Microwave Journal
www.microwavejournal.com/articles/35925-g-and-network-wide-innovations-providing-energy-savings-with-lower-power-consumption

5G and Network-Wide Innovations Providing Energy Savings with Lower Power Consumption

May 13, 2021

The impetus for reductions in energy emissions in the telecoms sector is anchored in the global fight to combat and mitigate climate change, as enshrined in the 2015 Paris Agreement. Urgency has grown markedly recently as governments seek to garner private sector commitments toward the central objective of keeping a global temperature rise this century to a maximum of 2°C above pre-industrial levels. This implies net zero for most countries by 2050.

In our report, “5G Energy Efficiencies, Green is the New Black” from GSMA Intelligence, we look at how the telecoms sector has emerged as a vocal supporter of proactive emissions reductions plans. This is enshrined in an industry-wide commitment to reach net zero emissions by 2050 and bolstered by a growing number of operators that have embraced carbon reduction efforts as a core business objective, with strict reporting targets.

For mobile telecom networks, 5G new radio (NR) offers a significant energy efficiency improvement per gigabyte over previous generations of mobility. However, new 5G use cases and the adoption of mmWave will require more sites and antennas. This leads to the prospect of a more efficient network that could paradoxically result in higher emissions without active intervention.

The mix effect of 4G LTE and 5G upgrades in emerging and advanced economies (led by the U.S. and China) will result in these technologies accounting for 60 percent and 20 percent of the global mobile connections base, respectively, by 2025. The impact of this shift will be a continued rise in mobile data traffic, estimated at 6.4 GB per user per month in 2019 and forecast to grow threefold on a per user basis over the next five years. When combined with the rising costs of spectrum, capital investment and ongoing radio access network (RAN) maintenance and upgrades, this means energy-saving measures in network operations are necessary rather than nice to have.

NETWORK COSTS AND PERFORMANCE

Irrespective of climate change, impetus for energy-saving measures from telecoms operators has grown because of sustained increases in network costs in a low revenue growth environment. The telco business model is based on network scale. In times of growing revenues, margins expand as the largely fixed cost base is monetized (positive operating leverage), unless the operator is sub-scale. This is broadly what happened in the 2G and 3G eras in the 1990s and 2000s when mobile phones were still new to people and subscriber growth consequently steadily rose. However, in periods of low or negative revenue growth, fixed costs are exposed, with resulting pressures on cashflow and longer-term investments. Network capital investments have increased to fund LTE and early 5G rollouts, while free cash flow margins have been mostly preserved through reductions in personnel and other costs.

Figure 1

Figure 1 Energy (electricity and fuel) is a significant cost element for mobile network operators as it is one of the largest operating costs (a). Mobile data traffic will grow almost threefold by 2025 and new 5G use cases will require more dense and complex networks (b). These are averages so will vary by operator. Source: GSMA Intelligence.

Given the industry imperative to invest in networks, Capex is followed more closely than the costs of ongoing maintenance (Opex). However, this is changing with the rapid adoption and incorporation of energy efficient technologies. Both offer material savings in Opex. Network Opex tends to account for around 25 percent of the operator cost base, or 10 percent of revenue. Around 23 percent of network opex costs are spent on energy consisting mostly of fuel consumption as shown in Figure 1a. Most of this spend powers the RAN, with data centers and fiber transport accounting for a smaller share.

The good news is that the shift from fossil fuels to renewables has started to feed through to Opex savings, as have the phased retirements of legacy 2G and 3G networks that are less energy efficient than LTE or 5G NR standards. Looking ahead, however, as LTE and 5G progressively account for larger shares of the overall global user base, data traffic rises are inevitable - and with this comes the pressure on energy consumption (see Figure 1b).

There is no one method of increasing energy efficiency or reducing power usage. Instead, a mixed approach is generally being used, comprising renewables, AI-driven network sleep states, more efficient batteries and decentralized site deployments with compute power pushed toward the edge. The results will feed through over a period of years.

ENERGY CONSUMPTION IN RUNNING A NETWORK

Past transitions to new wireless standards have entailed a significant improvement in the cornerstone metric of energy efficiency: kilowatt hours per gigabyte. Though 5G NR also offers a significant energy efficiency improvement per gigabyte over previous technologies, new 5G use cases and the adoption of mmWave will require more sites and antennas. This leads to the prospect of a more efficient network that could paradoxically result in higher emissions, without active intervention.

