ÎÚÑ»´«Ã½

Energy consumption makes up a significant portion of the mobile network operator’s OPEX, estimated at between – a figure that is even higher in heavy diesel usage regions, such as Southeast Asia and Africa.

In 2020, the annual global energy cost for running mobile networks was about , a figure that is likely to be significantly higher now due to subsequent global economic challenges such as the energy crisis and surging inflation. RAN (Radio Access Network) accounts for the largest portion of this cost, almost of the network’s energy consumption, making it a prime target for energy-saving efforts. The result reduces costs as well as the CO2 footprint.

Several intelligent RAN energy-saving mechanisms, such as , etc., have been introduced over the years to reduce energy consumption in the RAN’s active units. Adding AI/ML enables dynamic learning to intelligently control those mechanisms and improve energy saving while keeping an optimal user experience. Initial trials have shown up to additional savings over traditional techniques. However, for industrializing RAN energy saving for operators, the solution has to be scalable, reliable, and replicable across a multi-vendor network environment.

Challenges in building an industrialized RAN energy-saving solution.

Industrialization of RAN energy saving is difficult, and the key hurdles are:

1. Technical complexity – the mobile network consists of multiple technologies (2G, 3G, 4G, 5G). Energy efficiency must be implemented across these technologies and respective vendors.
2. Performance and QoS preservation – operators must ensure that QoS is closely monitored, and energy-saving measures do not impact user experience.
3. Compatibility and interoperability issues in a multi-vendor ecosystem – networks are commonly built with equipment from different vendors. The RAN energy-saving solution must adapt to the respective data formats, KPIs, energy-saving mechanisms and APIs. This challenge can be mitigated by improving standardization of network KPIs (usage, electricity consumption, quality of service) and energy-saving mechanisms and APIs.
4. Cost considerations – implementing energy saving solutions may involve upfront technology and infrastructure costs. Moreover, the energy-saving solution will have its own CO₂ footprint. Operators need to carefully assess the return on investment and develop strategies for cost-effective energy savings.
5. Measuring the savings – operators need to separate the contribution brought in by the AI-based solution from vendor-provided RAN-level mechanisms.

The way forward for operators

To evaluate the potential benefits and gain confidence in their ability to overcome challenges, operators must engage in trials, starting with a simulation based on real network data. A simulation is risk-free, with no impact on the network, and can help evaluate returns, building motivation to engage in field trials and then deploy a better-tuned, industrial solution. With this approach operators can realize immediate energy savings.

In order to be ready for the future, operators and NEPs need to also accelerate the adoption of Open RAN. Through standardized interfaces, RAN virtualization, and AI/ML enabled use cases, Open RAN has the potential to play a significant role in accelerating the industrialization of RAN energy saving and drive innovation.

With Open RAN, it is possible to introduce near-real-time energy saving techniques in O-RU such as , and more. These use cases provide deeper and more advanced energy-saving techniques that are not possible in traditional RAN. Additionally, O-RAN can optimize energy consumption in the O-Cloud infrastructure with intelligent workload optimization and CPU tuning.

How are we contributing to this endeavor?

ÎÚÑ»´«Ã½â€™s RAN Energy Saving offering uses an AI-native approach for a holistic and open solution to industrialize energy saving in RAN that is fully compliant with 3GPP and O-RAN Alliance specifications, as shown in figure 1 below. It is open, vendor-agnostic, and customizable as per operator needs for seamless integration in a multi-vendor telco network. It uses advanced AI-native techniques, utilizing ÎÚÑ»´«Ã½â€™s NetAnticipate framework, to correlate various metrics and provide real-time and predictive operational intelligence.

The underlying closed loop automation framework ensures that the solution self-learns to continuously improve energy saving in the network. It can prove this value in simulation, based on a continuous feed of real data, and then switch to real action on the network to gain savings effectively.

ÎÚÑ»´«Ã½ has participated in TM Forum’s moonshot catalyst project “. This included collaborating with leading CSPs and NEPs to address the faced by the telecom industry. All while ensuring scalability, reusing across multiple networks, reducing cost and risks by leveraging TM Forum’s best practices and standards.

Exciting opportunities on the horizon

The advent of 5G-Advanced and anticipated 6G technology is going to bring even bigger opportunities. New technologies are on the way, like and . Together, these will make future networks more energy efficient and natively sustainable. ÎÚÑ»´«Ã½ is actively engaged in research on these technologies with academic and industry partners, for building a sustainable tomorrow.

TelcoÎÚÑ»´«Ã½ is a series of posts about the latest trends and opportunities in the telecommunications industry – powered by a community of global industry experts and thought leaders.