Energy & Green Radio: challenges & sinergy

Transcription

1 Energy & Green Radio: challenges & sinergy Guido Riva Fondazione Ugo Bordoni

2 ICT: a low impact industry Mobile communications as well as fixed telecoms is itself a relatively low-impact industry when it comes to energy usage and carbon dioxide (CO2) emissions, despite its rapid growth. Estimates: approximately 0.14 per cent of global CO2 emissions and 0.12 per cent of primary energy use are attributable to mobile telecom. This compares with 20 per cent of CO2 emissions and approximately 23 per cent of primary energy use for travel and transport, for example. The annual CO2 footprint of the average mobile subscriber is around 25kg which is comparable to driving an average car on the motorway for one hour, or running a 5W lamp for a year. Source: Ericsson (2007)

3 Trends in Mobile Radio Evolution Service aspects Penetration of mobile devices exceeds 100% Consumers are hungry of higher and higher bit rates Diverging growth Traffic Subscribers expect to pay less for more Revenue Accordingly, cost per bit relatively decreases Time [yr]

4 Tarifs: relative cost per bit 10000, , ,000000!/Mb Bit rate [kbit/s] , !/Mb 1, , , , , , SMS dati a volume nuovi dati a volume Flat Telefonino Voce Internet a Tempo Internet a Tempo Internet a Tempo Flat PC Flat PC Flat PC

5 Power consumption In a typical cellular network, RBS plays the main role (till 80%) in power consumption MSC 20% Core Transmission 15% Data Center 6% Retail 2% RBS 57% Vodafone (2003) Edler (2008) Mobile handsets power drain per subscriber is much less than RBS

6 User need for greater data rate causes greater energy consumption The cost of energy is ever increasing Main drivers Transceiver Idling 19% Power Supply 16% Cooling Fans 13% 19 % 22 % 16 % 13 % 9 % 9 % 8 % Power Amplifier 22% 1 % 3 % Cabling 1% Transmit Power 3% Central Equipment 8% Combining/Duplexing 9% Transceiver Power Conversion 9% H.Karl (Sept.2003) EIA (2008) 2 complementary domains: Energy-aware innovative architectural solutions Energy-efficient hardware and software solutions

7 Holistic view of energy requirements Moving energy consumption from infrastructure to user devices might not be the best solution from the national economy perspective Manufacturing processes: embodied emissions in RBS and handsets Lifecycle assessments of mobile networks Improvements in power amplifier at the expense of power consumption of the other components is not a benefit from system viewpoint Ericsson (2007)

8 Technological improvements Ericsson (2007) System technologies Traffic Data Raw Materials (weight) Energy consumption System typologies Late 80 Analogue BS 21 voice calls (0,2 Mb/s) 2 tonnes 5 kw Building/shelter Nowadays WCDMA BS 150 users (11 Mb/s) 100 kg 0,5 kw Remote Unit

9 Main factors Approximately 2/3 of Cellular Networks energy consumption comes from RBS Approximately 2/3 of LCA impact comes from operational equipment Approximately 2/3 of the consumed daily energy is due to the transmission, whereas 1/3 to conditioning Lubritto ( )

10 TV Digital Switch Off Analog TV switch Off has a relevant impact from an energy viewpoint. A significative fraction of analog Tv sites are not turned on, after switch off. Moreover, digital TRx could theoretically lower their Transmitting power by 10 db with respect to their analog version; a typical value is 6 db Numero impianti accesi Prima dello switch-off Dopo lo switch-off [-10:0] [0:10] [10:20] [20:30] [30:40] [40:50] [50:60] [60:70] ERP prima dello switch-off [dbw]

11 Energy savings Some TRx were already digital (no power change). The overall ERP saved for each TRx ERP-class can be calculated. Power savings exceed 6 db, i.e. more than 75% saved Tx power. In terms of energy (GWh), savings can be roughly estimated around 30-40% ERP totale [dbw] Pre Switch-off Post Switch-off Active TRX Total ERP (dbw) 76,1 69, Prima dello switch-off Dopo lo switch-off [-10:0] [0:10] [10:20] [20:30] [30:40] [40:50] [50:60] [60:70] ERP per impianto [dbw]