One way to visualize a network is as a linear progression of stages, or phases, across which energy flows to provide power to base station sites, radio access nodes and backhaul links. Figure 2 outlines this journey, starting with energy sourcing from the grid and moving through to site and equipment consumption.

Figure 2

Figure 2 End-to-end energy loss from grid to the RAN. Source: GSMA Intelligence.

PHASE 1: ENERGY SOURCING

There are significant daily fluctuations in energy demand, while electricity supply is relatively static. This makes the price of industrial electricity vary on an hourly basis. Daytime energy prices in peak hours can be significantly higher than off-peak times during the night. Network operators can save money by buying energy at off-peak hours to be stored and used when data traffic peaks in the early evening, typically between 18:00 and 22:00. Operators can also sell excess peak-time energy back to the utility grid. This relatively nascent practice will require further investment in energy storage technology and batteries alongside partnership agreements with utility providers. Updated central energy management platforms to calculate and forecast site-level data traffic and consumption are also important to ensure each site has enough backup energy to operate safely, with no service interruptions.

PHASE 2: ENERGY CONVERSION

Energy utility providers sell AC electricity, while most of the site-level energy consumption happens in DC. For this reason, each cell site needs to have a rectifier module to convert AC to DC. Most cell sites in a typical portfolio are more than 10 years old and operate with less efficient passive infrastructure, including the rectifier module. New rectifier technology is much more efficient so it is a big opportunity for improving energy efficiency. The cost to upgrade is significant up-front but crucial to more efficient conversion and lower consumption over the long-term. New rectifiers are also key to cover the potential increased energy consumption before installing 5G equipment and ensure smooth capacity expansion.

PHASE 3: ENERGY TRANSPORTATION

When power is transmitted at high voltages, the efficiency of energy transportation increases (with a lower rate of leakage) because of a lower electric current in the conductors. Operators can improve efficiency and reduce power loss by increasing voltage with boosters, using power equipment closer to the load and decreasing the power supply distance.

PHASE 4: ENERGY CONSUMPTION

This phase represents the ‘lowest hanging fruit’ for efficiencies. Operators transform DC energy into radio waves and receive and process incoming signals. This category can be further broken down into areas such as architecture optimizations, signaling, network shutdowns, cooling and beamforming that wastes less energy than omnidirectional transmission.

SITE, RAN AND NETWORKWIDE INNOVATIONS

Alongside technical improvements to reduce energy leakage as power passes through the network phases, solutions to improve efficiency holistically across the network are available. Our analysis focuses on sites, the radio access level and broader network planning considerations. We exclude efforts targeted at handsets and other end-user terminals as these do not directly contribute to a mobile operator’s carbon emissions profile. Figure 3 summarizes the holistic network-wide approach.

Figure 3

Figure 3 Energy-saving methods differ between site level, RAN equipment and network planning. Source: GSMA Intelligence.

SITE INNOVATIONS - BATTERY SOLUTIONS

In off-grid areas, mobile operators are often forced to use diesel generators to guarantee the reliability of power supply for base stations. This is less than ideal considering generators emit high levels of carbon dioxide and have onerous cost implications associated with refueling, particularly if in hard-to-reach, sparsely populated areas (such as in African countries) requiring labor callouts and security protection.



Lithium batteries have emerged as a more environmentally friendly and cost-efficient alternative. These have a smaller and lighter form factor compared to traditional lead acid batteries, saving space after installation. Lithium batteries have a significantly longer expected life-span (five or six years on average). The commonly used lead acid batteries are expected to be efficient for a much shorter period - around three years.

Further favorable aspects of lithium batteries are the improved charge and discharge capacities and related savings potential from the battery configuration. Backup batteries are fully charged at all times and discharge only when there is a power outage. By using a cycle-type lithium battery capable of daily charge and discharge, smart power control with a DC power controller can be performed, enabling flexible and efficient power supply to radio equipment. Voltage boosting is also an option with lithium technology; this can help operators increase voltage, save on energy transportation and serve newly installed 5G AAUs from longer distances more efficiently. In the event of theft, the battery is designed to automatically stop any output of power, rendering it useless to criminals. The batteries are also fitted with GPS modules, making them easily traceable.

HYBRID AND RENEWABLE ENERGY

Solar has become a competitive alternative to diesel in off-grid areas as the price of photovoltaic panels has fallen and base station battery solutions have become more advanced. Site-level hybrid energy solutions involve a mix of solar/diesel/wind/electricity/hydrogen-rich fuel and grid, providing a more efficient way to power sites. Custom-made hybrid solutions have a dedicated variable speed motor and a DC alternator which reduces loss caused by energy conversion. Advanced algorithms can select optimal energy sources and achieve significant energy efficiency improvements for the estimated 1 million cell sites globally.

