Effects of joint macrocell and residential picocell deployment on the network energy efficiency Holger Claussen Bell Laboratories, UK

All Rights Reserved © Alcatel-Lucent | June 2008 Problem Overview  Increasing costs of energy and international focus on climate change issues have resulted in high interest in improving the efficiency telecommunications industry  Telecommunications is a large consumer of energy (e.g. Telecom Italia uses 1% of Italy’s total energy consumption)  This results in significant CO2 emissions which  contributes to climate change, and  will result in increased costs due to carbon taxation. Question: How can we contribute to reducing the energy consumption? (a) Directly by improving the efficiency of cellular networks (b) Indirectly by reducing the need for travel Example: rising oil prices in recent years source:

All Rights Reserved © Alcatel-Lucent | June 2008 Agenda 1.Direct effects of improving efficiency of cellular networks  Mixed macro-pico cell topology  Network power consumption with today’s technology  Potential Macrocell improvements  Potential Picocell improvements  Future network power consumption 2.Indirect effects of improving networks  Teleworking  Teleconferencing to reduce travel 3.Comparison of direct and indirect effects & conclusions

All Rights Reserved © Alcatel-Lucent | June Direct effects of improving efficiency of cellular networks

All Rights Reserved © Alcatel-Lucent | June 2008 Reduce the power required to operate our networks: Mixed macro-pico cell topology The concept  Use home-BS deployed by the user to supplement macro-cell coverage  Use the users internet connection as backhaul  Allow public access for home-BS  This results in no costs for the cell deployment, the site, electricity, and backhaul for the operator Objective of this investigation  Analyse the impact of such a mixed deployment on the total energy consumption and CO 2 emissions of the network

All Rights Reserved © Alcatel-Lucent | June 2008 Scenario assumptions Assumptions for user demand and distribution  Wellington, NZ + Suburbs (10x10km)  Population: (Wellington 160k, region 420k)  Mobile users: (95% of population) Population NZ = 3.7M, Vodafone: 1.9M, Telecom NZ: 1.6M  Usage: 740 min/user/month = 24 min/user/day = 8 calls/day (3 min)  Homes: assuming 3 persons per home  Demand is based on real measurements, extrapolated to the considered operator market shares of 10%, 20%, 30% and 40% Assumptions on emission and energy costs:  Electricity Emission factor = kg CO 2 / kWh  Electricity price = 86.3 Euro / MWh  Carbon emission trading value = 21 Euros/tonne CO 2

All Rights Reserved © Alcatel-Lucent | June 2008 Joint macro- and picocell deployment Home-cell deployment:  random in homes that have the distribution of the evening traffic  cell coverage area: 100x100m  up to 8 active users  Power consumption: 15W Macro-cell deployment:  Macrocells take care of the remaining users  Shared bandwidth: Supported active users per macro- cell depends on their requested data rate.  Different numbers of supported active users are considered: 30 (high speed data) to 240 users (voice) per macro-cell  This results in a VERY ROUGH approximation of the required number of macro-BS  User-distribution: max(dist_business, dist_evening)  users_covered_by_HBS  Power consumption: 2500 W for a 3 sector, 1 carrier base station (480 W power amplifier, 2020W base & control). A small fraction of randomly deployed home base stations can achieve a significant user coverage! With increasing home base station coverage, fewer macrocells are required to provide full user coverage By deploying picocells, the required number of macrocells is reduced to achieve the same capacity

All Rights Reserved © Alcatel-Lucent | June 2008 Energy consumption and CO 2 emission of different deployment scenarios – Today Challenge:  Macrocell coverage becomes less energy efficient compared to picocell coverage with increasing demand for high data rate services Approach:  A mixed deployment of macrocells for area coverage and picocells for the main demand reduces the total network energy consumption and CO 2 emission significantly. Model Results – Wellington NZ:  Maximum expected CO 2 reduction from direct effects (assuming 30 users/macrocell) would be up to approximately 2250 tons CO 2 /year for covering the full population in Wellington  Total carbon reduction value: Euros/year (for full population coverage)  The total saved energy costs: Euros/year (for full population coverage) Network energy consumption for operator with 40% market share (Today) 100% of energy costs paid by operator 97% of energy costs paid by end user The highest energy savings can be achieved when a small fraction of the customers have picocells deployed Picocell contribution increases linearly

All Rights Reserved © Alcatel-Lucent | June 2008 Macrocell improvements: Power Amplifier Efficiency improvements over time and their drivers Source: Georg Fischer, Bell Labs Nuernberg

All Rights Reserved © Alcatel-Lucent | June 2008 Macrocell Improvements: Efficiency improvements by change in architecture Current architecture: Digital RF-Signal Gen (Radio) PA Cable. Up to 2.5dB losses (40% of the power!) Diplexer 8-Element Antenna 1dB Loss in divider network (20% loss) PA, ~60% of Power lost in Heat Connector Cable, 0.5dB Loss (10%!) Tower-Top-architectures: Digital RF-Signal Gen (Radio) PA, (MUCH smaller) Diplexer Antenna (single Element) RF-Signal Gen (Radio) Diplexer RF-Signal Gen (Radio) Diplexer NO Cable Losses! Ground - Tower By Changing the architecture, min. 50% of the required RF-Power can be saved! But: more power in parallel radios etc. required… Source: Florian Pivit, Bell Labs Ireland

