INFSO-ICT EARTH. Deliverable D2.1. Economic and Ecological Impact of ICT

Transcription

1 INFSO-ICT EARTH Deliverable D2.1 Economic and Ecological Impact of ICT Date of Delivery of 1 st Revision: April 30, 2011 Editor(s): Gergely Biczók (BME), Albrecht Fehske (TUD) Author(s): Gergely Biczók, Jens Malmodin, Albrecht Fehske Participant(s): BME, EAB, TUD Work package: WP2 Estimated person months: 8 Security: Public Version: 2.0 Keyword list: Mobile communications, carbon footprint, ecological impact, economic impact. Abstract: It is commonly known that mobile communication networks will have an increasing ecological and economic impact worldwide, and recently, initiatives to reduce the energy consumption of network operation and reduce the carbon footprint of these networks have gained momentum. However, the overall carbon footprint of mobile communications has so far only been roughly estimated. This deliverable presents a study quantifying the global carbon footprint of mobile communications until 2020 and discusses the contribution of different parts of the networks lifecycle, such as manufacturing and operation of mobile devices, and manufacturing and operation of the radio access network. In the second part, the electricity consumption of worldwide radio access networks for different scenarios representing different levels of technological advances is estimated, showing that the targeted 50% reduction of the EARTH project can become essential for the sustainability of wireless communications. Implications of the electricity consumption on revenue models of operators are also discussed. Disclaimer: This document reflects the contribution of the participants of the research project EARTH. The European Union and its agencies are not liable or otherwise responsible for the contents of this document; its content reflects the view of its authors only. This document is provided without any warranty and does not constitute any commitment by any participant as to its content, and specifically excludes any warranty of correctness or fitness for a particular purpose. The user will use this document at the user's sole risk. EARTH WP2 D2.1: Economic and Ecological Impact of ICT 1 / 25

2 Authors Partner Name and Phone Budapesti Műszaki és Gazdaságtudományi Egyetem (BME) Gergely Biczók Ericsson AB (EAB) Jens Malmodin jens.malmodin@ericsson.com Technische Universität Dresden (TUD) Albrecht Fehske albrecht.fehske@ifn.et.tu-dresden.de EARTH WP2 D2.1: Economic and Ecological Impact of ICT 2 / 25

3 Table of Contents 1. Introduction Mobile Communications Ecological Impact Mobile Subscriptions, Traffic Demand and Network Infrastructure Carbon Footprint Modelling Global Carbon Footprint Forecast potential impact of the EARTH project on emissions due to ran operation Scenarios Of Global RAN Development Breakdown of RAN Operation Footprint Potential Impact of the EARTH Project on RAN Operation Footprint Machine-to-Machine Communication An Outlook Mobile Communications Economic Impact Energy cost and Revenue of Mobile Services in mature markets Energy Prices, Off-grid Sites and Alternative Energy Sources RAN Energy Consumption Summary and Discussion EARTH WP2 D2.1: Economic and Ecological Impact of ICT 3 / 25

4 List of Figures FIGURE 1. EXPECTED GROWTH IN AVAILABLE WIRELESS DATA RATES (WITH TECHNOLOGICAL MILESTONES) [5]... 6 FIGURE 2. THE NUMBER OF GLOBAL MOBILE SUBSCRIPTIONS PROJECTED UNTIL FIGURE 3. RAN SITE ELECTRICITY CONSUMPTION FORECAST FIGURE 4. GLOBAL CARBON FOOTPRINT OF MOBILE COMMUNICATIONS PROJECTED UNTIL 2020 (CI WITH A CONTINUATION OF CURRENT TRENDS AND A HIGH DATA SCENARIO) FIGURE 5. GLOBAL CARBON FOOTPRINT PER SUBSCRIBER AND PER GB OF MOBILE DATA TRAFFIC PROJECTED UNTIL FIGURE 6. GLOBAL CARBON FOOTPRINT DUE TO RAN OPERATION FOR THE PERIOD : BREAKDOWN INTO AREAS OF IMPORTANCE FIGURE 7. POTENTIAL SAVINGS CO 2 E IN GLOBAL RAN OPERATION FOR SCENARIO 3 AND SCENARIO 4 (BASED ON FIGURE 6) FIGURE 8. POTENTIAL SAVINGS CO 2 E IN GLOBAL RAN OPERATION FOR SCENARIO 5 (BASED ON FIGURE 6) FIGURE 9. AVERAGE REVENUE OF MOBILE DATA TRAFFIC ACCORDING TO [10] FIGURE 10. EXPECTED EVOLUTION OF ENERGY PRICES [17] FIGURE 11. GLOBAL RAN ELECTRICITY CONSUMPTION PROJECTED UNTIL 2020 FOR DIFFERENT SCENARIOS List of Tables TABLE 1. STATISTICAL DATA AND PROJECTIONS OF GLOBAL MOBILE NETWORK CHARACTERISTICS FOR 2007, 2014 AND TABLE 2. CARBON FOOTPRINT - COMPARISON TO SMART 2020 [6] EARTH WP2 D2.1: Economic and Ecological Impact of ICT 4 / 25

