140821 Security Level：Open How to reduce Telecom CO2 while traffic is avalanching? Case Study – Mature EU Country 2010-2020 Rev A Tomas Edler Senior Expert Energy Efficiency Huawei Technologies Sweden HUAWEI TECHNOLOGIES CO., LTD. www.huawei.com Abstract As telecommunication is evolving from basic voice and text services to data, media and entertainment, the WW and European CO2 targets are at risk. The energy consumption of ICT industry is already twice of air traffic energy consumption. Historically, telecom energy efficiency has improved substantially, but how long can we mitigate expected traffic avalanche? Based on best projections of traffic and usage trends and best solutions for capacity and efficiency, I will show results from a study on How we can reduce energy consumption of wireless networks while supporting traffic growth. The results are compared to other studies. HUAWEI Page 2 Content Traffic Growth Scope of study What is unique Country emulation Traffic variations over cells Traffic models used How to improve efficiency – Technology Generations – BS architecture – Dynamic Power Management – Modelling Hardware Efficiency – Spectrum & HW refarming – Traffic load HW utilizaion – Energy saving features RAN results Other studies Conclusion HUAWEI Page 3 Incredible Success of Wireless Communications Last 45 years: 1 Million Increase in Trafficz Martin Cooper’s law The number of simultaneous voice/data connections has doubled every 2.5 years (+32% per year) since the beginning of wireless Source: Personal Communications in 2025, Martin Cooper Martin Cooper Inventor of handheld cellular phones 4 HUAWEI Source: Wikipedia Traffic growth Mobile Data Growth Sweden: 90%/Y 2010-2012 2010-2020 estimates: CISCO: +61%/Y, 117X 2013-2018 METIS : 1000X Huawei study: 67/Y 2010-2020 How Can We Sustain this Growth? – Continuous network evolution – More traffic for the same price – Mitigate traffic growth with Energy Efficiency HUAWEI Page 5 Predictions for the Future Rapid Network Traffic Growth – 61% annual data traffic growth – Faster than in the past! – Exponential increase – Extrapolation: 20x until 2020 200x until 2025 2000x until 2030 How Can We Sustain this Growth? – Continuous network evolution – More traffic for the same price 6 HUAWEI Operator’s Demand Operator’s Demand Cooperate with Verizon, Qwest organize ATIS NIPP TEE standard Take energy consumption as key element CO2 emission reduced by 80% @ 2020 (vs.1996) HUAWEI CO2 emission reduced by 50% @ 2020 (vs.2006) Energy Consumption reduced 50% Improve equipment EE by 30% energy saving -- CEO Cesar Alierta Total power, petrol, water consumption Page 7 “Low energy is key performance”, said by TI CTO Stefano Pileri Develop energy saving plan for NGN2 Per-user power reduced by 20% @ 2020 (vs.2006) Introduction HUAWEI Page 8 Scope of study Holistic view, emulating field RAN evolution in central EU 2010-2020 Sample EU ”countries” Evolution of traffic, subscribers, UE’s and RAN 2010-2020. – Energy Consumption and Efficiency evolution Dense Urban/Urban/Suburban/Rural areas Deployment evolution 2G-4G, HW, SW and spectrum Total RAN aspects, i.e Base stations, site infrastructure (cooling, power, back-haul) HetNet in Dense Urban area. Energy Saving solutions ( HW/SW, BS, site, RAN) HUAWEI Page 9 What is unique in this study The holistic view including ”Non-Hype” realistic PS traffic growth, 67%/Y, 170X in period Traffic growth from a PS (Packet Switch) resource view Multi-RAT evolution 2-4G Spectrum refarming Higher utilization of RAN HUAWEI Page 10 Country emulation A typical mature central EU country Analysis area 100.000 sqkm 100000,0 Subs/ SQkm 10000,0 1000,0 Subs/ SQkm 100,0 10,0 DU