2010 European Wireless Conference

ICT & WIRELESS NETWORKS AND THEIR IMPACT ON GLOBAL WARMING Hans-Otto Scheck Nokia Siemens Networks Linnoitustie 6, FI-02600 Espoo, Finland phone: + (358) 7180 36329, email: [email protected] web: www.nsn.com

ABSTRACT

economical impact by comparing financial performance vs. energy consumption and in the attempt to find a correlation between ICT and national primary energy consumptions. In section 5 we finally draw possible conclusions from the previous chapters.

Global warming caused many countries to call for drastic energy savings within the coming years. At the same time we see a rapidly increasing demand on Telecom services. Estimations of the future Telecom and ICT sectors’ energy consumption are varying; nevertheless all predict an increase of energy consumption. On the other hand, the ICT sector is believed to be one of the solutions on the way to a low carbon society. This paper analyses the estimates of CO2 emissions from mobile networks and possible environmental impacts of ICT resulting from different user behavior. Finally an attempt is made to find an impact of ICT on the energy consumption of a complete nation. 1.

2.

ICT AS PART OF THE PROBLEM

The annual global primary energy consumption in 2006 was around 140,000TWh [IEA08] and the production of electrical energy around 17,400TWh. The energy consumption of the ICT industry was estimated at between 370-830TWh [IDA08], [GES08]. With other words, around 0.3-0.6% of the global primary energy or around 2-5% of the global electricity was consumed by the ICT industry.

INTRODUCTION

Initially energy consumption was discussed as part of Global Warming and was seen as one element of environmental responsibility. The European Parliament created a Code of Conduct for the ICT industry, other countries follow with similar measures and a number of global organizations are now developing standards and methodologies to measure and minimize ICT energy consumption. The recent economical downturn had its toll on Telecom equipment spending. Good progress has been made on more energy efficient Telco equipment; however, less old equipment has been out-phased and replaced with new, more efficient equipment. On top of the original demand for more efficient new equipment came an increasing pressure to optimize and improve the efficiency of already used Telecom equipment. Despite all efforts for further improving the ICT efficiency, growing energy consumption has been estimated. On the other hand, Telecom services have the potential to reduce the energy consumption of other sectors. From an environmental perspective the total global energy consumption matters and therefore, the impact of the ICT industry on the society and the resulting overall energy consumption has to be understood. This paper analyzes the energy consumption from following perspectives: Section 2 describes the impact of mobile network operation, including a review of current assumptions and estimates from various sources. Section 3 presents the potential positive impact of Telecom and ICT, including energy savings by using ICT solutions. Section 4 analyzes the

978-1-4244-6001-4/10/$26.00 ©2010 IEEE

Figure 1 – ICT energy share of global primary energy

A continuous growth in telecom services and further drastic increase in the amount of transferred data has been predicted. Remarkable in all these figures is the relative large variance or uncertainty of the ICT energy consumption. To a large extent this uncertainty comes from missing methodologies how to define energy consumption or even which elements should be considered in the calculation. As shown in figure 2 the expected energy consumption growths estimations for ICT varies from +13% for Germany, +110% for Japan [NIK09] and +70% for North America by 2015 [IEA09]. As all three countries are on a similar in-

911

dustrialization level, it’s hard to believe that their future development differs that much.

commonly agreed that radio base stations are the main power consumers of the cellular network as shown in figure 3.

Figure 3 – Global energy consumption of cellular networks Figure 2 – Different ICT energy growth estimations for different countries

These discrepancies can be understood better by a closer analysis of one segment of the ICT sector, where we have relative detailed data from various sources: mobile networks. A mobile network can be split into four key elements: user equipment, cellular base stations, network controller and backbone network as shown in figure 3. One of the first and often cited sources for mobile network energy consumption comes from the analyst firm ABI. Their estimate for the energy consumption of mobile networks in 2008 is ~180TWh [ABI08]. This figure is based on estimations of the number of base stations and the average power consumption of base stations. Unfortunately, there exists no average base station, or even a clear definition of a base station. It might vary from a single sector mast-mount site consuming some 200W to a multiple-carrier, multi-band, multi-standard site with around 10kW power consumption. Estimates made by Nokia Siemens Networks are based on two different methods: first, we calculated the energy consumption based on the number of our sold products, their typical power consumption, their typical lifetime and our market share on mobile base stations. In a second estimation we calculated the average annual energy consumption per mobile subscriber based on the reported energy consumption from some telecom operators and the respective number of subscribers. Multiplied with the global number of subscribers we finally got a second value for the global cellular base station energy consumption. Surprisingly, these two methods matched well with each other, but resulted in ~60TWh annual energy consumption which is only 1/3 of the ABI estimate. Also the SMART2020 report acknowledged the difficulties in calculating network energy consumption by stating “no data were available from each provider on specifically how much energy their networks consumed” [SMA08]. These very different results of a small and relative well defined part of the very complex ICT sector shows, how difficult accurate estimates are, and that currently published energy figures should be used with care. However, it is

