How KDDI used AI to cut RAN energy consumption in half
Case study: AVA for energy efficiency
KDDI’s goal for reducing carbon emissions
KDDI is Japan’s second-largest communications service provider (CSP), providing mobile services to more than 60 million subscribers across the country. Like many other CSPs, KDDI wants to help create a more sustainable, decarbonized future.
Reduce power consumption in base stations
In its Green Plan, KDDI aims to reduce carbon emissions by fifty percent in 2030 compared to 2019. One way it intends to achieve that target is by reducing power consumption in its base stations.
KDDI’s demand for an AI network energy management system
What KDDI needed was an AI-powered network energy management system that could assess real-time demand and traffic patterns, then automatically adjust the amount of power being consumed by RAN resources to match demand. To ensure the system could do all that without compromising the premium user experience expected by subscribers, KDDI worked with Nokia to trial Japan’s first-ever AI-controlled RAN.
How Nokia helped KDDI with telco AI and AI energy management
Shutting down unused network elements during periods of low traffic is an excellent way for CSPs to realize significant energy savings. With this pilot project, KDDI used the AI-driven AVA for Energy Efficiency to perform the precise predictions needed to balance power consumption, network performance and customer experience requirements.
Nokia AVA for Energy Efficiency uses AI to analyze and anticipate changing traffic volumes in the sites and cells of a RAN —to determine when radio resources can be powered down to reduce energy consumption. It also coordinates across multiple neighboring cells to achieve the best overall power savings within a coverage area.
AVA for Energy Efficiency benefits
On average, KDDI reduced power consumption by up to 50 percent in low-traffic environments and by up to 20 percent per cell. In addition:
AVA for Energy Efficiency delivered energy savings without any network performance degradations (such as traffic overflow in adjacent cells) or alarms, despite the wide sleep-state windows each day — for minimum impacts on the user experience.
It took only four weeks to set up the project, including the data pipeline needed for the AI algorithm — for faster energy saving results.
Learn more about AVA Energy Efficiency solution
Please complete the form below.
The form is loading, please wait...