More broadly, the shift to purchasing renewables will accelerate as prices continue to decline and future contracts enable long-term lock-ins. Operators can buy more of their energy from larger, centralized renewable energy sources and achieve long-term power purchase agreements or use their assets and produce their renewables in their cell sites.

RAN AND NETWORK EQUIPMENT INNOVATIONS

Turning off equipment even for a short period of time or putting it into a sleep mode when there is no traffic to serve, saves energy. With 2G, 3G and 4G, there are recurring transmissions of always-on signals called cell-specific reference signals to secure cell coverage and a connection with users. Significant energy-saving is possible from decreasing resources allocated to signaling and its ‘ping-pong’ effect between user equipment and the cell site.

The 5G NR standard allows more components to switch off or go to sleep when the base station is in idle mode and requires far fewer transmissions of always-on signaling transmissions. Overall, these factors allow deeper sleep periods for a longer time, which - everything being equal - confers a significant saving on network energy consumption per bit of data.

Massive MIMO requires an increased number of antennas compared with traditional MIMO technology. Laboratory tests suggest that the increased number of antennas improves energy efficiency, transmitting and receiving more data for a given amount of energy.

NETWORK-WIDE PLANNING AND OPTIMIZATION - SUNSETTING LEGACY NETWORKS

A key challenge is to square the improvement in energy efficiency per bit of data in 5G networks with the inevitability of rising traffic and the risk that overall power consumption could still increase. In this sense, strategies to reduce energy emissions have to be considered at an overall network planning level, incorporating all generations of mobility and their associated spectral elements.

Sunsetting legacy 2G and 3G networks is a major means of emissions reduction. The energy per bit of data with each new mobile generation is constantly improving, so sunsetting 2G and 3G networks can boost overall network energy efficiency. As legacy mobile technologies approach the end of their lifecycles, the importance of decommissioning and refarming certain spectrum bands to LTE or 5G is growing. 5G is particularly attractive as it is more efficient than legacy generations, given the NR standard.

Although the exact difference in energy efficiency between 5G and previous technologies varies, laboratory tests suggest 5G has a significant efficiency advantage. It can also save energy and space through using fewer active antenna units and other networking elements. The process of sunsetting is already in progress and will likely continue in a staggered manner over several years to balance the risk of stranded network assets if take-up of LTE and 5G tariffs lags expectations.

AI-DRIVEN NETWORK PLANNING

AI-driven network management and planning applications are not a particularly new concept, but many vendors and network operators have recently launched energy management solutions that leverage AI and advanced data analytics to optimize energy consumption. AI can help operators increase energy efficiency and deal with the 5G era’s increased data traffic in terms of network planning and optimization.

AI can also help in network planning by gaining insights from coverage areas, building heatmaps for network usage and recommending an optimal location for new cell sites. New algorithms could also help understand spatial and temporal patterns in the ever-changing nature of mobile data use and predict future usage profiles in different coverage areas. AI algorithms can support the interplay between indoor and outdoor small cells, Wi-Fi hotspots and macro sites to maximize energy efficiency.

AI-DRIVEN NETWORK OPTIMIZATION

Equipment vendors have started to offer AI-driven energy-saving solutions as an extension to existing network management platforms. Algorithms for power-saving in base stations can already be used to shut down power amplifiers, transceivers and other network elements. However, AI can improve efficiency and lengthen sleep periods.

Base stations are the ‘low-hanging fruit’ for such applications as they consume more than 70 percent of total energy. As each is unique, optimizing their operation one-by-one would be labor-intensive. AI was introduced to enable more precise energy-saving based on traffic and other site-related conditions, improving efficiency and reducing the manpower required. Large-scale deployments have shown an increase in power-saving activation of more than 80 percent.

Many device companies are now using machine learning and AI to optimize functions such as antenna tuning and power amplifier biasing to improve efficiency of the transmit and receive chains in the radio. This is a future trend to improving efficiency at the device level.

In the 5G era, the energy optimization offered by the first and second generations of algorithms is not sufficient to deliver the needed energy-saving to keep up with growing data traffic. The third generation of AI-driven energy-saving solutions can take account of the different efficiency levels of frequency bands and factors in that the power efficiency of different networks can vary. The new AI can help base stations direct services to the optimal network, resulting in greater network energy efficiency.

