A holistic approach to address RAN energy efficiency

Over the past 10 years mobile traffic has grown by a factor of almost 300. With the technology advancements service providers have managed to handle this traffic growth with only a 64 percent increase in global energy consumption - from 91 Terawatt-hour (TWh) to an estimated 150TWh. While mobile networks already today enable a net reduction in overall energy consumption for society, it is fair to say that mobile networks themselves do consume a considerable amount of energy. The estimate of 150TWh mentioned above is equivalent to 0.6% of global electricity usage or the annual electricity usage of a small country like Sweden – and for service providers it corresponds to a US$25 billion yearly spend.

As we move ahead with 5G, the energy stakes are set to get even higher. While 5G technology promises new and better services, this can only happen if operators deploy radios that support the new 5G frequency bands at a broad scale. Hence, our and other telecom vendors’ challenge becomes how to scale 5G to meet consumers and industries expectations for new and improved services, manage the four-fold traffic increase expected by 2025, while simultaneously aiming to reduce the absolute network energy consumption.

Ericsson Radio System energy efficiency approach

The nature of the Radio Access Network (RAN) is to provide nation-wide coverage and service capacity. As the coverage and capacity offered by a single radio is limited, this is achieved by deploying thousands of radio sites from which signals can be transmitted and received. The number of radio sites and radios can often be a factor of 10,000 or more when compared to the number of core network nodes. Due to this huge difference in numbers, it is not surprising that the RAN represents more than 75% of the service provider’s network power consumption. RAN energy efficiency is and will grow in importance for service providers because it is the only way to keep energy consumption under control, while still delivering an excellent experience to the consumer and the industry.

Energy efficiency is on top of our agenda and it - together with the increased capacity demands - guides both the development of new components, products, and features. During the last couple of years, we have taken significant steps for example improving the energy efficiency of Massive MIMO radios with up to 50% and our multi-band radios with up to 20% compared to the earlier generation. At the same time our basebands are 30-60% more energy efficient than competition when delivering the same capacity. Together, this sets us on good path towards taking the next step to significantly reduce the power consumed per delivered GB.

Ericsson’s ambition is to provide high quality, high performing 5G experiences while simultaneously aiming to reduce network energy consumption towards 2025. To achieve this goal, we have multiple solutions to address RAN energy efficiency. These span the entire product lifetime, from both product design to operation in the field. Our approach can be summarized by:

• Designing the most energy efficient and capable products
• Optimizing energy consumption of deployed products with advanced software
• Operating the site intelligently

Let’s look at each of these in more details.

Designing the most energy efficient and capable products

The transition from 4G to 5G brings a huge increase in compute requirements – they have increased by a factor of more than 150. So, what is driving this change in 5G? Well, if we look at the processing requirements, what is driving 5G is really the Digital Front-End, layer one processing and beam forming. If we take LTE as an example, back in 2010, LTE mainly consisted of a single 20 MHz carrier, with radios equipped with two receive and two transmit branches, and a transmission time interval (TTI) of 1 millisecond. Fast forward to 2021 with 5G, we typically have a 100 MHz carrier, which is five times the bandwidth that we had ten years ago. Additionally, with massive MIMO radios, we have 64 receive and 64 transmit branches – 32 times that which LTE radios had at the time – and a transmission time interval (TTI) of 0.5 milliseconds. In other words, you have half the time to do 160 times more processing.

While general compute solutions can be used for 5G, to truly deliver 5G performance with the highest energy efficiency, you need purpose-built silicon. Ericsson’s System on a Chip (SoC) design is the optimal solution to achieve this. The dedicated, purpose-built Ericsson Silicon enables us to create smaller and lighter radios. When done right, the Silicon will be high performing, while simultaneously consuming less energy and generating very little heat. Within the Silicon itself, we have managed to increase the energy efficiency by a factor of seven from 2016 to 2021.

Another crucial component for achieving energy efficiency is the power amplifier (PA) in the radio, which is used to generate the signals to be transmitted. Typically, the PA stands for more than 60% of the radio’s power consumption although this naturally varies across products. As a result, the efficiency of the radio hardware can be optimized for the specific output power or configuration used by continuously integrating more discrete steps into a single package, adopting new technology such as high-efficiency GaN and wideband PA technology multi-band radios. Of course, these technology improvements need to be completed with the right software solutions – both residing in the radio and in the RAN Compute. This way, we can significantly lower the power consumption of the radio.

The highly efficient power amplifiers, in combination with the Silicon’s low heat generation, enable us to use passive rather than active cooling for the radio - while still maintaining the highly desirable small and low weight form factor.

