What is 5G Energy Consumption?

The Information and Communication Technology (ICT) industry currently accounts for approximately 4% of the world’s electricity consumption. With 5G projected to increase capacity up to approximately 1000-fold and high frequency millimeter wave (mmWave) transmission driving exponentially higher cell density, this percentage could potentially exceed 20% by 2030, or an astounding 150 quadrillion Btu each year. 

Increased consumption has raised the importance of 5G energy savings for operators and service providers who already dedicate a considerable portion their OPEX budgets to power. At the same time, consumer sentiment and government regulations are continually steering the industry toward greener electricity sources.  

• Does 5G Consume More Power than 4G? 
  Based on data bits per kilowatt, 5G networks are 90% more efficient than their 4G predecessors. However, huge increases in density and traffic are expected to negate these savings, leading to a net 5G energy consumption potentially 4 to 5 times higher than 4G. New RAN elements like Massive MIMO and beamforming also shift and concentrate 5G power consumption, with a typical 64T64R massive MIMO configuration drawing over three times as much power as a conventional 4G LTE radio. 
• What 5G Means for Energy
  Edge computing, mmWave transmission, and widespread IoT adoption have made the increase in energy consumption unavoidable. This has led to renewed efforts to reduce waste and improve component and system energy performance. The 3GPP has defined energy efficiency KPIs for the entire 5G network as part of Release 17. Innovation and out-of-box thinking are being applied across the telecom industry to enable: 
  • Energy-efficient data processing, chipsets, and cooling to reduce baseline energy consumption.
  • Intelligent and dynamic control of the RAN and Core to selectively power down or reconfigure underutilized resources and reduce waste.
  • Improved installation and test processes to identify and eliminate wasteful sources of signal loss, RF interference, and overlapping cell coverage.
• 5G and Renewable Energy  
  The telecom industry accounts for approximately 2% of annual CO2 emissions worldwide. This will most certainly increase without a strong commitment to 5G renewable energy. The GSMA is dedicated to helping the telecom industry target net zero emissions by 2050. Over two dozen telecom operator groups have already committed themselves to science-based emission standards. Energy saving initiatives are pivoting to incorporate greener sources or produce more renewable energy off-grid. The improved performance and affordability of solar and wind power are expected to facilitate incorporation of renewable energy sources while also reducing OPEX. 

   

The 5G network is a dynamic system that consumes energy continually and responds to spikes in network activity. Over 70% of this energy is consumed by RAN antennas, radio units, and base station elements. Core/edge cooling and computing processes, amplifiers, and backhaul are additional areas where 5G energy efficiency can be improved. 

• 5G Base Station Power Consumption: With each base station carrying at least 5X more traffic and operating over more frequency bands, 5G base station power consumption is at least twice that of a 4G. For perspective, each 5G base station is estimated to consume about as much power as 73 households. The addition of high energy active antenna units (AAUs) contributes to this increase.  
• 5G Antenna Power Consumption: Massive MIMO antenna arrays, requiring an additional 1000 watts of power per sector, also influence 5G RAN energy consumption. RAN power can be reduced by limiting massive MIMO deployments to high traffic, urban areas. The integration of more C-band MIMO antennas lessens the impact, since less cells are required to cover the same physical area.  
• 5G Data Center Power Consumption: With 5G applications tethered to the cloud for Core computing, analysis, and storage, data center efficiency is an important consideration for 5G energy savings. Hyperscalers have improved their efficiency through virtualization and economies of scale, but data center operations can still account for up to 30% of 5G energy consumption.  

The explosive growth in edge computing installations to support the IoT also calls for ongoing hardware size and power reduction, along with improved automation to eliminate in-person operators. 

The concept of Open RAN was intended to promote open interfaces and standardization for 5G RAN infrastructure. Open-source software running on white box hardware improves both interoperability and innovation. The heightened focus on RAN energy consumption has added efficiency to the list of important Open RAN characteristics.  

