. Nokia, for example, said it has more than 50 customers for its AVA solution. These include China Mobile, KDDI, and Telefónica Germany.

“The telco industry has accepted that AI is central to any effective energy reduction strategy. With rising traffic, increased energy prices and the sheer complexity of modern networks there is no longer a debate about whether to use AI or not,” said Andrew Burrell, Head of Portfolio Marketing, Business Applications, Cloud and Network Services at Nokia.

He said the industry has “moved beyond the proof-of-concept stage” and is now in “an intermediate stage where live deployments are not always scaled across the entire network.”

“PoCs remain a required step in the deployment of any new AI solution, but there are multiple competing solutions from vendors and, unlike some years ago, CSPs are not getting stuck in ‘PoC purgatory,’” he said.

According to Tim Hatt, Head of Research at GSMA Intelligence, AI-driven energy efficiency applications are focused on shutdown and sleep solutions in the RAN as base stations are the “low-hanging fruit,” accounting for most of an operator’s energy use.

“Besides shutdown solutions, AI will be able to improve energy efficiency with new features: load balancing; more intelligent beam forming; reducing interference; and better spectrum utilisation,” he said.

Furthermore, he said operators are also starting to use AI in other energy saving ways outside the RAN, such as for “predictive maintenance and enhanced troubleshooting to reduce the number of site visits and save on fuel and human resources; network planning support to not only save on resources but also create a more optimal end results; optimised fuelling and reduction of generator run hours; and equipment lifetime optimisation.”