DU 0,2% ISD, km U SU % of sites R U 11% % of subs 5 R 27% 4 R 20% 3 ISD, km 2 SU 62% 1 SU 36% 0 DU HUAWEI U SU R Page 11 U 42% DU 2% Traffic load variations over cells Low traffic hours Wireless Network Traffic is not evenly distributed over cells High traffic hours Traffic 50% of traffic Based on data from EARTH project Source: LTE for UMTS - OFDMA and SC-FDMA Based Radio Access By Harri Holma, Antti Toskala Wiley 2009; ISBN 978-0-470-99401-6 45% of traffic 5% of traffic 10% HUAWEI 50% 100% of Cells Page 12 Traffic Models Traffic Evolution, Busy Hour traffic Voice coder data more demanding. 12.2 kbps voice = 30 (GSM)-75(3G) kbps ”Equal PS resource” for 2G/3G mix considered Year Total kbps PS kbps CS kbps E BH CS kbps/E of PS Daily model Traffic BH Mid Low Average H/day 15 4 5 - 10,00 Load 0,31 0,2 0,1 0,25 2010 0,56 0,113 0,45 0,015 2012 0,84 0,316 0,525 0,015 2014 1,48 0,881 0,6 0,015 2016 3,14 2,46 0,675 0,015 2018 7,63 6,88 0,75 0,015 2020 20,11 19,21 0,9 0,015 Growth 35,7 170,0 2 1 30 35 40 45 50 60 2 Total 36X Growth Subscriber Traffic evolution PS 170X Growth +67%/Y Total kbps CS kbps 1,00 CS kbps Load % Avg 25% 15h 30 20 10 0,10 2010 2012 2014 2016 2018 2020 4h 5h HUAWEI Hr/Day Voice X-over 2013 80% Page 13 Voice 5% How to improve Efficiency Technology Generations BS architecture Dynamic Power Management Modelling Hardware Efficiency Spectrum & HW refarming BS load utilization Energy saving features HUAWEI Page 14 Efficiency Evolution – Technology Generations Massive MIMO? Efficiency evolution for 1000 Traffic growth 10x per 5Y (58%/Y) Source: Green Touch kbps/W typical Macro Base station deployment Transferred bits/W LTE-A 100 8X LTE WCDMA /HSPA 10 5Y Total cell average WCDMA/DCH 1 EDGE 0.1 HUAWEI LTE-NG GSM GPRS Substantial improvement of spectral & BS eff Page 15 ”Less to win” on BS eff. Close to Shannon BS Architectures Power efficiency* Macro BS RBS site power efficiency - Example PCI Power consumption Index: Macro => DBS Feeder loss elimination RRU natural cooling Avg AC powersite / RF Powerantenna Total Eff ”Efficiency”: 4% PCI = 25 Eff ~ RF Power Traffic model Climate model RBS Site Backhaul Mains Input Power Climate Eq. Rect. RBS BS HEX or ventilation. Eff. 65% EE ~ double With other HW Eff. 85% Eff. 50% Eff. 15% Energy efficiency* PCI RBS site power efficiency – Example – RRU 2010 improvements 3-5X EE. PCI Power consumption Index: Total ”Efficiency”: 20% E PCI =5 f Avg AC powersite / RF Powerantenna Eff f Traffic model Climate model ~ RBS Main RRU Mains Input Power Backhaul HEX Eff. 95% HUAWEI RF Power Eff. 90% Rect. Eff. 92% Page 16 Base Band Eff. 25% BB& RRU Natural Cooling Field data – Traffic vs Power Consumption Source: Bath Green Radio Conference 2009, Orange presentation Power Consumption - not scaling with traffic Potential for power saving - scaling with traffic HUAWEI Page 17 DPM case - Scaling with load Equipment view: - Scaling with load Power Consumption BB, ”Radio” Poor Scaling! • Cooling – scales • PA – fair scaling PA • Base Band, ” Radio” i.e. ”TRX”: Target - Poor scaling Cooling • How to improve? • Where to improve ? HUAWEI Avg Page 18 Load Max Modelling of HW Efficiency Macro Base Station parts Site parts HUAWEI Page 19 HUAWEI Page 20 HUAWEI Page 21 HUAWEI Page 22 Spectrum & Hardware evolution 2010 RAT/BW 2010 RAT/BW Ban d Macro BS 2012 HW refresh DU U SU R 7 - - - - 8 - - - - 9 G4C G2C G2C G2C 18 G4C G2C G2C G2C 21 U5 U5 U5 U5 2013-2020 New RAT/BAND/HW New New New 26 New Micro/Pico BS 21 - - - 100% Macro Base Stations HUAWEI New - - Macro/DBS Note DU U SU R L10 - - - 7 Cap & Coverage L10 - L10 L10 8 Cap & Coverage G3C U5 G2C U5 G1C U5 G1C U5 9 GL Multi-RAT Radio Units L20 L20 - - 18 Cap U20 