Building on more than 20 years of partnership, Ericsson and FET carried out a service continuity trial to see how AI and machine learning-powered solutions could minimise energy usage while enhancing network performance.
Ericsson’s Service Continuity AI app suite helps to optimise the network and solve issues before they occur by using advanced algorithms to identify potential adjustments.
In the early part of the trial FET used the energy-saving solution during off-peak nighttime hours, but later decided it could be relied upon to apply the solution across a full day, including busier periods.
Nello Califano, head of strategy Ericsson Network Services, says: “The collaboration showed that FET was able to make radio access network (RAN) energy savings of 25% with the solution running over the course of a full day at a balanced level of aggressiveness. Crucially, this occurred without any abnormal impact on network performance KPIs. In other words, energy efficiency was improved significantly without any noticeable impact for the network’s end user.”
Data from the collaboration shows that a more aggressive approach could deliver as much as between 32% and 46% RAN energy savings. Deployed over a longer timeframe, there is potential for the AI to become more effective as it gains more input data from the network and learns accordingly, which could lead to further savings.
Communications service providers (CSPs) are estimated to be responsible for about 2% of global energy demand, making the need for greater energy efficiency in the telecom sector more pressing than ever.
Chee Ching, president of Far EasTone (FET), says: “Radio Access Network stations’ electricity consumption accounted for 71% of our electricity bill every year, regardless of savings in other network components – that’s huge. Anything Ericsson can do to improve network energy efficiency is of major value therefore.”
She adds: “We have always been looking for innovative ways to improve energy usage. The solution Ericsson deployed worked very well: with a relatively conservative level of activeness the AI solution made a 25% energy saving – that’s a lot. The result is already very impressive and doing it at a larger scale could be even more impressive still.”
How the Ericsson solution works
The deployed solution uses machine learning prediction model, regularly analysing data in order to make a decision: should it close a component down, should it activate a component, or should it do nothing?
That decision is predicated on predictions based on the live network data it is trained on, and also active monitoring of activity in neighboring cells.
In addition, FET deployed Ericsson’s Node Radio Power Efficiency Analysis service to create a map of all the cells in the network and calculate the energy efficiency of each cell. The data was then analysed to identify potential waste and faults, allowing FET engineers to intervene and repair where necessary.
The collaboration showed that the solutions can be trusted to deliver energy efficiency benefits while also helping to maintain service quality, and Ericsson and FET now plan to roll out the solutions at a larger scale.
Follow us and Comment on Twitter @TheEE_io