All Rights Reserved © Alcatel-Lucent | June 2008 Picocell improvements Introduction of idle mode procedures  In urban areas public picocell deployments can quickly result in high over provisioning of capacity (20% of customers with picocells can serve up to 80% of the total demand) Switch off picocells temporarily in areas where sufficient capacity is already provided by other picovells Other possible areas for improvements  more efficient processing  Power saving states when only partially loaded not considered here

All Rights Reserved © Alcatel-Lucent | June 2008 Assumptions:  Macrocell efficiency is improved by 33% by improved PAs and architectural improvements.  Picocells dynamically switch off when the area in which they are deployed already provides sufficient coverage and capacity. Results:  The improved efficiency results in a significant further reduction of the total energy consumption, energy costs, and CO 2 emissions.  Reductions of up to 70% compared to a macrocell network with today’s technology are feasible, for high data rate demand in urban areas (30 user/macrocell).  The benefits of a mixed macro- and picocell topology will increase further as both technologies mature. Energy consumption and CO 2 emission of different deployment scenarios – Future improved Technology Network energy consumption for operator with 40% market share (Future improved Technologies) 100% of energy costs paid by operator 85% of energy costs paid by end user Higher macrocell efficiency Higher picocell efficiency When picocells dynamically switch off based on demand, more than the optimum number of picocells can be deployed without significantly increasing the energy consumption Picocell contribution does not increase linearly anymore

All Rights Reserved © Alcatel-Lucent | June Indirect effects of improving communication systems:  Teleworking  Teleconferencing to reduce travel

All Rights Reserved © Alcatel-Lucent | June 2008 Effects of teleworking  A reduction of approximately 6MWh in Energy usage can be achieved by telework per person compared to full time office work  The main savings result from reduced travel.  The energy for heat and light is similar in all cases  The reduction results in a value of 33.6 Euros under the carbon emission trading scheme per year  The average travel cost reduction per year is far greater at 698£ = 938 Euros (assuming car travel with 7.1 l/100km and kg CO 2 /km, and price of unleaded petrol pence per litre)  At a national level, the effect of 5 million people working at home would save about 8 million tonnes of CO 2, 1.4% of UK total CO 2 emissions. based on study by BT Laboratories

All Rights Reserved © Alcatel-Lucent | June 2008 Teleworking Example: Wellington + Suburbs  Approximately 65% (130000) of the population are working.  Assumption: New communication technologies would enable an increase of 10% (13000) of the working population to work from home.  For Wellington this would result in a CO 2 reduction of up to tons of CO 2 per year.  This would correspond to a carbon reduction value of Euros per year.  The total travel cost reduction would be 12.2 x 10 6 Euros per year

All Rights Reserved © Alcatel-Lucent | June 2008 Reduced travel due to Teleconferencing Example: Wellington International Airport  Wellington Airport ( airport.co.nz/html/business/statistics.php)  flights per year, approximately 95% are domestic, and 5% international  55% of flights (60500) are business related  Assumption: A typical domestic flight distance is 800km and emits kg CO 2 /km resulting in a total of tons CO 2 emission. Total emission of flights = tons CO 2  Assumption: A typical international flight distance is 8000km and emits kg CO 2 /km resulting in 187 tons. Total emission of 3025 flights = tons CO 2  Per 1% reduction of business flights in Wellington would result in a reduction of tons of CO 2 per year.  This would correspond to a carbon reduction value of Euros per year.  In addition the corresponding reduction in travel costs would be roughly 87 x 10 6 Euros (assuming: domestic flight = 150 Euros, International flight = 700 Euros)

All Rights Reserved © Alcatel-Lucent | June Comparison of Direct and Indirect effects & Conclusions

All Rights Reserved © Alcatel-Lucent | June 2008 Comparison of direct and indirect effects of improved networks Results:  Improving the efficiency of network equipment can directly reduce both OPEX and CO 2 emissions  Indirect CO 2 and cost reduction as a result of improved networks can be far greater that the direct effects.  Examples of indirect effects resulting from improved networks are:  increase in teleworking  replacement of business trips by teleconferencing Opportunity:  Teleworking and teleconferencing have an enormous potential to reduce both costs and CO 2 emissions if the user experience is improved. Possible solutions are:  improve technology to provide the required higher data rates to homes and offices  reduce the data rates required for high quality video conferencing (e.g. by improving compression). Potential CO2 reduction for Wellington per year Potential cost reduction and carbon value for Wellington per year Direct effects Indirect effects

All Rights Reserved © Alcatel-Lucent | June 2008 Conclusions Direct effects: Architecture improvements  A mixed macro- and picocell architecture can significantly reduce the energy consumption of cellular networks in urban areas where macrocells are capacity limited  Effect expected to increase in the future when both technologies mature  Attractive for operators since energy for picocells is paid for by the user Indirect effects: Teleworking and reducing travel  Improving telecommunication systems can reduce energy consumption indirectly by improving teleworking and video conferencing  Reducing travel has a very high impact on the energy consumption and emissions  Indirect effects have a much higher impact as direct effects. Reducing the environmental impact can be achieved best by a combination of (a) Improving the efficiency of eetworks (b) Improving communication systems to promote teleworking and reduce travel

All Rights Reserved © Alcatel-Lucent | June