912

A similar challenge is the calculation of the energy consumption of the user segment (mobile phones). Based on the average mobile phone usage as reported for example by the Japanese Ministry of Internal Affairs and Communications and typical power consumption figures provided by suppliers the global energy consumption would be below 0.1TWh. When assuming that every second subscriber might keep his charger connected to the power grid even after the phone has been charged the resulting energy consumption is about 2TWh, or roughly 20 times as much. Above energy consumptions estimates are limited to the product use-phase of a telecom network. If we take the total lifecycle footprint into account the impact of the user segment increases significantly and also the uncertainties become even larger. Infrastructure products have a life span of over 10 years; therefore the embedded energy is small compared to the use phase energy. Because of the short lifetime of user equipment, their footprint is usually determined by manufacturing and related operations. The multiple functions of user equipment (a mobile phone is a camera, music player, etc., a PC is used not only for document processing but can replace a fixed line phone and fax machine) make a separation in Telecom, ICT and user equipment nearly impossible and deliver arbitrary results based on the chosen segmentation and assumptions. 3.

ICT AS PART OF THE SOLUTION

The ICT industries have a significant impact on our society, from business culture to private live. Therefore, the environmental impact of the ICT can’t be evaluated simply from the energy consumption. The ICT industry has been recognized as the key to a low carbon economy. It has been estimated that ICT could reduce CO2 emissions in other sectors of approximately five times as much as the ICT’s own emissions and deliver about 1/3 of the expected total abatements in 2020 [SMA08]. The five most important ICT related abatements are expected to come from dematerialization, smart motors, smart logistics, smart buildings and smart grids. Under dematerialization we understand the substitution of high carbon products with low carbon alternatives e.g. re-

mote work vs. commuting, videoconferencing vs. travel or electronic billing instead of paper bills. As already shown in the previous section, already the power consumption estimate of the different parts of a mobile network is difficult and inaccurate. Naturally, even larger difficulties and uncertainties exist in estimating the potential savings. As in the previous chapter, we’d like to analyse the accuracy of estimated abatements in a simple, defined example. One of the key contributors to global warming is travelling by car. Remote work has the potential to drastically remove the need for commuting and hence emissions. As the example in figure 4 shows, annual reductions in the order of 1t CO2 per person could be easily achieved. The table includes many assumptions and the results vary accordingly. The potential savings depend naturally on the travelled distance. The assumed distance from home to office might be on average too low or more than 15% office days might be needed. Also about the potential extra energy consumption of the home office can be argued. Unlike in the previous chapter where we noticed significant discrepancies between the various estimates, the estimated savings of ~1t agree nicely with results presented by ETNO & WWF [ETN08].

mote office. Alternatively the money could be spent on a flight to the Canary Islands, a popular escape route from the Scandinavian winter. The resulting emissions of 600-900kg CO2 will compensate for most of above savings. The example clearly shows how much the actually achieved savings depend on the initial assumptions and behaviour (figure 5). Properly invested we could be soon on the way to a low carbon society, or alternatively, all the ICT savings could be compensated by the rebound effect. The estimation above is made for a single person. For the global impact we have to estimate how many people would be able but also willing to work remotely rather then at the office. There are clearly many jobs which can’t be handled remotely, but even those who could face a radical change in their working practice and management methods.

Figure 5 - CO2 emissions from commuting, home office, home office including rebound effect and home office including reinvestment of savings

4.

Figure 4 – Example: CO2 emission savings by remote work

Although we agree on the figures for remote work savings, the impact on the environment isn’t given. Any energy saving is followed by cost savings (which is in many cases a far more important driver than environmental protection). The saved money is now available for other expenditures, with the unfortunate result of potentially additional emissions. This is called the rebound effect and might have a significant impact. The person in above example would save in the order of 357liters of gasoline, equivalent to about 450€ at today’s gasoline costs in Europe. How will the money be spent? If the 450€ are invested in solar panels, the energy generated will save annually another 60-100kg of CO2. The cumulative savings of two years invested in solar panels could provide all or even more of the additional energy needed for the telecom service for the re-

913

ICT AND NATIONAL ENERGY CONSUMPTION

Calculating the current and future energy consumption and abatements of the ICT sector is a challenging task, and the results depend on many unpredictable factors. However, ICT isn’t a new invention. Mobile phones and the internet are around for over one decade. Therefore we should be able to see first effects on the energy consumption of the leading industrialized countries. In this section we analyze the development of energy consumption and economy on the example of Germany. The global primary energy consumption is rising, first driven by OECD countries, now by the rapid development in China and India. These countries are currently transforming from agricultural to industrial nations. Although this includes a transition to an ICT society, it doesn’t ”leapfrog” the step of industrialization, which is reflected in the relative low fraction of ICT energy consumption in these countries. Between 1995 and 2005 the electricity consumption in Germany increased by about 13% and remained relative constant over the recent years. At the same time primary energy consumption rose only by ~3% and dropped between 2006 and 2008 [AGE08] as shown in figure 6. This decline can only

partly be explained by the financial crises starting in autumn 2008.

