Green Radio - The Case for More Efficient Cellular Base Stations Pei-Jung Chung University of Edinburgh, UK p.chung@ed.ac.uk
Presentation Overview Introduction: Why Green Radio? Green Radio Objectives A Novel Base Station Location Planning Strategy Conclusions
Why Green Radio? An operator view Vodafone UK has an average load of 50MW Vodafone uses 1 million litres of diesel per day Vodafone (and Orange) have pledged to reduce CO2 emissions by 50% by 2020 57% of energy use is in Radio Access Data Centre 6% MTX 20% Core 15% Retail 2% RBS 57%
Where is the Energy Used? Operating electricity is the dominant energy requirement at base stations For mobile user devices, most of the energy is expended during manufacture CO 2 emissions per subscriber per year 3 9kg CO 2 Operation 2.6kg CO 2 4.3kg CO 2 Embodied energy 8.1kg CO 2 Base station Mobile 3. Tomas Edler, Green Base Stations How to Minimize CO 2 Emission in Operator Networks, Ericsson, Bath Base Station Conference 2008
Base Station Power Consumption 2003 H. Karl, An overview of energy-efficiency techniques for mobile communication systems, Telecommunication Networks Group, Technical University Berlin, Tech. Rep. TKN-03-XXX, September 2003. [Online]. Available: http://www-tkn.ee.tu-berlin.de/ karl/wg7/ag7mobikom- EnergyEfficiency-v1.0.pdf
Power Consumption Power Consumption per Base Station 2007/08 Target (2010) GSM 800W 650W WCDMA 500W 300W Nokia Siemens Networks
Energy Consumption The Base Station is the most energy intensive component of a 3G mobile network. A typical 3G Base Station consumes about 500 W with a output power of ~40 W. This makes the average annual energy consumption of a BS around 4.5 MWh (which is lower than a GSM BS). A 3G mobile network with 12,000 BSs will consume over 50 GWh per year. This not only responsible for a large amount CO 2 emission it also increases the system operating energy.
Energy Consumption The Challenge Since 2006, the growth rate of data traffic on mobile networks has been approximately 400% per year. With increased deployment of mobile broadband e.g. smartfone (iphone, blackberry, android, dongle), traffic will grow at least at this rate in coming years. This growth demands a much higher energy consumption than today. The challenge is how to design future mobile networks to be more energy efficient to accommodate the extra traffic.
Possible Solutions Green Radio Can we benefit from the use of the information below in the design of future mobile networks? Mobility pattern (location, speed and direction of mobile user) information Characteristic of multimedia traffic (traffic classification) Hour by hour variation in traffic Transmission power scaling (distribution) in order to use renewable energy for BSs.
Green Radio as an Enabler Costs Voice Data Traffic Time Revenue Diverging expectations for traffic and revenue growth Trends: Exponential growth in data traffic Number of base stations / area increasing for higher capacity Revenue growth constrained and dependent on new services Energy use cannot follow traffic growth without significant increase in energy consumption Must reduce energy use per data bit carried Number of base stations increasing Operating power per cell must reduce Green radio is a key enabler for growth in cellular whilst guarding against increased environmental impact Traffic / revenue curve from The Mobile Broadband Vision - How to make LTE a success, Frank Meywerk, Senior Vice President Radio Networks, T-Mobile Germany, LTE World Summit, November 2008, London
Green Radio Scenarios Developed World Developed Infrastructure Saturated Markets Quality of Service Key Drive to Reduce Costs Emerging Markets Less Established Infrastructure Rapidly Expanding Markets Large Geographical Areas Often no mains power supply power consumption a major issue
Key Research Questions What would be an appropriate green network architecture? A low power wireless network & backhaul that still provides good quality of service What are the best radio techniques? Need to consider across all layers of the protocol stack that collectively achieve significant power reduction
Green Radio Objectives Two key targets: 1. To identify a green network architecture - a low power wireless network & backhaul that still provides good QoS 2. To identify the best radio techniques across all layers of the protocol stack that collectively address the aspiration of achieving significant power reduction.
Architecture: Mobile VCE Technical Approach Energy Metrics & Models Primary and derived energy metrics to accurately quantify consumption Communications energy consumption models for the radio access network (RAN) architecture Energy Efficient Architectures For RAN technology, compare large versus small cell deployment Assess scenarios for placement of relay nodes Efficient backhaul in support of identified architectures Multi-hop Routing Bounding energy requirements by strict end-to-end QoS Exploiting delay tolerant applications and user mobility for energy reduction Frequency Management Identification of energy efficient co-operative physical layer architecture using emerging information theory ideas to remove interference Applying Dynamic Spectrum Access (DSA) to minimize energy consumption by utilising bands with low interference Solar-powered relaying allocating resources to match combined traffic and weather patterns
Energy Efficiency Macro Micro Pico Femto Step1: Large vs. small cells applying the energy metrics Step2: overlay Source & Network Coding and/or Cooperative Networking Step3: Evaluate from the following perspectives.? RRM BER/FER vs Eb/No Link Budget Mobility/Traffic Models Packet scheduling, handover, power control, load control Differentiated QoS, fast fading effects, UE speed, MIMO Energy consumption is proportional to distance UE movement, traffic types & mixes
Femtocells inside Macrocells? 4.4 Power Consumption per user (W) 4.2 4 3.8 3.6 3.4 3.2 3 2.8 30 users/ macrocell 60 users/ macrocell 120 users/ macrocell 240 users/ macrocell Embodied plus operating energy: power consumption per user for different levels of macrocell activity & support 2.6 2.4 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Fraction of Customers with Femtocell Curves: different user density, RHS all femto i.e. macro switched off. A lower user density in macrocell gives larger power saving (as energy overhead is shared over smaller number of users). The trade-off (minima) points for 30, 60, 120 and 240 users/macrocell imply 80%, 60%, 40% and 20% respectively for femtocell coverage. Corresponds to energy reduction ratio between 34.6% and 2.6%.