This in turns leads to a more robust solution with less site visits e.g., for repairs as a consequence, which again have positive net impact on the environment. As a real-life example, with AIR3227, Ericsson and Vodafone cut their energy consumption in half when compared to the previous radio generation in a trial in London.

Figure 1: New Massive MIMO radio AIR 3227 improve energy efficiency instantly

The final technology trend worth highlighting is multi-band technology that allows us to effectively combine the functionality of several radio units into a single physical unit. The energy efficiency and the size/weight can be greatly improved by integration of more components into the same radio- as in our recently launched Radio 6626 - exploiting that traffic seldom or never peak in all bands simultaneously. One example is multi-band PA technology allowing power to be dynamically shared between different bands. This way, you can still get full peak power output in a band, but you do not need to dimension the PA for peak power in all bands and can thus use a less power consuming PA enabling power pooling.

Turning to RAN Compute, the Ericsson Silicon family SoC is a key component to offer a high performing solution that at the same time optimizes weight, size, energy consumption. Compared to the industry benchmark, the RAN Compute portfolio consumes 30 to 60% less power. (see figure below)

Figure 2: Industry leading RAN compute power consumption

Optimize the energy consumption of deployed products with advanced software

Once the radio products have been deployed in the network it is vital for the radio equipment to:

1. Maximize the amount of data transferred per consumed Watt when it is used
2. Consume very little power when it is idle and not used
3. Use innovative software solutions

In Ericsson we have developed an integrated hardware and software architecture distributed across the radio and RAN compute that allows us to take superfast decisions (each micro-second) considering the site configuration as well as instantaneous traffic conditions in a multi-standard environment. This architecture offers multiple ways to control and optimize RAN energy consumption using advanced software capabilities exploiting capabilities, algorithms and information distributed across the different RAN equipment. Examples of the recent and efficient features for optimizing energy consumption are:

• The deep sleep software functionality can reduce energy consumption by up to 70% of the Massive MIMO radios’ consumption during low traffic hours.
• Micro-Sleep TX reduces energy consumption by switching off component (e.g., Power Amplifiers) when no transmission is required. Thanks to the integrated architecture full network performance is performed by having the capability to switching them on again in microseconds, in time for the next transmission whenever new data arrives. As network performance can be secured it is applicable for all traffic scenarios and for every hour of the day and it is for example supported for any combinations 2G, 4G, 5G, NB-IoT.
• Moving traffic to the most energy efficient bands. Carrier Aggregation expands mid-band coverage with the help of low band. By moving data to a more energy efficient 5G band where service providers often have a 100MHz carrier, the energy per transferred bit can be reduced a factor 10. At the same time as consumer experience is significantly improved.
• Getting the most out of the hardware that remains on. Innovative solutions like Ericsson Spectrum Sharing offers service providers a way to deploy 5G on existing 4G band without sacrificing performance and at the same time re-uses already deployed hardware and to enable nation-wide 5G coverage from day one.

Operate the site intelligently

By making use of artificial intelligence (AI), service providers can operate site infrastructure more efficiently and minimize energy consumption proactively. As we described above advanced power saving software features such as Deep Sleep and Micro-Sleep TX can reduce the energy consumption of Remote and Massive MIMO radios. Service Continuity complements this capability introduced in RAN by dynamically determining when and on which sites to activate the functionality. Our intelligent RAN power-saving solution models the entire network’s cells based on the traffic behavior such as data volume and users. It uses advanced machine learning techniques, determining the deep sleep power-saving feature activations and deactivations autonomously across the network on a radio level granularity, without blocking or soft locking the cells. As a result, the sleep times can be extended by at least 13% and further increase power savings.

Another example is the Energy Infrastructure Operations (EIO), an energy management solution that uses AI and advanced data analytics to optimize energy consumption across the entire network infrastructure, including passive equipment. By using AI and advanced analytics, service providers can achieve an approximate 15% reduction of energy-related OPEX, and an approximate 30% reduction in energy-related outages.

Why our holistic approach matters

There is not one silver bullet that solves the RAN energy efficiency conundrum. We need to look holistically at the entire network and addressing RAN energy efficiency from all angles.

Ericsson will continue to deliver high performing network solutions, combined with world class energy efficiency. We believe it will be feasible to scale up 5G, while simultaneously aiming to reduce total energy consumption towards 2025. This can be achieved through a combination of state-of-the-art hardware, advanced software, and intelligently operating the network, and at the same time delivering the best end-user experience for the consumers and the industry.

Read more:

Explore the Tech Unveiled blogs and videos

Visit the Ericsson breaking the energy curve

Read more about Ericsson Silicon

Read more about  Revolutionizing Massive MIMO