Some readers may find this blog a bit provocative because it illustrates that  the assumptions and experiences that we have about how to achieve massive energy savings, are not necessarily true in for telecom networks.

The reduction of carbon footprints and energy bills require major changes in the way we live and work. Here are just a few examples of how we will have to radically change to achieve these goals:

• Making less flight travel to save CO2 emissions,
• Switching from individual cars to public transport to get to work and
• Consuming less and different goods involving a far lower carbon footprint along their entire product lifecycle
• A lower carbon footprint of buildings has to be bought by significant investments in new heating systems and the insulation of walls.

Network energy savings do not require changes in the network

The good news is this is very different for mobile telecom networks. Massive energy savings do not necessarily involve large scale infrastructure transformation. As pointed out in my last blog, 30% energy savings are possible within weeks just with the help of AI by using software that is running as an overlay control of existing radio infrastructure and network controllers.

Communication Service Providers (CSPs) are not forced into large-scale hardware re-deployments, comprehensive network modernization or architecture re-designs. To start, all that is required is inventory, configuration, traffic profiles, alarms data coupled with an AI algorithm with Machine Learning capabilities.

This provides an easy deployment without pre-investments. If an outcome based business model is involved, where the payment to the supplier of the energy management solution is purely linked to the achieved savings, there is no business risk at all.

This is excellent news for CSPs but what is the impact on the end users?

More good news: Network energy savings do not affect network performance or the user experience

Running network resources that are not in constant use provides redundant resources which maintain high network availability even if other resoures fail however this also leads to significant amounts of wasted energy. Network traffic fluctuates across time and location.  This means thatdifferent parts of the RAN infrastructure in a givenarea,such as whole sites, certain sectors and/or certain layers, can be put into sleep mode for defined timeblocks.

The question that arises is how can we guarantee that network performance and customer experience aren’t suffering at any time when parts of the network resources are unavailable? How do we make sure that resources are powered up again right in time when traffic is peaking again? In other words, how do we ensure that network performance requirements and energy consumption are perfectly aligned?

A problem of such complexity calls for an AI driven, automated energy management solution. AI based energy solutions can predict precisely the right time to power off resources and power them on again. Just-in-time waking is hard to achieve with static or rules-based methods, usually requiring wide wake windows or extensive use of stand-by mode to shorten wakeup times. This is necessary to prevent a poor customer experience when traffic starts increasing again. On the other hand, AI can manage the complexity of aligning energy saving needs and customer experience requirements.

That was exactly the point for China Mobile who needed a solution that would cut energy consumption and control costs without compromising the customer experience. The company realized it needed a comprehensive energy efficiency plan to reduce emissions and lower costs -  but was adamant those mitigations could not affect the customer experience or compromise network performance. China Mobile is using Nokia AVA Energy Efficiency solution for:

• Predictive and dynamic management of passive and active components rather than applying fixed schedules for powering on and off — to gain much finer-grained control over energy consumption and not affect network performance or the quality of customer services
• Predictive closed loop actions for faster, automated responses to changing conditions — maintaining quality and energy optimization — instead of relying on manual interventions that cause delayed responses
• Automated remote antenna control to adjust coverage dynamically in accordance with shifting capacity requirements

Using Nokia AVA AI as a Service, China mobile was able to permanently balance energy savings and performance requirements, allowing KPIs to be pre-set, with savings calculated by the AI system.

Different KPI setups can be used for base stations of different areas such as urban or rural. An AI system permanently monitors the network performance, stopping the saving any time an unexpected network performance degradation occurs. and correcting the cell states in case of faults. The system also provides operators with the ability to manually intervene at any time as needed.

Network energy savings and performance management require adaptations all the time

Network performance is not a static KPI. Therefore the AI system dynamically calculates the threshold based on the required performance every day for every site. This leads to widest possible energy saving windows while delivering the desired customer experience. In the example below, the AI system is able to expand the saving window by 86% on the first day and by 57% on the second day, on top of what has been calculated by the legacy BTS and SON (Self Organizing Networks) power saving features that rely on strict traffic dependent rules.

AI systems do not only focus on the energy saving potential of single radio sites but also coordinate multiple neighboring sites to achieve the best overall power saving result within the coverage area. For example, in case of cross coverage, one site can get entirely powered off during times of low traffic while the area is sufficiently served by the adjacent sites.

There is a plethora of further innovative methods for balancing both energy consumption and network performance. Depending on coverage and capacity requirements, AI dynamically adjusts the azimuth and elevation angle of antennas to focus energy accurately and precisely where the capacity demand is. With full automation, there is no need for personnel to go to the station to move the antenna.

Conclusion: Contrary to most areas of daily life, energy savings in telecom do not require massive lifestyle change and do not have an impact on the services and the experiences customers are used to. Required change are hidden by AI as it masks the underlying complexity and does the difficult work.

To learn more about Nokia AVA for Energy Efficiency AI as a service solution, take a look at our webpage.