5 Acronyms and Abbreviations EARTH BS BSC BRIC CAGR CI CO2 CO2e CPU DVD EIA GGSN GSM HLR ICT IPCC ITRS LCA MIMO MSC M2M OPEX PC RNC SGSN WCDMA Energy Aware Radio and network technologies Base Station Base Station Controller Brazil, Russia, India and China Compound Annual Growth Rate Continuous Improvement Carbon-dioxide Carbon-dioxide equivalent Central Processing Unit Digital Versatile Disc Energy Information Administration Gateway GPRS Support Node Global System for Mobile Communications Home Location Registry Information and Communication Technology Intergovernmental Panel on Climate Change International Technology Roadmap for Semiconductors Life Cycle Analysis Multiple-Input Multiple-Output Mobile Switching Center Machine-to-Machine OPerational EXpenditure Personal Computer Radio Network Controller Serving GPRS Support Node Wideband Code Division Multiple Access EARTH WP2 D2.1: Economic and Ecological Impact of ICT 5 / 25

6 1. INTRODUCTION Rarely have technical innovations changed everyday life as fast and profoundly as the massive use of personal mobile communications. Over the past two decades mobile wireless services grew from niche market applications to globally available components of daily life: the first GSM phone call took place 1991 in Finland, and only 15 years later there were over two billion GSM users. According to ITU, the total number of mobile subscriptions in the world will have passed 5 billion by the end of 2010, more than 70% of the population of the planet. By comparison, there are only about 1.2 billion fixed line subscribers worldwide. Also, now the number of people accessing the Internet amounts to only 2 billion worldwide which is slightly more than one fourth of the global population [1]. The driving force behind this rapid development was the growing importance of connectivity for social and economic interactions. In addition, processing power and storage capacities of mobile devices have doubled approximately every 18 months according to Moore s Law, or rather the International Technology Roadmap for Semiconductors (ITRS). This considerable growth rate in turn renders the use of ever more powerful information and communication systems and devices attractive for the mass market. In order to be able to transport an exponentially rising amount of available data to the user in an acceptable time, data transmission rates in both the wired internet and wireless networks including cellular, local area and personal area networks have been rising at approximately the same speed by about a factor of ten every five years, as illustrated in Figure 1 for wireless systems. The same trend is assumed for the near future as well. FIGURE 1. Expected growth in available wireless data rates (with technological milestones) [5] The ever-increasing demand for wireless services and ubiquitous network access, however, comes at the price of a growing carbon footprint for the mobile communications industry. The whole Information and Communication Technology (ICT) sector has been estimated to represent about 2% of global CO 2 emissions and about 1.3% of global CO 2 equivalent (CO 2 e) emissions in 2007 [1], [19]. This study estimates the corresponding figure for mobile networks to be 0.2% and 0.4% of the global CO 2 e emissions in 2007 and 2020, respectively. It is important to note that these figures include emissions from the whole life cycle, not only emissions related to the operation of mobile networks. In terms of absolute figures, the Smart 2020 Report EARTH WP2 D2.1: Economic and Ecological Impact of ICT 6 / 25

7 has shown that the overall ICT footprint will less than double between 2007 and 2020 [6], with the footprint of mobile communications more than doubling between 2007 and This document shows an even steeper increase, it predicts that the footprint of mobile communications could almost triple from 2007 to 2020 corresponding to more than one-third of present annual emissions of the whole UK. In addition to minimizing the overall footprint of mobile communications, there is a strong economic drive to reduce the energy consumption of these networks. Mobile data traffic is going to increase dramatically over the next five years, essentially due to a dramatic increase in mobile video traffic, e.g., live streaming or YouTube [10]. When considering a rising overall energy consumption of the networks driven by the computational complexity of advanced transmission techniques and increasing number of base station sites required for high data rates in demand on the one hand, and steadily increasing energy prices on the other, it becomes obvious that energy consumption is critical for the operational costs (OPEX) of mobile telecom providers. In this report we present valuable insights into Life Cycle Analysis (LCA) and carbon footprint modelling of mobile communication networks alongside with absolute figures of the footprint and electricity consumption of mobile devices as well as network equipment. Furthermore, we quantify the global carbon footprint of mobile communication networks until 2020 and discuss trends in contributions due to RAN operation, mobile devices manufacturing and data transport. The report also provides a discussion on the impact of RAN energy consumption on operator revenue models. A prediction of the global RAN energy consumption depending on the energy efficiency of base station sites is presented, showing that the targeted innovation of the EARTH project is essential for the sustainability of wireless communications. The results of this study are calculated using most reasonable assumptions on models and parameters available today, which in certain cases might be subject to changes in the future. 2. MOBILE COMMUNICATIONS ECOLOGICAL IMPACT Assessing and quantifying the future ecological impact of mobile communications itself requires studying a series of factors related to production, operation, and distribution of mobile communication networks and services. From an ecological perspective, a key factor of interest is the overall carbon footprint measured in carbon dioxide equivalent, CO 2 e. Generally, a carbon footprint is defined as the amount of CO 2 e emissions calculated with the Global Warming Potential (GWP-100) indicator defined by the International Panel on Climate Change (IPCC) [7], [8]. In case of mobile communications, a large part of the carbon footprint stems from electricity consumption during operation and manufacturing, i.e., from the production and distribution of electricity, the extraction, production and distribution of consumed fuels, the construction and operation of the power plants and the grid, and finally, all related waste treatment. In 2007, global direct CO 2 emissions from electricity production/consumption amounted to 0.54 kg/kwh, while total life cycle CO 2 e was 0.62 kg/kwh. In addition, trends of characteristic figures affecting the carbon footprint such as the number of global mobile customers, the size of network infrastructure, or the mobile traffic volume are interesting to study in their own right. Estimates of those figures up to 2014 as presented in the following are based on projections from analysts Gartner and ABI Research [1], [3]. Estimates of characteristic figures presented for the period 2015 to 2020 are based on modelling and extrapolation of current trends and data, conducted within task T2.1 of the EARTH project. For the convenience of the reader, statistical data for 2007 and projections for the years 2014 and 2020 for parameters related to the ecological impact of mobile networks are summarized in TABLE 1. In addition, to the ecological impact from the life cycle of mobile telecommunication networks and terminals there is also an increasing awareness of the potential of ICT industry in general and mobile communication in particular to reduce CO 2 emission in other sectors primarily through reduction in e.g. transport and travels. This aspiration has e.g. been manifested in the resent Guadalajara Declaration [22] signed by more then 40 ICT companies, which was presented by Ericsson during the UN Climate Change Conference COP 16 in Mexico. EARTH WP2 D2.1: Economic and Ecological Impact of ICT 7 / 25

