- A growing number of MNOs are benefiting from AI-powered RAN energy-saving tech
- BT and Three UK have recently deployed cell sleep software
- Ericsson has integrated a number of energy-saving features in its radio system portfolio
A number of mobile network operators (MNO) are adopting what is becoming an increasingly efficient way to reduce how much energy their radio access networks use: turning radio capacity off when it is not required.
They are now benefiting from artificial intelligence (AI) and machine learning capabilities that facilitate the autonomous management of radio networks, rather than relying on manual intervention.
In the U.K., for example, BT is using energy-saving "cell sleep" technology that puts certain 4G LTE capacity carriers to sleep when the capacity is not needed, based on predicted periods of low traffic that have been established for each site through machine learning.
According to BT, the technology has been implemented nationwide at sites operated by EE, its mobile subsidiary. An even lower power state, "deep sleep," can also be activated if required, for example during overnight periods of extremely low demand.
The use of machine learning is key here, as it enables the system to be automatically turned on and off as required. As noted by BT, the system “automatically wakes up during busy periods, and is also configured to react to unexpected surges which might occur during scheduled sleep modes.”
Greg McCall, chief networks officer at BT Group, observed that there is “huge potential” for energy savings across networks by dynamically matching power consumption against network usage.
“The optimization and rollout of cell sleep technology to over 19,500 sites across the U.K. is a significant milestone in achieving this, and an important development in countering the massive growth in data consumption we’re seeing across our networks,” McCall said.
BT said the technology is expected to produce energy savings of up to 2 KWh per site per day, or 4.5m KWh per year across EE’s estate. It added that its networks account for around 89% of its total energy consumption, and increasing network energy-efficiency is integral to BT’s ambition to become a net zero carbon emissions business by the end of March 2031.
Other measures included the switch-off of EE’s legacy 3G network earlier this year and upgrading to more energy-efficient RAN equipment, “including the deployment of ultra-lightweight Massive MIMO technology which uses up to 40% less energy than the previous generation of radio equipment.”
Three UK also announced last week that it has been working with Ericsson to improve network energy performance, with measures including switching off radio components on 4G and 5G networks.
BT did not specifically mention any vendor names, noting only that the "cell sleep" and "deep sleep" functionality is provided by the respective RAN equipment supplier on each of EE’s sites.
Ericsson and Nokia are now the main suppliers of BT’s 5G RAN equipment. According to a report from Denmark-based consultancy Strand Consult, 65% of BT’s 4G RAN network was provided by China-based vendors such as Huawei. Meanwhile, the group has stripped Huawei equipment from its core network, and has previously indicated it is on target to meet a deadline of 2027 to remove Huawei from the radio access part of its mobile network.
Energy-saving portfolio
Ericsson has certainly been talking about cell sleep and related technology for some time, previously noting that the RAN accounts for nearly about 80-85% of overall energy consumption.
“Depending on the geographical location and varying data traffic loads, it would be wise to put some of the capacity cells into sleep and wake them based on the traffic demand,” the vendor said in a January 2023 blog.
According to Ericsson’s web site, other MNOs to have employed Ericsson technology to put RAN cells to sleep include Elisa in Finland, Singtel, Canada’s Rogers, Indosat Ooredoo, Umniah in Jordan and Taiwan’s Chunghwa Telecom.
As well as cell sleep, the vendor’s radio system portfolio includes features such as energy metering, micro-sleep Tx, low energy scheduler solution and MIMO sleep mode.