5G Base Stations: The Energy Consumption Challenge

With the rapid growth of data traffic and the emergence of various network connectivity requirements, Mobile Network Operators (MNOs) in many countries are deploying their 5G networks to be ready for the next generation of the industrial revolution. These countries include South Korea, China, the U.S., the U.K., Saudi Arabia, Spain, UAE, Australia, Germany, and many others. As of Q2 of 2020, 110+ commercial 5G networks have been deployment globally, and solely in China, 650,000+ 5G base stations have been installed, and the terminal connections have reached over 160 million.

Although 5G is gaining momentum, several deployment and operational challenges have been troubling MNOs. Amongst these challenges, the most notable one is the energy consumption of a 5G base station due to the implementation of the massive MIMO technology and the level of network densification that need to provide a positive user experience. Early deployments indicate that 5G base stations require 2.5-3.5 times more power compared to a 4G one. Moreover, C-band, i.e., 3.4 GHz to 4.2 GHz, is deemed as the most popular 5G spectrum because of its balanced network coverage and rate performance. However, the deployment of a 5G network in this band still requires 3-4 times more installed 5G base stations to provide the same coverage as a 4G network.

According to ABI Research analysis and certain infrastructure vendor statistics, the typical three 5G massive MIMO 64T64R AAUs at a site need to consume more than 2600 watts, which are nearly 30% of total power consumption of the main equipment of a base station deploying a mix of 2G, 3G, 4G, and 5G radios. Apart from the radio unit, the most power-hungry component is the baseband processing unit, where several baseband processing functions, such as DPD, CFR, beamforming, and channel coding, consume more than 70% of its consumed power.

Traditional RAN vendors run baseband processing functions on special design hardware, that are typically powered by energy-efficient Application Specific Integrated Circuits (ASICs) or System on a Chip (SoC) platforms such as Nokia’s Reefshark. However, with the advent of network virtualization, many baseband processing functions are expected to run on commercial off the shelf (COTs) server with acceleration card(s). For a distributed RAN architecture, virtualized baseband functions allow MNOs to deploy 5G/4G/3G/2G networks on shared hardware to cope with different connectivity requirements. For a centralized RAN system, cloudified infrastructure management can flexibly allocate radio resources based on specific traffic needs.

When considering the deployment of a 5G base station, COTs server plus acceleration card(s) based baseband processing architecture still cannot compete with the ASIC or SoC based processing equipment in terms of the energy consumption, especially for those power-hungry time-sensitive processing functions as we mentioned above. On one hand, MNOs are trying hard to improve network performance, especially when it has now become clear they cannot charge extra for a 5G subscription. On the other hand, they are still struggling to deal with high energy consumption of 5G base stations, even though baseband processing functions are running on ASIC or SoC based hardware.  

By introducing more advanced radio access technologies, such as dynamic radio resource control and network automation, 5G network can significantly improve energy-efficiency, e.g., energy per bit, over previous generations of wireless technologies. However, high energy-efficiency does not necessarily mean lower energy/electricity consumption for 5G base stations. Besides, the adoption of C-band or mmWave spectrum requires more site and antenna per site of a 5G network deployment, which results in greatly scaled energy/electricity consumption cost. MNOs are working hard to find efficient solutions to solve the problem. From the technology point of view, some promising solutions are listed below.

• Artificial Intelligence (AI) and Machine Learning (ML) enabled traffic planning and dynamic radio resource management to dynamically control the length of radio and baseband processing hardware sleep mode.
• Renewable base station energy supplementary solution, such as solar or wind, to guarantee more reliable and green power supplies and reduce the environmental impact.
• The development of more advanced technology in terms of material to reduce the energy cost of massive MIMO radio units and the creation of different antenna/radio solutions to adapt different traffic requirements.

Alongside technical improvements, governments should be positioned and help reduce the electricity bills and offer subsidies for base station constructions and maintenances. Besides, the promotion of effective RAN sharing strategies among different MNOs can also help reduce the total energy cost and ease the burden of MNOs.

Although the energy consumption of 5G base stations is higher than any previous generations, technology and strategy innovations mentioned above would help MNOs stabilize or even reduce the non-renewable resources consumption and generate a greater positive environmental impact. The accomplishment of these targets will need MNOs and their network partners to be more open and proactive at testing and developing innovative approaches, thereby accumulating more experience and being ready for new challenges.