• Open RAN Efficiency Gains: Open interfaces allow power saving measures to be rolled out across the supply chain quickly. A disaggregated, software-centric Open RAN also supports the shifting of computing processes to cloud data centers with greater economies of scale. Successful multi-vendor PlugFest results have shown that Open RAN can deliver the improved energy efficiency 5G providers are striving for, even under peak traffic conditions. 
• The RAN Intelligent Controller (RIC): The software-defined RIC optimizes RAN functionality and facilitating onboarding of third-party applications. The RIC receives real-time feedback from the RAN, then uses machine learning (ML) and artificial intelligence (AI) to support 5G low energy performance by balancing resources, strategically turning off cells and/or frequencies, and recalibrating network slices to optimize QoS. 
• TeraVM AI RSG Tester: Prior to deployment in the field, xApps and rApps for the RIC can be trained to optimize 5G energy savings in the lab. To maximize this opportunity, developers need to emulate RAN hardware, software, and traffic conditions to test RIC performance under real-world conditions. The VIAVI TeraVM AI RSG Test helps to reduce O-RAN deployment and operating costs with optimized efficiency out of the gate. This virtualized test tool allows developers to accurately assess the effectiveness of RIC applications and updates in the lab. 

Improved 5G energy efficiency with a shrinking carbon footprint can be achieved through a combination of innovation, conservation, and intelligence. Precision in antenna alignment along with fiber certification and monitoring to reduce signal loss and retransmission are other conspicuous opportunities for savings. While the O-RAN Alliance, 3GPP, and Next G Alliance are committed to 5G renewable energy and reduced consumption, they acknowledge that more can be done. 

• What General Practice can be applied? 
  The intelligence of the cloud and RIC can be used to optimize 5G power consumption by responding to demand in real time and applying resources where they are needed most. New 5G hardware and real-time network feedback support common sense 5G low energy practices including: 
  • Switching off inactive cell sites or placing radio units into sleep mode. 
  • Reducing power to lesser used antennas. 
  • Adjusting the digital tilt of massive MIMO arrays to optimize beamforming accuracy and efficiency. 
  • Deploying “lights out” data centers and edge computing locations to reduce baseline energy consumption. 
  • Establishing renewable energy sources close to cell sites to minimize power supply distances. 
  • Sunsetting inefficient legacy 2G and 3G networks. 
• Energy Efficient Architecture   

  Beamforming practices improve spectral efficiency, wider bandwidth supports higher cell capacity, and network function virtualization (NFV) reduces hardware operation and cooling power. Liquid-cooled base stations to replace wasteful air conditioning units, redesigned chipsets to improve computing efficiency, and gallium nitride amplifiers to increase power density are among the innovations contributing to 5G energy savings at the component level.

  Small cells extend the reach of 5G networks within dense urban areas and indoor venues. Improved battery technology, cooling, and backhaul connectivity present huge opportunities for RAN energy savings with millions of 5G small cells expected to be deployed over the next decade.

• Data Center Efficiency   

  The migration of cloud native 5G applications to hyperscale data centers introduces additional opportunities for network energy savings and reduced CO2 emissions. The IoT and artificial intelligence, backed by the combined power of 5G and hyperscale computing, enable unmanned data center operation and real-time monitoring to optimize cooling profiles and reduce power consumption.  

  Lights-out operation allows these data centers to be deployed in colder, remote locations (such as Iceland) with built-in cooling benefits. The sheer size and physical attributes of hyperscale data centers should lend themselves well to colocated solar, wind, or geothermal renewable power sources. Hyperscalers including Google, Amazon, and Microsoft have pledged carbon neutrality by 2030. 

• Location Intelligence 

  Effective 5G energy reduction practices can be established with the end user and QoS in mind. Location Intelligence translates real time data on subscriber and IoT device location, performance, and traffic patterns into targeted RAN energy saving opportunities. A digital twin is used to process the data and model adjustments offline so that user experience KPIs are not compromised, and may even be improved. This intelligence can also be used to plan new radio deployments and gather valuable market insight for operators.