Skip navigation

Automating MIMO energy management with Machine Learning and AI

AI helps reduce energy consumption cost
The opportunity for Machine Learning in this case is the ability to identify patterns. It's a fantastic way to learn from experience, like when we are children.
- Francisco (Paco) Martin, Vodafone Group Head of Radio Products

Ericsson Vodafone SS MIMO

Today's mobile customers want to stream and share content with the world – and they don't want to wait. With 27EB of data being used across 6 billion mobile subscriptions per month, additional hardware is needed for networks to keep up with increasing user demands. However, running additional radios at cell site locations takes time, money and electricity.

What if a software module powered by a Machine Learning (ML) algorithm, could solve this problem by putting radio transmitters into a power-saving Sleep Mode when user traffic falls below a certain level, and then re-activating them when traffic surges, automatically?


  • MIMO hardware at radio towers increases network capacity, at additional cost in power.
  • Engaging MIMO Sleep Mode when units aren't active could conserve energy across the network.


  • A Machine Learning (ML) algorithm was developed to observe, predict, and respond to user traffic.
  • Six trial radio units were identified across urban and rural locations.
  • ML algorithms were exposed to four weeks of traffic data at each site, then allowed control over Sleep Mode activation at each site.
  • Manual activation by operators was used as the benchmark to beat.


  • ML Sleep Mode management delivered 14 percent savings in energy consumption at each site, outperforming manual management.
  • Customer KPIs were maintained during automated tower activation or deactivation.

Optimization through automation


The algorithm was trained on real data from six radio sites in Portugal using MIMO technology to identify traffic patterns. ML then took control of power when activity at a radio site dropped below a learned threshold, customer data was routed to a nearby site, and the low-activity unit powered down until demand increased.

The results were conclusive, achieving energy savings of up to 14 percent with ML, at no impact to users.

Tailoring energy efficiency at each site could be introduced across the Vodafone network, and the 14 percent energy saving at the international level could make significant savings in cost and environmental impact.