How to run sustainable networks without sacrificing user experience

The increase in 5G base stations and deployment of large-scale multiple input multiple output (MIMO) technology have led to a significant increase in power consumption. Each 5G base station uses 2.5 to 3.5 times more energy than 4G base stations, which is a huge sustainability test for operators in terms of the construction and maintenance of base stations. Consequently, building eco-friendly 5G networks is one of the hottest topics in the field of operations management.
AI for a greener telco – Phase II is the second phase of a Catalyst proof of concept project which aims to show how the technology can be used to create a sustainable 5G future. The first phase was developed last year and shown at Digital Transformation World in May, 2019 (watch the summary video below). The premise was that as energy accounts for 30% of all operational and maintenance network costs, this is a serious issue both in terms of cost and sustainability that must be addressed.

Learning from LTE

The first phase carried out an assessment of energy usage and built models to find ways of using AI to make LTE base stations more energy efficient. The team developed intelligent LTE cell energy management to reduce the network’s OpEx and carbon footprint. It also measures the customer experience using an emotional connection score guard to prevent any negative impact on customers due to network’s performance. The solution devised by the team also addressed how to build energy efficiency into 5G network planning.

China Telecom championed the first phase of the project and is the also the second phase’s champion. The communications service provider (CSP) is supported by participants AsiaInfo, Beijing University of Posts and Telecommunications, BOCO Inter-Telecom, and ZTE.

As Bing Qian, Technical Director of China Telecom and the project’s co-leader, explains, “The vision of our project is to build a smart, energy-saving network that can create a stable customer experience for subscribers, lower power consumption for telco operators, and mitigate the greenhouse effect for our Earth.”

Different approaches

The team wanted to leverage and evolve the best elements of different approaches to energy saving (ES):

• Beyond one size fits all – Refine historical pattern recognition from different times to identify when energy could be saved, but take this down to individual cell level


• Intelligent ES prediction – Use time and space key performance indicators (KPIs) to reset energy saving parameters in advance


• Precise ES – Refine the solution based on users’ behaviour and cell-scene data, but without sacrificing the cell’s coverage to avoid damaging users’ experience


• Protective ES – Monitoring the users’ perception KPIs and enabling immediate activation of base stations to protect their experience as necessary.

This is the Intelligent ES Framework the team developed to achieve those goals.

One use case, Chengdu City’s high-tech zone which has 221 small cells and 13,770 pico remote radio units (RRUs) – at peak saved 9.5% and overall average of 8.7%. The estimated energy saving data in Chengdu City itself, which has 3,000 small indoor cells, and 180,000 pico RRUs was 8%, equating to a saving of 3.3 million kWh a year.

Using the KPI approach, the network remained highly stable throughout. The shutdowns and activations were automated using AI.

Many different factors

This second phase explored coordinated use of 4G and 5G to save energy, as well as four different kinds of shutdown, depending on the circumstances – symbol, channel and carrier shutdown plus hibernation mode.

The team built a user perception density center to collect and analyze data about users’ perceptions as well as their their positioning data. The analyzed data is input into a 5G coverage simulation which dynamically adjusts Massive MIMO parameters for more efficient usage of 5G resources.

As the graphic below shows, the automation and evolution of energy saving is a long-term process, divided into five stages, with the potential to save more energy as it progresses. AI and automation play an ever-bigger part, until ultimately energy saving is managed by a fully autonomous network without manual intervention.
The team demonstrated their project during the TM Forum Catalyst Digital Showcase in July 2020.

Now available on demand:
Watch the Catalyst team showcase their project