8 This will further increase the pressure on the industry gain knowledge in this area and be leading by example, by keeping tight control on its own CO 2 impact to appear as a credible partner. TABLE 1. Statistical data and projections of global mobile network characteristics for 2007, 2014 and Global values Total subscriptions (millions) 2950 A 5600 A G+ data subscriptions (millions) 30 C 700 C 1600 Total mobile phones sold (millions/yr.) 1150 A 1400 A 1700 Share of smartphones in total sales / in use (%) 9 A / 7 29 A / / 33 Laptops and netbooks sold (millions/yr.) 23 A 280 A 510 Worldwide base station sites * (millions) 3.3 B Average power consumption per site (kw) Total global RAN power consumption (TWh) Share of diesel generated power in total RAN power (%) Mobile data (including voice as data, million TB/yr) 0.8 C 45 C Share of mobile data in total mobile traffic (%) 37.5 C 98 C 99.6 Share of mobile data traffic in all IP traffic (%) 0.9 C 5 C Mobile data per avg. subscription (GB/sub./yr) Mobile data traffic per 3G+ data subscription (GB/sub./yr) 5 C 27 C Average data traffic per base station (Mbit/s) A: Gartner Research, B: ABI Research, C: Cisco Visual Networking Index, *: per site and standard, actual site count is less but sites require more power due to co-location 2.1. MOBILE SUBSCRIPTIONS, TRAFFIC DEMAND AND NETWORK INFRASTRUCTURE A fundamental driver of the overall carbon footprint of mobile communications is the number of worldwide mobile subscriptions. According to statistical surveys, the number of mobile subscribers worldwide increased with a compound annual growth rate (CAGR) of 24%, from around 500 million to over 4 billion between 2000 and 2009 [4]. As depicted at the end of the section in FIGURE 2, there are expected to be about 1.4 billion and over 2 billion regular 3G+ (broadband for phones/smartphones) subscriptions in 2012 and 2014, respectively. The number of GSM subscriptions decreases rapidly after A peak GSM (in terms of number of subscriptions) is expected to occur around Also, a steady increase for 3G+ data subscriptions (USB dongles for laptops/pcs) can be observed. The total number of subscriptions is expected to grow to about 6 billion in 2014 and tends to reach global penetration of 100% after Penetration of 100% means that the number of subscriptions matches the number of inhabitants. If only the relevant age group between 10 and 65 is taken into account, a quasi-100% penetration can be expected already during Note that penetration EARTH WP2 D2.1: Economic and Ecological Impact of ICT 8 / 25

9 rates over 100% have already been reached in western countries like Italy (165%), the UK (122%), Spain (119%) or Germany (118%) by 2010 [9]. The other fundamental driver behind the footprint of mobile communications is the traffic volume. With the proliferation of smartphones and cellular data subscriptions for laptops, mobile data traffic shows a steep growing pattern. The most influential type of application is video streaming: it generates high data volumes, and user experience is already quite good (and is improving continuously) with advanced user devices and advanced networking technology. Note how mobile data will dominate all mobile traffic (the share of voice is decreasing); also, mobile data is projected to account for 11-17% of all IP traffic in As for per user statistics, an average mobile subscription (across all technologies) is expected to correspond to 8 GB of traffic per year in 2014, and it will reach 65 to 100 GB in Cellular data subscriptions (USB dongles) will carry even more traffic per user, due to more capable terminals (laptops). ABI research reported in 2008 that about 3.3 million base station (BS) sites were in operation in 2007 counting every standard [3]. The number of actual physical sites is of course lower as more than one standard and more than one operator can be located at the same site. In many cases 3G sites have been built at already existing 2G sites. It is estimated that there are about 4.6 million sites, counting all standards in 2009, located at 3.6 million actual physical sites. The total number of delivered base station cabinets is more than 5.5 million [3]. FIGURE 2. The number of global mobile subscriptions projected until CARBON FOOTPRINT MODELLING In order to quantify the future carbon footprint of mobile communications we consider all generations of cellular mobile networks including all end-user equipment accessing the networks, all business activities of the operators running the networks, and the use of fixed network resources as a result of the data traffic from mobile network users. The overall carbon footprint model of mobile communications can conveniently be broken down into six categories as follows: Manufacturing of mobile devices refers to the manufacturing of mobile terminals also including regular mobile phones, smartphones and laptops, based on actual sales the same year. Mobile devices operation refers to charging of batteries and stand-by consumption of chargers left plugged in for all mobile phones and smartphones. Here, charging and grid operation for laptops but no docking stations, extra monitors or other peripherals are included. EARTH WP2 D2.1: Economic and Ecological Impact of ICT 9 / 25