Power Efficient Hardware Base station efficiency Climate control 65% Power supply 85% Power Amplifier / transceiver 45% Feeder cables 50% Baseline overall efficiency 12% Advanced base station architectures Multi-mode and multi-standard Maximise equipment and base station re-use Integrated remote radio antenna Masthead PA eliminates feeder loss Integration avoids interconnect losses Passive thermal cooling Integration allows energy reductions Masthead electronics to avoid cable losses Target > 20% overall efficiency Advanced power amplifier techniques Target: > 60% PA efficiency
Macrocells vs Micro/ Femtocells In a conventional macrocell, a base station serves more than one hundred users. A microcell or femtocell base station serves a few users in a home or office environment. - better QoS, - lower transceiver energy, ca 0.1 ~ 1% of a conventional macrocell base station. Great potential in minimizing the overall long term energy of the network! *This part is supported by EPSRC grant EP/E018939/1 ``Bridging the Gaps Between Engineering and Mathematics.
Deployment of Smaller Cells Conventional Macrocell
Extra Energy Costs When evaluating energy efficiency of microcells or femtocells, one needs to consider additional energy consumption required in the large number of base stations. Trade-off between low transceiver energy and additional maintenance powers. Further, the minimal QoS in the network has to be always satisfied.
Base Station Location Planning for Minimal Power Consumption We focus on optimizing the number and locations of the base stations to achieve high long-term energy efficiency. Numerical results show that deployment of microcells improves energy efficiency significantly. The best planning is achieved through facility location optimization and stochastic programming.
System Model The problem is to place several smaller base stations in a conventional urban cellular network (in which all mobile users are served by one base station). No interference among users. Power consumption in the downlink. Pathloss term in modelling physical channel. Non-uniform user distribution.
Problem Formulation Assume N users on average in a region, locate M base stations so that the overall energy consumption is minimized, i.e. minimize (P c + P t ) subject to QoS requirement power constraint where P c is the power for operating the central equipment, P t is the total transceiver power in the network.
QoS Constraint The receive power at each user i needs to meet the minimum value Pi to ensure the QoS. We define the QoS as the minimal data rate. Given the pathloss 1/ L ij between user i and base station j, the QoS is guaranteed by M j = 1 P ij / L ij P P ij where denotes the power transmitted by base station j to user i.
Power Constraint at Base Station Each base station has power constraint on transmitting messages due to power amplifier limitations: P ij P up where P up is the power limit of the base station. The energy required for communication back to the core network is counted as operation power P c.
User Distribution User distribution is crucial to the base station planning. Our approach considers non-uniform user distribution, which is realistic in practice. Since we are concerned with the long term effect of the system, we consider user patterns taken across large time scales (e.g. days). The long term effect is addressed by assigning the probability of occurrence π k to the user pattern k.
Mathematical Formulation
Simulation Scenario 13 base station locations ( ) are predefined in a given region. Parameters are derived from the Long Term Evolution (LTE) standard.
Optimal Base Station Locations Scenario 5 selects 9 of the 13 BS locations.
Simulated User Distribution (total number of users N=30) region scenario 2 scenario 5 scenario 7 scenario 8 1 0 5 0 15 2 0 2 0 5 3 0 3 0 0 4 0 3 0 5 5 1 5 30 5 6 0 2 0 0 7 0 2 0 0 8 29 3 0 0 9 0 5 0 0
Energy Efficiency Our study shows that stochastic programming leads to more than 96% reduction of the power consumption on radio frequency generation in all cases!! Significant energy efficiency can be achieved by smart deployment of microcell base stations. More details can be found at the website http://www.maths.ed.ac.uk/ergo/preprints.html
Conclusions Green design is a key enabler for rapid data rate growth in mobile communications whilst guarding against increased environmental impact. We show that optimization of base station locations can reduce the energy consumption by more than 96%! Future work will consider more realistic cellular networks and various transmission protocols.
Acknowledgements Institute for Digital Communications, University of Edinburgh, UK Peter Grant, Steve McLaughlin, John Thompson, Ioannis Krikidis School of Mathematics, University of Edinburgh, UK Pablo González-Brevis, Jacek Gondzio Department of Electrical Engineering, Princeton University, USA Yijia Fan, Vincent Poor Mobile VCE Colleagues
Thank You!