Artificial intelligence can reduce your carbon footprint by 30 percent now
The promise of an instant 30 percent energy savings sounds a bit like the lofty promises made by weight loss programs, which most people recognize as being unrealistic. But many communications service providers (CSPs) have committed to greenhouse gas reduction goals that are just as ambitious, including carbon neutrality by 2050 or sooner. And just as someone trying to lose weight needs to choose the most effective training and diet options to meet their goals, CSPs must also choose the best strategies that will enable them to shed carbon weight from their networks and achieve their Paris Accord commitments.
Yet the pressure doesn’t come only from the Paris Accord. It also comes from the cost side. Energy consumption already accounts for about half of all network operations costs, making it a very large cost block that requires attention at the corporate level. Unfortunately, most CSPs do not always have full transparency into where energy is spent — and many energy bills are not even entirely correct.
And the problem is expanding, driven by exponential traffic growth and the rising number of sites. Added processing capability at the edge of the network is leading to increased numbers of energy-consuming data centers. Regulations, extra demand from a growing number of electric vehicles and the transformation toward renewable energy production are also driving energy prices up.
These rising energy costs will put even more pressure on margins at a time when CSPs need to invest in 5G rollouts and are not prepared to manage further financial strain. For these reasons, CSPs need to act urgently to slash their energy bills and their carbon footprints. This blog looks at the ways they can do that.
Where to start? What strategy is most effective?
As I pointed out in my last blog, CSPs have considerable scope to cut energy consumption. Data transfer accounts for only around 15 percent of the energy used by a mobile network. That means the rest is simply wasted because equipment or whole sites sit idle during periods of low traffic and auxiliary systems — especially cooling systems — are often misconfigured.
There is real potential to reduce this waste and improve a network’s energy efficiency. Besides sourcing energy from renewables, there are five strategic opportunities to reduce the carbon footprint of a mobile network:
Base station efficiency: More efficient baseband processing and high-performance radio front-end amplifiers can achieve impressive energy savings. With such product related efficiency gains the greenhouse gas emissions from customer use of Nokia products fell from 42.9 Million metric tons CO2 emissions in 2016 to 32.4 tons in 2020 despite steep traffic growth. Such product innovation related efficiency gains require CAPEX and the physical deployment of new hardware.
Site optimization: Beside implementing more renewable energy production at the site, elimination of feeder and cooling losses provide a major opportunity to reduce the carbon footprint of radio sites. Japanese operator KDDI used Nokia’s Liquid Cooling AirScale baseband solution to cut its cooling energy consumption by more than 70% compared to traditional air conditioning systems.
Network architecture evolution: In addition to implementing energy-saving measures at individual sites, CSPs can also make improvements at the network architecture level. For example, cloud elasticity can improve resource utilization to avoid idle resources. But it should be noted that this is not a short-term solution. For many CSPs, cloud transformation is an ongoing strategic process and full savings will take time to achieve.
Network modernization: Phasing out legacy technologies at the right time can optimize CAPEX and OPEX and can often be justified by energy savings alone. Depending on the average age of the equipment in the network, a state-of-the-art base station may consume less than half the energy of the station it replaces.
AI-driven energy savings automation: AI can analyze large energy-related data sets such as traffic patterns, consumption patterns, status of network resources, site inventory, energy bills and weather information to make the best energy-saving decisions across active network elements as well as passive components such as cooling systems. The result is coherent energy control that dynamically adapts energy consumption to traffic levels while maintaining a premium user experience.
Why is AI-driven automation the most effective energy-saving strategy?
AI can optimize power savings for every base station in every sector, every day. For example, conventional energy-saving methods operate based on pre-defined static shutdown windows and are not able to handle complex savings scenarios. Nokia AVA for Energy Efficiency, on the other hand, predicts low traffic periods and shuts down resources such as frequency carriers or even whole sites in case of overlapping coverage. We have seen that AI-based solutions achieve two to five times more savings than non-AI systems that perform temporary shutdowns based on fixed schedules.
Real-world experience shows that AI-driven automation can be implemented in a matter of weeks, making it the most immediate opportunity for large cost savings. We have seen savings in real networks from seven percent to 30 percent. When the AI solution is multi-vendor, these savings apply to not just the equipment of a single RAN vendor but the whole network.
As a total-site, software-based solution, an AI system can be set up quickly to minimize all kinds of energy waste. It requires no large-scale deployments, network modernization or architecture re-design, and has no hardware impact. Once the data pipeline is established and the system trained a bit, it acts as an overlay and uses available data to start saving energy right away.
How to make AI-based energy management happen?
AI-based energy management systems are a mature technology so technical implementation is unlikely to be a barrier. That said, there are factors that CSPs should keep in mind to help facilitate their adoption:
Simple insertions overcome organizational complexity: Energy is a complex topic. Its consumption, measurement and cost responsibility are typically spread across multiple organizational units and functions, including network planning, network operations, field operations, facility management and procurement. Most organizations have no “Chief Energy Officer” overseeing the creation and implementation of an energy saving strategy. But AI-based solutions are quick and easy to implement, and they yield results fast. As a result, these solutions often act as a catalyst for a wider transformation.
Outcome-based business models overcome business risks: Traditional CAPEX-based investment models require proper business case calculations that consider business risk. With an outcome-based model, payment is directly linked to achieved savings, removing the business risk in advance.
Although no two networks are the same and energy saving results will vary, we have heard reports from customers that the results are “beyond expectations”. In other words, strategic AI-driven energy savings automation has proven itself to be the fast track to shrinking the carbon footprint of networks.