U5 - - 21 Cap L20 - - - 26 Cap 21 Cap Micro/Pico BS U10 Reuse of sites +u-Base stations Ba nd - - 50% Macro 50% DBS Page 23 - Traffic Load – Utilization of BS 30,0% Busy Hour Load 2010 25,0% 20,0% G 15,0% U 10,0% 5,0% 0,0% DU U SU R 30,0% Busy Hour Load 2020 25,0% 20,0% G 15,0% U L 10,0% 5,0% 0,0% DU HUAWEI U Page 24 SU R Example of Energy Saving Features Feature name Symbol Power Saving Description & applicability (techno, frequencies, equipment, release ..) The eNodeB can shut down the PAs in the time of empty symbols. MBSFN sub-frame could be used to reduce the reference signal further so that more empty symbols are available for PA to shut down RF Channel Intelligent In MIMO mode, the carrier for a cell is transferred through different transmission channels. When no traffic is Shutdown on the cell, the carrier can be switched off on part of transmission channels. Inter-RAT energy saving When there is light traffic in an area that is covered by multi-RAT carriers, some of LTE carriers can be blocked, and all services can be automatically taken over by other RAT carriers that remain in service. Intra-RAT energy saving During low traffic hour, deactive some cells and let neighbour cells expand to the coverage hole. Small cell energy saving When low traffic in an area covered by Macro and Micro/Pico carriers, some of small cell can be blocked, and all services can be automatically taken over by macro carriers that remain in service. Dynamic Cell Power Off (GSM dual band only) 900 MHz/1800 MHz GSM dual-band network. In a specified period, if the traffic is low and a 900 MHz cell can carry all the traffic in the coverage area of an 1800 MHz cell, then the 1800 MHz cell can be powered off to reduce the power consumption of the BTS. HUAWEI Page 25 RAN Results Total energy saving 50% Improvement mechanisms MWh/Y Energy Consumption, 100.000 Sqkm Network 500000 - Eff gains 2G – 4G: ”RAT” gain, Single carrier => 450000 multi carrier, Flexibility of multi RAT BS 400000 EC, 65% CAGR - Refarming of spectrum and Hardware - Utilization eff gain 350000 - Energy saving features 300000 - ”Dynamic power management” ie shut down of 250000 redundant resources. Issue - How to cope with traffic growth beyond EC, 65% ESF* 200000 2010 2012 2014 2017 2020 2020? - The ”total” traffic growth will increase, as PS Total energy saving 50% dominates Traffic growth PS only 170X - Less to win on ”RAT-gain” and utilization Traffic growth ”total”: 36X - New technology candidates: ”Total” Efficiency gain: 72X - LTE-A, HetNet for dense traffic areas, Massive MIMO, 5G technologies, mm wave frequencies.... HUAWEI Page 26 80,00 RAN Results – Energy Efficiency 70,00 60,00 Energy Efficiency, TB/MWh Area DU 1,9 U 1.0 SU 0,16 R 0,19 All 0,26 Year 2010 7 2,4 0,3 0,4 0,5 2012 10,8 3,5 0,7 0,8 1,1 2014 26 11.3 2,8 3,7 4.6 2017 E Eff. Growth kb/J Factor 75 165 39 44,2 97 45 12 26 77 14,3 31 75 18,5 41 70 2020 2020 20102020 50,00 U SU R DU All 40,00 30,00 20,00 10,00 0,00 2010 HUAWEI 2012 2014 2017 2020 Other studies Femto offloading Macro BS 1/2 1,9X SE* w. Femto  Energy Efficient High Capacity network by offloading hi QoS users to Femto. HetNet with ”regular” users and few heavy, high QoS users. 