Energy Metrics and Figures of Merit for Planning EE Communication Networks Professor Timothy O Farrell Chair of Wireless Communications Department of Electronic and Electrical Engineering The University of Sheffield, UK 1/16
Presentation Outline Introduction Remarks on current EE framework A new EE framework Figures of Merit & Metrics Radio Access Network case studies Macrocell densification Microcell densification HetNets vs. Small Cells Conclusions 2/16
Introduction Energy consumption is a concern in communication networks, in particular in cellular mobile radio networks. Despite intense global research since 2009, there still isn t a consistent evaluation framework for unambiguously comparing the EE of different network topologies for network planning purposes. For example, a question pertinent to 5G For best EE should only small cells be deployed or HetNets? largely remains an open question! 3/16
Current EE Framework Essentially, one joint EE metric and figure of merit (FoM) has emerged the Energy Consumption Ratio (ECR). ECR! E M J Bit 1 EE = 1 ECR = M E E is energy consumed, M is data volume The rational being EE relates useful output (bits) to energy consumed (J). Then, EE is measured in various forms: EE = M / T E / T = S P = Throughput Power Bit J 1 4/16
Remarks Initially, E was the transmitted signal energy; later total energy to account for the equipment overhead energy. T is a common observation or coherence period. The EE calculated depends on how S and P are defined and where and when they are measured. Some concerns when comparing networks: As E is total energy, E(M) = ECR x M is problematic (e.g. M = 0); Between two network topologies, does the network with the largest EE consume the least energy if the data volumes differ (i.e. M 2 M 1 ) 5/16
New EE Framework E 1 RAN 1 {E 1 {o} M 1 T T 1, P 1, A 1, n 1, α 1, C 1 } I/P Energy O/P Energy Data volume Observation Time BTS Tx duration Total BTS Power BTS coverage area Number of BTSs BTS Load factor BTS mean capacity E 2 RAN 2 {E 2 {o} M 2 T T 2, P 2, A 2, n 2, α 2, C 2 } α i = T i T {i = 1,2} 6/16
Figures of Merit (FoM) Consider a FoM based on the ratio ECR 1 :ECR 2 ECR 1 ECR 2 = EE 2 EE 1 = E 1 M 1 E 2 M 2 = E 1 E 2 M 2 M 1 This suggests three essential FoMs Energy Ratio ER = E 1 E 2, Throughput Ratio TR = M 2 M 1 Joint Energy Throughput Ratio ETR = ER TR = E 1 E 2 M 2 M 1 7/16
Remarks on FoM ER compares energy consumption definitively. TR compares data volumes definitively. ETR is a joint FoM which compares energy consumption if and only if TR = 1 (i.e. M 2 = M 1 ) since for this condition ETR = ER. Therefore, comparing network designs based only on EE defined in terms of ECR is problematic. We now require metrics for ER, TR and ETR. 8/16
Metrics (Homogeneous, 1-sector cell plan) For ETR: For TR: ETR = ECR 1 ECR 2 P 1 S 1 P 2 S 2 T = M 1 n 1 α 1 C 1 = M 2 n 2 α 2 C 2 TR = M 2 M 1 S 2 A 2 S 1 A 1 since S i = α i C i {i = 1,2} and n 2 n 1 = A 1 A 2 For ER: ER = ETR TR = E 1 E 2 P 1 A 1 P 2 A 2 9/16
Metrics 10/16
RAN Case Studies P bts = αp rh + P oh 0 α 1 P P rh = n s n a P trx + tx η pa η cl P oh = P ac + P bh + n s { P ps + P } pu C = Blog 2 1+ 1 α + mkdγ N SIR min GP tx C bts = 1 A A C da 11/16
Macrocell Densification ER TR ETR RAN ER RAN TR RAN ETR Macrocell Baseline 12/16
Microcell Densification ER TR ETR RAN ER RAN TR RAN ETR Macrocell Baseline 13/16
Small cells vs HetNets ER TR ETR 10 MHz Macrocell benchmark HN1 = Macro + Micro HN2 = Macro + Pico HN3 = Micro + Pico 14/16
Conclusions Designing for EE RANs requires at least three FoMs (ER, TR and ETR) as determined by three metrics (P/A, S/A and P/S). Having an accurate power consumption model of the network equipment (including backhaul) is essential to the design process. IMEC/Greentouch power models (www.imec.be/powermodel) released 18 June 2015 - Power model for today's and future base stations. And of HetNets vs. small cells? HetNets (as in adding small cells to existing macrocell infrastructure) is more about increasing capacity rather than reducing energy consumption. Replacing macrocells with small cells can reduce energy consumption (up to a point) as well as increase capacity. Some next steps: Combining our analysis with ETSI ES 203228 in a Switzerland case study; Adding cost km -2 as an additional FoM and metric. 15/16
Thank you Questions? Acknowledgements: 1. Simon Fletcher, NEC Telecom MODUS, Leatherhead, Surrey, UK. 2. Konstantinos Samdanis, NEC Laboratories Europe, Heidelberg, Germany. [1] T. O Farrell and S. Fletcher, Green Communication Concepts, Energy Metrics and Throughput Efficiency for Wireless Systems, Green Communications: Principles, Concepts and Practice, First Edition, Edited by Konstantinos Samdanis, Peter Rost, Andreas Maeder, Michela Meo and Christos Verikoukis. 2015 John Wiley & Sons, Ltd. Published 2015 by John Wiley & Sons, Ltd. (in print) 16/16