10 RAN sites manufacturing and construction means manufacturing of all electronic equipment as well as site equipment like diesel generators and batteries, and construction of site infrastructure such as antenna towers and site housings. RAN sites operation refers to the total electricity consumption of base station sites, control sites (BSC/RNC) and core sites (MSC, HLR, SGSN, GGSN etc.). A total site view is used that also includes transmission, cooling, rectifiers, backup power, etc. Diesel consumption for off-grid site operation and as backup power is also included. The RAN operation model is illustrated in FIGURE 3. Electricity consumption figures due to RAN operation for resulted from a large-scale investigation of installed base stations, while further numbers are extrapolated based on the investigation. Current trends, including more 3G than GSM base stations, more main-remote and micro base stations, lower power consumption and better power management are reflected by the base station model. Moreover, more efficient cooling techniques and power systems are also taken into consideration. The solid orange line represents the electricity consumption of newly deployed sites with an 8% annual improvement, while the solid red line depicts the electricity consumption of an average site (taking into account the mix of older and newly deployed sites) with the same annual improvement for new sites. Dashed lines assume that the consumption of newly deployed sites is frozen at the 2010 level (worst case). Gradual modernization of sites is also accounted for (green line). A gradual modernization of off-grid sites run on diesel-generated electricity is also modelled. It is predicted that all diesel generators in 2020 will work in a so-called hybrid mode with batteries. This is not visible in FIGURE 3, which only shows the electricity consumption of the base station. Operator activities cover operation of offices, stores, vehicle fleet and business travel related to all operators business activities. Data centers and data transport refers to the use of or allocation of other network resources based on the data traffic generated by mobile network users. 16 MWh per year and base station site Average installed site eq. The 1.5 kw site New site eq. The 1 kw site No annual improvements 8% annual improvements Old site eq. taken out of service (per new site eq.) No annual improvements FIGURE 3. RAN site electricity consumption forecast In order to estimate the footprint of manufacturing and operation of mobiles, we consider three categories of devices accessing the network: regular mobile phones, smartphones, and laptops. The carbon footprint model of regular mobile phones is based on Ericsson s cradle to grave LCA studies of mobile phone manufacturing EARTH WP2 D2.1: Economic and Ecological Impact of ICT 10 / 25

11 (including transportation to the customers, waste treatment and recycling) resulting in an average of 18 kg CO 2 e per device [18]. The operation is estimated to 2 kwh/year based on charging every 60 th hour equal to 40% of battery capacity every day and a stand-by scenario 50% of the remaining time. It must be noted that modern mobile phone chargers have low stand-by power consumption in the order of 0.1 W. According to recent trends, we assume that both the manufacturing and operation emissions remain constant, i.e., we assume that technological improvements at the component level are used for provisioning of phones with better performance and more functions. Corresponding values for smartphones are 30 kg CO 2 e for manufacturing and 7 kwh for operation. The same assumptions and principles used for regular mobile phones also apply to smartphones. Based on a comprehensive review of LCA studies of PCs, the manufacturing related emissions of laptops are estimated to 240 kg CO 2 e per average device [19]. The electricity consumption of the average annual use is estimated to be 40 kwh in Based on current trends, it is projected that manufacturing and operation emissions are reduced by five percent per year. An important aspect of the model for laptops is that 50% of the life cycle carbon footprint is allocated to wireless networks, as they are also connected to other networks at home and at work. Exact figures of the carbon footprint of RAN site manufacturing and construction are based on Ericsson s LCA of network equipment [18]. Figures on emissions and energy consumption due to RAN site operation, operator activities, data center operation, and data transport are based on a broad operator investigation covering networks that service about 40% of global subscribers. The annual electricity consumption per average subscription is about 17 kwh. The construction of new base station sites every year as well as the removal of old site equipment is taken into account throughout the studied period. Surveying existing network equipment reveals that energy consumption of new base station sites is reduced by about 8% on average compared to equipment installed the year before due to technological advances; we refer to this annual reduction as the Continuous Improvement (CI) scenario later on. Under this assumption (also taking into account the installation of new and removal of old equipment) the average base station site power amounts to about 1.7 kw in 2007 and is reduced to about 1.2 kw in According to analyst company Berg Insight, there could be about 12 million femtocells with about 70 million users "on a regular basis" in 2014 [14]. Extrapolating from this, there could be about 100 million femtocells with about 500 million users in The power consumption of a femtocell today is around 10 W, will be around 6 W in 2012, and it can be assumed that a femtocell in 2020 will still consume 5 W, if also the power consumption for the fixed line connection is included. The 100 million femtocells in 2020 will then consume 4.4 TWh in 2020, i.e., about an extra 5% on top of the EARTH CI scenario for the global RAN network (please refer to Section 3.3). Given the high uncertainty, the impact of femtocells is not included in the energy consumption model or on RAN data traffic. While femtocells add a predictably small direct amount of energy to RAN consumption, their positive impact on data traffic and capacity, resulting in an energy consumption decrease could be indeed larger GLOBAL CARBON FOOTPRINT FORECAST The estimated global carbon footprint of mobile communications as a whole, including end-user equipment, operator s business activities and data traffic, for the period is depicted in FIGURE 4. Initial variation in the data is explained by the use of current production statistics between 2007 and Four major trends can be observed. According to the projection, the overall carbon footprint of mobile communications increases almost linearly until 2020 with an annual increase of 11 Mton CO 2 e, an increase equivalent to the annual emissions of the whole country of Luxembourg or 2.5 million EU households. The emissions in 2020 amount to more than 235 Mton CO 2 e, which corresponds to more than one-third of present annual emissions of the whole UK. (Globally speaking, mobile communications will account for 0.6% of global direct CO 2 emissions and 0.4% of global CO 2 e.) Relative to 2007, the overall carbon footprint of mobile communications will increase by a factor of 2 until 2014 and 2.7 until In case of only EARTH WP2 D2.1: Economic and Ecological Impact of ICT 11 / 25