5 X SE w. Femto Femto BS for heavy users at ”random worst” location. For a legacy network, EE increases substantially by adding Femto’s to comply to traffic growth. EE gain 1,9X for ISD 500m and 7X for ISD 800m. The larger ISD, the higher EE gain. Highest gain with the first 20% femto’s, than gain is declining. If Femto’s are deployed randomly, there is no * SE: Spectral Efficiency EE gain. HUAWEI Page 28 7 X SE w. Femto Other studies Femto offloading Macro BS 2/2  Energy Efficient High Capacity network by offloading hi QoS users to Femto. Adding Femto’s & Backhaul to 100% of Heavy users adds 10% to RAN Energy 1,8 X EE w. Femto 3 X EE w. Femto consumption, so Energy Saving gain is still substantial. HUAWEI Page 29 Other studies Pico offloading Macro  Energy efficiency improvement through pico base stations for a green field operator. Macro ISD adapted for traffic density, considering pico BS. Pico BS located at hot spots. ~15% EE gain, 2 Pico, constant traffic density Limited efficiency gain. HUAWEI Page 30 Other studies Pico offloading Macro  Energy efficiency gains throuigh traffic offloading and traffic expansion in joint macro pico Macro ISD constant. Pico BS added at hot spots for traffic growth. Limited efficiency gain. 5 X Efficiency when DBS PRB offloaded 50% Eff gain when DBS PRB not offloaded HUAWEI Page 31 Other Studies – Massive MIMO Energy Efficiency Reference: E. Björnson, L. Sanguinetti, J. Hoydis, M. Debbah, “Optimal Design of Energy-Efficient Multi-User MIMO Systems: Is Massive MIMO the Answer?,” Submitted to IEEE Trans. Wireless Communications. Golden Combination: Large array gain Many simultaneous users Fractional pilot reuse Low Per-Antenna Power Use handset technology 32 HUAWEI Comparison to other studies Ericsson – Vodafone study  P. Frenger, Y. Jading, and J. Turk, 2013, “A case study on estimating future radio network energy consumption and CO2 emissions”, Results: Energy Consumption 2020: -60% ”compared to today”. Similarities: Smart refarming of Hardware when LTE is introduced. Differencies: Ericsson /Vodafon study deploys less LTE Macro BS and more LTE Pico nodes. HUAWEI Page 33 Conclusion 2010-202: 50% Energy Reduction of RAN Energy Consumption is possible, if – Spectrum is refarmed and LTE introduced – HW refreshed, old 2G and 3G equipment phased out for LTE and Multi-RAT BS. – New efficient BS architecture and Hardware – Energy saving features - Dynamic power management – HetNet (micro/pico BS) is used in dense areas > 2020: A challenge to further reduce RAN energy consumption – HetNet, Higher order MIMO, Massive MIMO deployment to support capacity and further spectrum refarming /Hardware swap. HUAWEI Page 34 Recommended reading RECOMMENDED READING  Energy efficient high capacity HETNET by offloading high QoS users through femto Usman, M. ; Vastberg, A. ; Edler, T. Networks (ICON), 2011 17th IEEE International Conference on DOI: 10.1109/ICON.2011.6168500 Link: http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-98289 Publication Year: 2011 , Page(s): 19 - 24  Energy efficiency gains through traffic offloading and traffic expansion in joint macro pico deployment Arshad, M.W. ; Vastberg, A. ; Edler, T. Wireless Communications and Networking Conference (WCNC), 2012 IEEE DOI: 10.1109/WCNC.2012.6214158 Publication Year: 2012 , Page(s): 2203 - 2208  Energy efficiency improvement through pico base stations for a green field operator Arshad, M.W. ; Vastberg, A. ; Edler, T. Wireless Communications and Networking Conference (WCNC), 2012 IEEE DOI: 10.1109/WCNC.2012.6214157 Publication Year: 2012 , Page(s): 2197 - 2202  P. Frenger, Y. Jading, and J. Turk, 2013, “A case study on estimating future radio network energy consumption and CO2 emissions”, available at: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6707825  E. Björnson, L. Sanguinetti, J. Hoydis, M. Debbah, “Optimal Design of Energy-Efficient Multi-User MIMO Systems: Is Massive MIMO the Answer?,” IEEE Transactions on Wireless Communications, Submitted for publication [Available online: http://arxiv.org/pdf/1403.6150] HUAWEI Page 35 140821 Thank You www.huawei.com www.huawei.com HUAWEI TECHNOLOGIES CO., LTD.
© Copyright 2018