Major vendors are currently offering solutions that can make energy savings of 5 to 15 percent on the RAN. New software can forecast data traffic based on historical patterns, weather, events nearby and other factors, before identifying the necessary thresholds, activation and sleep periods. Based on the information, the algorithm can shut down power amplifiers, transceivers and other network elements to save energy. Alongside this power-saving potential, AI-driven shutdown solutions can constantly monitor customer experience, network availability and data traffic to ensure there is no impact on network performance.

AI can also reduce energy consumption outside the RAN - in central offices, shops and data centers - by continuously calibrating the optimal settings of heating and cooling systems, pumps and fans. Engineers can use AI-driven building management systems to prioritize work, reduce unproductive travel time, identify equipment issues, avoid costly unscheduled callouts and help ensure network reliability. Going forward, AI-driven energy-saving platforms are expected to focus more on data harvested from user devices. Anonymized coverage and data traffic insights from devices can help optimize the network further and adjust more capacity layers.

Overall, AI-driven network shutdown solutions can be broken down into three areas:

  • Module (transceiver, baseband processing, etc.): The AAU components can be shut down in real-time during idle periods
  • Equipment (AAU, RRU): Equipment can be completely shut down during periods of low traffic, usually at night
  • Network: Large-scale, AI-driven solutions can schedule data traffic between different 5G bands (for example, from C-Band to sub-3 GHz bands) or between 5G and 4G, in a similar way to the smart data mode seen in new smartphones.

CONTENT CACHING NEARER TO THE END-USER

A surge in the popularity of video streaming over the last five years has made placing content caching facilities closer to end users strategically important - to maintain quality and for competitive reasons. Most video content passes through content delivery networks (CDNs), which transfer media across hundreds of servers worldwide. CDNs can reduce power demand as a video stream only has to travel through the network once to reach thousands of customers.

The CDN market historically mostly comprised independent groups such as Level 3 and Akamai, but major internet and consumer tech companies (Google, Facebook, Apple, Amazon, Netflix) have established their own servers to ensure control over their own content.

Reducing the distance between cache points and users results in improved latency, which preserves the customer experience for high- and super-high-definition video. As fixed-wireless access over 5G gains traction as a last-mile alternative for home broadband in some markets, the requirement for caching nearer to end-user premises would become even more pressing. CDN analytics platforms and network management systems can together capture, locate and analyze trends and events across the RF, RAN, backhaul and core, providing operators with unprecedented insight to optimize their network, save energy and monitor the customer experience.

THE WAY FORWARD

Alongside technical improvements to reduce energy leakage as power passes through the network phases, a range of measures are available to improve efficiency holistically across the network. These include the following:

  • User equipment and devices - energy consumption and extended battery life of end-user terminals, mostly handsets
  • Site-level innovations - new lithium-ion battery solutions, rectifiers, liquid cooling, air-con systems and simplification of site set-up
  • RAN and network equipment innovations - AI-driven software focused on maximizing sleep states to avoid unnecessary energy consumption in the RAN
  • Network planning and optimization - including the sunsetting of legacy 2G and 3G networks and long-term purchasing contracts for renewable energy.

The big picture for operators of ultimately reducing emissions to net zero depends on wrapping energy efficient technologies into a broader ‘green’ strategy that encompasses all facets of operations. To put teeth behind public commitments, many large operators have implemented key performance indicators and reporting targets in line with the independent Science Based Targets initiative.

Emissions reduction goals have been set in a phased approach to first reach carbon-neutral status before the more difficult and ambitious objective of net zero. Our analysis indicates that progress has generally been solid so far, enabled by advances in the renewable energy markets.

Despite this progress, reporting targets are not yet in place in most operators. There are also several persistent barriers, including emissions data availability and tracking mechanisms, lack of partnerships with energy sector producers and, in some cases, outdated organizational structures that augur for more cross-team working and less hierarchy.

The data aspect is of particular importance; we hope this research will help raise awareness of the issue. The construction of comprehensive data ‘pipelines’ with associated analytics would help uncover costly anomalies. Deploying smart sensors at various points of the network would help measure equipment-level energy consumption, battery status, active hours of generators, fuel levels, outside and indoor temperatures and air conditioning. Operators would need to build their comprehensive and real-time data repository, but we believe this would be money well spent. With reliable measurements and data pipelines established, big data applications can monitor and adjust network power - a key ability for the software-defined networks set to be the default option in the 5G era.

For the full report, 5G Energy Efficiencies: Green is the New Black, visit:

www.mwjournal.com/gsma