12 minor efficiency improvements in base station sites and end-user equipment, the footprint would even increase by a factor of more than three. In contrast, the footprint of the ICT sector as a whole is expected to increase by a factor of only 1.72 during the same 13-year period [6]. FIGURE 4. Global carbon footprint of mobile communications projected until 2020 (CI with a continuation of current trends and a high data scenario) Over the whole studied period, the overall carbon footprint remains dominated by the individual footprints of RAN operation and production of mobile devices. Accordingly, initiatives aiming at carbon footprint reduction of the mobile communications sector should focus on these two areas. While RAN operation is by far the largest contributor in 2007, mobile device manufacturing will become increasingly important until both have an equal share in the overall carbon footprint in The reason here is that smartphones and laptops represent a strongly increasing fraction of the devices accessing the network, a trend driven by the demand for advanced wireless services and applications, especially video. Compared to regular phones, smartphones have an almost two times, laptops a ten times higher carbon footprint. This effect is not sufficiently considered in [6], which claims a reduction in the fraction of CO 2 caused by mobile devices. The footprint of data centers and data transport will experience by far the strongest growth among all contributions until 2020, due to the drastically increasing volume of mobile data traffic in the coming years. The scenario considered here uses a high data traffic model with a CAGR of 60% between 2015 and Reducing the CAGR of mobile traffic volume to 50% during that period would lead to 33% less CO 2 e emissions from data centers and data transport in Dividing the global carbon footprint of mobile communications (FIGURE 4) by the number of subscriptions (FIGURE 2) yields the carbon footprint per mobile subscription. In case of an average subscription, the carbon footprint will increase just slightly from 28 kg CO 2 e in 2007 to about 31 kg CO 2 e in 2020 (the red curve in FIGURE 5). In contrast, the data traffic per average subscription will increase from 0.3 GB/year in 2007 to about 100 GB/year in 2020 (see TABLE 1). Putting these two trends together, the carbon footprint per GB of data traffic drops quickly from nearly 100 kg CO 2 e per GB in 2007 to 0.3 kg CO 2 e per GB in 2020, a decrease by about a factor of three hundred (the black curve in FIGURE 5). EARTH WP2 D2.1: Economic and Ecological Impact of ICT 12 / 25

13 FIGURE 5. Global carbon footprint per subscriber and per GB of mobile data traffic projected until 2020 Despite the strong connection of increasing data rates and higher energy consumption, scientific and technological progress managed to keep pace with the increase, a fact that is also observed for the last decade of 3G developments. In that period, possible data rates and actual data traffic increased by a factor of one hundred while the energy consumption of base stations dropped five times per provided channel capacity according to Ericsson. This trend could continue as the EARTH project keeps working on energy efficiency in mobile networks [20] POTENTIAL IMPACT OF THE EARTH PROJECT ON EMISSIONS DUE TO RAN OPERATION FIGURE 4 shows that RAN operation dominates the global carbon footprint of global mobile communications today, and it will still be responsible for nearly one-third of the carbon footprint in The EARTH project focuses on RAN operation and is expected to have major impact on the energy consumption (and hence the carbon footprint) of new base stations Scenarios Of Global RAN Development In order to assess the potential impact of technological advances between 2007 and 2020 we compare five scenarios that reflect roll-out and adoption of new technologies until We expect the predicted footprints for different scenarios to corner-case the real footprint as well as the energy consumption of RAN operation that will be observed until We define the scenarios as follows. 1. There are no significant technological advances reducing the energy consumption of base station sites after This scenario corresponds to the case where all improvements on component level contribute towards increasing spectral efficiency rather than reducing energy consumption as can be observed for mobile phones. 2. Newly installed BS equipment consumes 8% less electricity per year than sites installed in the previous year for the whole studied period. A reduction of 8% in electricity needed per year corresponds to the annual average that can be observed for currently deployed equipment. Note that a continuation of this trend does not come for free and stems from efforts to increase hardware EARTH WP2 D2.1: Economic and Ecological Impact of ICT 13 / 25

