Vodafone UK trialled Ericsson AI and machine learning (ML) software on 5G sites in capital city London, cutting power consumption of the radio units by a third, and tackling key operator concerns around expenditure and the environment.

The operator tested various 4G and 5G sleep modes along with a heatmap covering radio units, a combination Ericsson states assesses power consumption then seeks to optimise it by forecasting likely demand.

Vodafone chief network officer and network director Andrea Dona said the Ericsson software enabled it to “significantly improve energy efficiency without impacting” quality of service.

He cited obvious potential benefits for Vodafone involving financial and environmental costs, knowledge the operator will employ it to “improve the efficiency of our network as we build 5G coverage across the UK”.

Ericsson’s 5G deep sleep set-up employs AI in predictive algorithms which unlock an ultra-low energy hibernation state the vendor asserts can reduce energy consumption by 70 per cent during periods of low traffic.

A 4G cell sleep mode orchestration function models the performance of a network to optimise snooze states, with the radio power efficiency map employing ML to highlight underperforming sites.

Ericsson UK and Ireland CEO Katherine Ainley said its work with Vodafone was a key step in moves to “make networks smarter and more sustainable”.