12 Spectral (Bandwidth) Efficiency Shannon s theorem fixes the limit capacity C in a noisy Gaussian channel Spectral Efficiency [bit/s/hz] Shannon capacity Spectral efficiency SNR [db] η B = C B = log 2 1+ S N Example 2 Available bandwidth: 10 MHz Achievable SNR: -20 db Maximum capacity: 14.4 kb/s Spectral Efficiency [bit/s/hz] C = B log 2 1+ S N Example 1 Available bandwidth: 10 MHz Achievable SNR: 30 db Maximum capacity: ~100 Mb/s Shannon capacity 100,000 10,000 1, ,100 0,010 0,001 SNR [db]

13 Power (Energy) Efficiency The Shannon limit capacity is higher when SNR is higher, e.g. more power is equivalent to more capacity. So we can define Power Efficiency as: Power efficiency [bit/s/w] or [bit/j] η P = C S = B S η B = η B E b = 1 N 0 As signal propagates, its power S (and also Energy per bit, E b ) decreases and so the achievable capacity ,00-60 Signal Power [dbm] NF=7 db η B SNR We assume a receiver NF=7 db. The achievable capacity depends on the available bandwidth as well as on S Capacity [Mb/s] , , ,00 100,00 10,00 B=10 MHz B=! B=1 GHz 1,00 0,10 0,01

14 Emf power limits and capacity Italian law poses a 6 V/m threshold to emf emissions, defining a respect volume around a Tx site. This poses a limit to the achievable capacity even when bandwidth approaches infinity. Let us assume a consumer device: receiver NF=7dB; antenna gain=0 dbi. With a bandwidth of 10 MHz, we have a maximum capacity of 300 Mb/s Capacity@900MHz [Mb/s] Capacity@2000MHz [Mb/s] B.eff.@900 B.eff.@ ,00 40,00 35,00 Of course, as signal power will decrease, so will do the corresponding optimum capacity ,00 25, , ,

15 Efficiency and propagation We expressed the power efficiency as the ratio between the capacity and the signal power at a point. From a network viewpoint, is better to express this efficiency as the ratio between the capacity and the power we have to transmit from the antenna to obtain that signal power at the receiver (i.e. the Transmitting Power P T ). We need to introduce the effects of frequency f and distance d on path loss L, decreasing the ERP (the product of Tx antenna gain and P T ). 180,00 S = ERP + G L(d, f ) L(d,1950) =137,4 + 35,2 log 10 d Hata model h b = 45 m, h m = 1,5 m Path Loss and SNR [db] 160,00 140,00 120,00 100,00 80,00 60,00 40,00 20,00 L(d,1950) SNR [db]@1950 0, ,00 Rx distance from Tx [km]

16 Efficiency Trade Off Bandwidth Efficiency Power Efficiency 1,6000 7,0000 1,4000 6,0000 Bandwidth efficiency [b/s/hz] 1,2000 1,0000 0,8000 0,6000 0,4000 5,0000 4,0000 3,0000 2,0000 Power efficiency [normalized b/s/w] 0,2000 1,0000 0, , SNR [db]

17 Some general considerations The increasing cost of energy challenges the traditional business model. Bandwidth efficiency is important in order to offer high data rate services, but should not be the only leading criteria: power efficiency is also worthwhile as energy associated OPEX costs increase and health emission regulation sets intrinsic limits to the achievable cell capacity. Two coordinated approaches could be envisaged: A gradual bandwidth expansion, possibly achieved with both new use bands (prone to delay and uncertainties) and new network architectures (pico- or femto-cells, relaying, heterogeneous networks, ) A specific attention to less bandwidth-efficient requirements and more power-efficient needs, whose cost per bit is more favourable.

18 Spreng s triangle D. T. Spreng, a physicist, hypothesized that for any given task, f1(t)f2(e)f3(i) = a constant t=time, E=energy, I=information required to perform the task To carry out a given task, you can save energy by taking more time or using more information, or you can save time by using more energy, and so on. Despite its original scope, the main suggestions lays in the role played by t,e and I: two vertex corresponds to bandwidth and power-efficient situations. But the main challenging scenarios are in the remaining part of the figure: the application of ICT to other economic/technical fields

19 Thank you