14 efficiency in close relation with from Moore s Law. We refer to the reduction of 8% per year as continuous improvements. 3. Scenario 3 assumes that newly deployed base stations consume 50% less energy each year than in Scenario 2. The 50% reduction in energy consumption per site must be seen as an average value due to the combined effects of improved hardware as well as better use of the equipment through improved radio resource management, smarter deployment, etc. This Scenario captures potential impact of EARTH results through deployment of new equipment. Since the improvements only affect newly deployed base stations, the saving effect is visible in the overall RAN only over time according to the roll-out model. 4. Since alternative energy has become a very important and increasingly funded area, a strong research push is foreseen in the coming years. Based on the new per site electricity consumption of Scenario 3, after 2012, all newly installed off-grid sites will be equipped with alternative energy modules (e.g. solar panels and batteries). In addition to the energy coming from a renewable source, diesel consumption with regard to the transportation of diesel to remote sites also decreases. There is also a large potential for so-called community power, when mobile batteries (not for mobile phones, but large batteries for base stations) can be charged in times of excess energy. This excess energy can possibly be as large as the energy required by the base stations themselves. 5. The last scenario is considered optimistic: Here we model the effect of EARTH results on the installed base of equipment. The savings are loosely based on the combined effects of large-scale swapping out, base station site sharing among mobile operators, site modernizations and advanced operation techniques such as sleep modes. A more detailed description is provided in the remainder of this section. Although this scenario is not backed up by detailed studies, the impact of the EARTH project on the energy consumption of new base stations described in Scenario 3 can catalyse swapping out and network sharing processes making this scenario more realistic Breakdown of RAN Operation Footprint The electricity generated by diesel, both for off-grid sites and as backup, only accounts for about 10% in 2007 but is responsible for about 30% of all CO 2 e emissions due to unoptimized operation and losses. The main reason for the slowdown of the growth of diesel related emissions seen in FIGURE 6 between 2012 and 2020, is that more and more off-grid sites will be upgraded to hybrid diesel/battery operation. Diesel consumption is predicted to still be responsible for nearly a quarter of all CO 2 e emissions in 2020, when all off-grid sites are modeled as hybrids. Another important aspect is that new base stations are more energy efficient compared to already installed ones, as clearly shown in FIGURE 3 and FIGURE 6. New base stations are also frequently installed on existing sites, and benefit from already installed power and cooling systems. At the same time, new base stations provide more capacity (this is not shown in the figures). The replacement process of base station equipment from old to new is modeled conservatively, starting in 2011 and growing moderately. This growth then slows down and reaches a level in 2020, which is derived from the average base station production 15 years earlier. Note, that only a relatively small ratio of all base station equipment ever produced has in fact been taken out of service and scraped. EARTH WP2 D2.1: Economic and Ecological Impact of ICT 14 / 25

15 No improvements in energy cons. Cont. improvements of 8%/yr Mtonnes CO 2 e New base stations (-8%/year) Installed base up to Installed base of all diesel sites will be modernized with hybrid operation to : 0.14% of global direct CO % of total global CO 2 e 2020: New diesel cons. 0.18% of global direct CO % of total global CO 2 e Installed base diesel consumption FIGURE 6. Global carbon footprint due to RAN operation for the period : Breakdown into areas of importance Potential Impact of the EARTH Project on RAN Operation Footprint A scenario where new base stations produced in the period of consume only 50% of the projected energy is shown in FIGURE 7. In addition, all new off-grid sites could be operated with renewable energy as a result of the low energy consumption of new base stations. The installed base of equipment installed already before 2013 will provide the largest challenge in terms of future energy consumption and its related carbon footprint. Note, however, that deployment of new base stations could also significantly increase the RAN footprint if historically observed improvements do not continue or slow down in the future (see dashed areas in FIGURE 7). To what extent can EARTH solutions also play a role for the installed base? A key here is the low energy consumption of future base stations that EARTH helps providing. The most important ways to reduce the global RAN carbon footprint of the installed base can be summarized as follows. Swapping out old base stations for new. This is fairly obvious. The exchange of old base station equipment for new could be done faster than what has been the case so far. Sharing base stations among operators. This can be done primarily for rural coverage, where the benefits are the largest, or for complete networks. Network sharing also makes swapping out more attractive. Site modernizations that improve rectifiers, cooling and other energy consumers (like lighting) and reduce energy consumption of base station sites. These modernizations also fit well with other site changes like swapping out and network sharing. EARTH WP2 D2.1: Economic and Ecological Impact of ICT 15 / 25

16 Network management techniques, designed for new base stations can possibly also be introduced in already installed base stations. For instance, there has been a number of sleep modes developed for GSM that have been able to run on older generations of base stations. However, the most advanced sleep modes developed for GSM require new or almost new hardware. FIGURE 8FIGURE 7 shows a scenario where the carbon footprint reduction of the installed base also reaches 50% in 2020, as a result of the possible actions described above Scenario 1: No improvements Scenario 2: Cont. improvements Mtonnes CO 2 e Scenario 3 CO 2 e saving potential Scenario 4 CO 2 e saving potential New base stations (-8%/year) Installed base up to Installed base of all diesel sites will be modernized with hybrid operation to New diesel cons. Installed base diesel consumption FIGURE 7. Potential savings CO 2 e in global RAN operation for Scenario 3 and Scenario 4 (based on FIGURE 6) Scenario 1: No improvements Scenario 2: Cont. improvements Mtonnes CO 2 e Scenario 5 CO 2 e saving potential New base stations (-8%/year) Installed base up to Installed base of all diesel sites will be modernized with hybrid operation to New diesel cons. Installed base diesel consumption FIGURE 8. Potential savings CO 2 e in global RAN operation for Scenario 5 (based on FIGURE 6). EARTH WP2 D2.1: Economic and Ecological Impact of ICT 16 / 25