reason for this, but other factors like structural changes and political energy saving measures play a significant role, too. If the German energy productivity (energy consumption versus GDP) was improved because of an increasing usage of information technologies, we should see significant differences between countries with a different ICT penetration. While the energy vs. GDP figures are comparable with those of countries like Sweden, Germany and the U.S., we see a dramatic difference when comparing the figures with China and India: these countries spend nearly four times the energy for the same GDP as shown in figure 8. In contrast to their high energy per GDP, the GDP per capita is significantly lower in these countries.

Figure 6 – Energy consumption development in Germany

The global ICT energy consumption was calculated in chapter 2 to around 0.3-0.6% of the global primary energy consumption. The primary energy consumption of Germany and the German ICT sector were ~4000TWh and ~57TWh respectively (2008), resulting in a share of ~1.4% or somewhere between 2.5 and 5 times the global average. Not unexpectedly, the ICT sector plays a significant larger role in Germany compared to the global average. During the time from 1991 to 2008 gross domestic product (GDP) was continuously growing from 1535B€ to 2496B€ [STA09], or inflation corrected from 1761B€ to 2271B€ in Germany. Dividing the primary energy consumption by the GDP, the resulting figure presents an “energy consumption of productivity”. As shown in figure 7 the needed energy per GDP has been reduced continuously during the recent 19 years. This is also reflected in the total primary energy consumption. While electrical energy consumption kept rising during the 90’s we see a clear stagnation from 2005 onwards. Even more important, the total primary energy consumption was reduced during the recent years, falling in 2008 to levels from the early 90’s as shown in the previous figure 6.

Figure 8 – Energy productivity, or amount of spend energy per GDP

Also the key industrial countries have an energy consumption per capita which is much higher than the global average, figure 8 demonstrates a dramatic advantage in the energy productivity for the key industrial countries. This might indicate, but doesn’t prove the impact of ICT on energy productivity.

Figure 9 – Amount of spend energy per GDP vs. internet usage

The figure 9 shows the energy/GDP ratio vs. the internet usage of the EU countries. All countries with an internet usage of >75% spend less than 3kWh/$GDP, countries with a lower internet usage spent a significantly higher amount of

Figure 7 – Gross domestic product (GDP) and energy consumption / GDP development in Germany

Obviously Germany became more energy efficient during this time. The increased usage of ICT might be one possible

914

though the examples of Germany and the European countries are far from conclusive, it might be considerably more accurate to analyse energy consumption and economical development of countries and regions instead of referring to potential mitigating effects of the ICT sector.

energy per GDP. While Internet usage is not a guaranty for high efficiency, it clearly is one essential factor. 5.

CONCLUSIONS

Estimating the energy consumption of the ICT or even of the Telecom sector is a challenging task with widely varying results. The ICT sector has a clear potential to reduce primary energy consumption, but the achievable savings are not only more difficult to estimate than the ICT’s own consumption, the outcome depends dramatically on human behaviour and how we utilize the resulting financial savings. If properly done it could lead us rapidly to a low carbon society, but there is the clear risk that rebound effects compensate a significant amount of the potential benefits. ICT is an enabling technology, but in order to utilize its full potential behavioural changes are needed. The example of Germany shows that economical growth is possible with reduced energy consumption. The reasons for the reduced energy consumption and improved energy economic efficiency could be manifold. The significant difference in economic efficiency and the correlation of internet usage with economic efficiency indicates that there exists a correlation between the usage of ICT and improved energy efficiency. More research has to be done to verify if this assumption is true, or if this is only an indirect effect. Al-

REFERENCES [ABI08] ABI Research, 3Q08, Mobile Networks Go Green [AGE08]Arbeitsgemeinschaft Energiebilanzen e.V. 2008, Energieverbrauch in Deutschland [ETN08] ETNO / WWF, Saving the climate @ the speed of light, 2008 [GES08] GeSI press release, June 2008 [IEA08] International Energy Agency, Key World Energy Statistics 2008 [IEA09] International Energy Agency “Gadgets and Gigawatts – Policies for Energy Efficient Electronics, 2009. [IDA08] IDATE news 453. [NIC09] METI, Nikkei Electronics Asia 2/2009. [SMA08] SMART 2020: Enabling the low carbon economy in the information age, The Climate Group, 2008 [STA09] Statistisches Bundesamt Deutschland, Statistisches Jahrbuch 2009, Volkswirtschaftliche Gesamtrechnungen

915