17 2.5. MACHINE-TO-MACHINE COMMUNICATION AN OUTLOOK Cellular machine-to-machine (M2M) communication enables wireless data communication between machines via mobile networks. According to various industry sources (e.g., ABI Research [11] and Cisco [10]), the current number of global M2M subscriptions is around 70 million, and they are projected to triple to over 200 million in 2014 with a CAGR of 26%, a growth rate similar to that of mobile subscriptions in recent years. Vehicles and utility meters are expected to be the most numerous M2M devices. Despite this growth potential we expect M2M communication to have only a marginal impact on the carbon footprint of mobile communications. Due to low activity levels and typically low required capacity per link, M2M data traffic is expected to be negligible compared to traffic originating from human users and their video demands. Assuming a traffic of 50 bytes every 100 seconds per device and one thousand devices per person as envisioned by the World Wireless Research Forum (WWRF) [12], the M2M traffic per person is on the order of 0.5 kb per second negligible to the estimated 100 GB data traffic per person in A four-to-one ratio of M2M to human subscriptions is considered to be handled easily already by currently deployed networks. If in addition only the actual modem part is allocated to be part of the mobile network, which seems reasonable, the carbon footprint due to manufacturing and operation of M2M communication will be small, even for a vast number like 7 billion existing devices in 2020 as envisioned by the WWRF. An exception might be video cameras, e.g. for security applications, sending real time video. Here, the key question is to what extent mobile networks will carry such traffic, which is beyond the scope of this study. 3. MOBILE COMMUNICATIONS ECONOMIC IMPACT Besides the ecological impact (carbon footprint) of the ever-evolving mobile communication systems, there is also a significant economic impact that mobile operators have to face. This impact is composed of three ingredients: the changing revenue model of mobile services, energy prices and alternative energy, and, most importantly, the growing electricity consumption of the mobile RAN ENERGY COST AND REVENUE OF MOBILE SERVICES IN MATURE MARKETS The average revenue per user (ARPU) earned by operators ranges from under ten dollars to several tens of dollars per month in developing and developed markets, respectively. Overall, ARPU is globally decreasing. For instance Vodafone Germany reports an ARPU shrinking annually by over 6% on average from around 30 Euros 2000 to around 16 Euros in 2009 [16]. The hundred-fold increase in data traffic per subscriber (TABLE 1) as projected in this and other studies, unfortunately, cannot be expected to translate into corresponding revenue per user, since the revenue per unit traffic varies widely between service types. While classic mobile messaging generates revenue on the order of a several tens of dollars per MB, web browsing related traffic generates average revenue in the range of only several dollars per MB. Moreover, the revenue per MB of streaming video services is about two orders of magnitude less, at around one cent per MB and the proliferation of flat rate data subscriptions will ultimately bound potential revenue gains of rocketing mobile data volumes (see FIGURE 9). Similar trends can be observed for classic voice traffic, where operators report that revenue per call has dropped in the order of 5% annually over the last few years and combined flat rates for fixed and mobile connections are becoming standard. Considering these trends it is clear that the revenue models of mobile operators are becoming more and more connected to the direct operational costs. In this regard, the energy bill for network operation constitutes a significant factor. Assuming an average of 1.7 kw power need, 800 subscribers per site with an ARPU of 20 dollars per month, and energy costs of 15 cents per kwh, common for mature western markets, the annual energy bill can be straightforwardly calculated to be around the staggering amount of 1% of the overall network earnings before interest and tax (EBIT). An increasing number of subscriptions cannot be expected to change the situation in fully penetrated markets. Annual customer surveys have shown that the share of mobile communications in annual expenditures per physical customer is constant, and thus the attainable revenue per person is flat. The ARPU thus tends to decrease according to the increase in subscriptions once EARTH WP2 D2.1: Economic and Ecological Impact of ICT 17 / 25

18 certain market maturity has been reached as then the fixed revenue per person divides into multiple subscriptions. Taking into account, that a quasi full mobile penetration can be expected globally during 2012 (FIGURE 2), it becomes obvious that overall network operators revenues can only stay flat in the future. An increase in network energy consumption by a factor of 2.5 until 2020 without any technological improvements, predicted by this study (FIGURE 4), translates directly into a cost increase, which cannot be compensated with earnings staying flat. Considering a CI scenario with decreasing electricity consumption of annually 8% for newly installed sites, the energy cost in 2020 already doubles; an additional increase of energy prices by 50% would triple them. Moreover, current trends of network densification further foster that development, and relying on energy savings as accomplished in a CI case is no longer sustainable; dedicated research efforts are clearly needed Cent / MB Cent / MB Cent / MB 10 Cent / MB 12 1 Cent / MB 1 Video Streaming Video download Web/other Text messaging FIGURE 9. Average revenue of mobile data traffic according to [10] 3.2. ENERGY PRICES, OFF-GRID SITES AND ALTERNATIVE ENERGY SOURCES As a large portion of the OPEX of mobile operators goes to paying for electricity, energy prices are of high importance for them. We show expected energy prices in FIGURE 10 using the international energy price forecast from the Energy Information Administration (EIA) of the US Department of Energy [17]. Although energy prices are hard to predict accurately, a clear linearly increasing trend could be observed, indicating a strong possible motivation for mobile providers to cut down on electricity consumption. Moreover, despite the rapid spread of mobile communications around the globe, today there are still about one billion people worldwide that do not have access to any telecommunications services. Particularly in Africa and the Middle East, the lack of infrastructure (transport and electricity grid) prevents provisioning of cost efficient access to wireless communications. One reason behind the lack of mobile services is the unfavourable ratio of energy cost and revenue (discussed above for mature markets) considering an ARPU of 3 dollars per month and a high energy cost of 20 cents per kwh common for developing countries, arriving at around 10% of overall revenue (EBIT). Since ARPU, strongly depending on regional economic strength, expectedly increases very slowly in developing countries, the key lever here is the cost of electricity. If there is no electrical grid to power network equipment, more often than not operators have to resort to diesel-powered sites. The transport of diesel to remote sites roughly doubles the cost per gallon compared to the price at the pump, which often forbids feasible business models. EARTH WP2 D2.1: Economic and Ecological Impact of ICT 18 / 25

19 USD per million Btu (2008) electricity crude oil FIGURE 10. Expected evolution of energy prices [17] Here, alternative energy sources can provide a more favourable alternative. However, as of today, the number of BS sites powered by alternative energy is quite small. For instance Alcatel-Lucent reports to have installed only about three hundred solar powered BS sites globally as of 2009, but envisions a global market potential of more than one hundred thousand sites in 2012 [13]. In order to facilitate the availability and use of mobile communications around the globe, the exploitation of alternative energy sources for site operation is absolutely essential. Highly energy efficient base station equipment is the key enabler for alternative energy technologies to be applicable RAN ENERGY CONSUMPTION An exponentially rising demand for mobile data rates and traffic volume cannot be met by increasing system bandwidth alone, but in addition requires increased spectral efficiency of the network. Very high spectral efficiency, however, can only be provided through higher site densities and advanced transmission schemes (such as, e.g., multi-user MIMO and base station cooperation techniques). These schemes in turn require advanced signal processing and hardware complexity, and, consequently, generate an increasing electricity need for the BS site equipment. In the light of globally increasing electricity prices it becomes even more important to consider energy efficiency of the network as a design criterion. Since the base stations have the largest share in the overall network electricity consumption, improvements in BS efficiency can have a significant impact on overall electricity needs. In order to quantify the overall electricity consumption of worldwide RANs between 2007 and 2020 we again consider the scenarios defined in Section Results for different scenarios are illustrated in FIGURE 11. Scenario 1 (no improvements) might be pessimistic, but it can serve as a projected upper bound of RAN electricity consumption for energy-efficiency research initiatives. Scenario 5 by definition is a bit optimistic, since it considers the installed base of equipment before However, high savings through new technologies could be an enabler for such a scenario. Comparing Scenario 1 and Scenario 5, the difference in global RAN energy consumption is around 20% in 2014, while it shoots up to almost 60% in Total electricity savings amount to about 70 TWh per year in 2020, which equals to the total annual electricity consumption of the countries of Ireland and Portugal together. These numbers themselves show the tremendous potential of energy efficiency research in the field of mobile communications. EARTH WP2 D2.1: Economic and Ecological Impact of ICT 19 / 25

20 TWh per year Technology Potential x total traffic 20 3x base stations 1x RAN energy cons Scenario 1: No improvements Scenario 2: Cont. improvements Scenario 3: New technologies Scenario 4: Alternative energy Scenario 5: Large swap of eq. FIGURE 11. Global RAN electricity consumption projected until 2020 for different scenarios. The EARTH project aims at identifying some of the key levers for mobile radio communications to stay sustainable. Adding the planned 50% EARTH improvements to the current positive trends, it might even be possible to provide for 600 to 1000 times the amount of mobile data (0.8 million TB in 2007 to million TB in 2020, data and voice) and three times as many base station sites (3.3 million to 11.2 million) with zero increase in the total electricity consumption of all RANs globally. 4. SUMMARY AND DISCUSSION Mobile communication networks will have an increasing ecological and economic impact. This study quantifies the overall carbon footprint of mobile communications, as well as the overall RAN electricity consumption until Our projections suggest that the overall carbon footprint of mobile communications will almost triple between 2007 and 2020 if no additional means for reduction are taken. The overall footprint of 86 Mton CO 2 e as presented in this report is in contrast to the one presented in [6], where the footprint in 2007 with 150 Mton CO 2 e was largely overestimated. This overestimation is rooted in their use of electricity consumption data for older base station sites and a network model, which was more suitable for fixed, wireline networks. The footprint predicted in [6] for 2020, however, seems to be underestimated compared to the current study, as the future number of mobile subscriptions was lower compared to more recent projections [2], and the larger footprint of smartphones and laptops were not considered (see TABLE 2). We also show that mobile device manufacturing, driven by the increasing number of smartphones and laptops, will have approximately the same footprint as RAN operation by TABLE 2. Carbon footprint - comparison to Smart 2020 [6] Carbon footprint [Mton Co 2 e] This study (CI 2020 scenario) Smart 2020 Total active subscriptions 7.6 billion 4.8 billion Mobile phones (incl. smartphones) Laptops (USB dongles) EARTH WP2 D2.1: Economic and Ecological Impact of ICT 20 / 25