1. A! Aalto University Comnet Energy Efficiency Techniques & Challenges for Mobile Access Networks Maliha Urooj Jada Supervisor: Jyri Hämäläinen Work carried out @Aalto University Date: 19-05-2011

2. A! Aalto University Comnet Outline  Motivation  Research Question  Power Efficiency Model for Mobile Access Networks –Equations behind results –Result Analysis  Impact of Femtocells to the WCDMA Network Energy Efficiency –Energy Model –Comparison Scenarios –Methodology –Result Analysis  Case Study: Idle Mode Feature Requirement in Femtocell  Conclusion and Future work

3. A! Aalto University Comnet Motivation "In last few years energy efficiency for mobile networks has become important for both industry and academia"  Although Mobile networks do not have considerable share in the overall energy consumption of –ICT Sector >> Responsible for 2% to 10 % of the world energy consumption.  However, the cellular companies are looking for adequate alternatives that can improve the energy efficiency; reducing energy costs while also improving their corporate image (being environmentally conscious).  With the time, new technologies (such as LTE) will be deployed. Along with the enhancement in network’s capacity, the energy consumption will also increase.  Hence, there is a need to take an action from now to improve the energy efficiency of existing mobile networks otherwise energy consumption will keep growing in proportion to the increase in peak hour traffic.

4. A! Aalto University Comnet Energy Saving Aspects in Mobile Networks/ Research Questions  There exists two approaches towards Energy Efficient Mobile Networks –To find apropriate solutions to the energy efficiency challenges for already existing networks. –To design future networks from energy (and cost) efficient perspective (Greenfield perspective).  Estimations for the future Telecom’s energy consumption admit large variations. –Also, difficulties and uncertainties exist in estimating the potential savings due to the different methodologies used.  The largest fraction of energy consumption and accordingly CO 2 emmissions occurs in Base Stations. The ways to minimize the BS energy consumption –Improvement in BS energy efficiency –Usage of system level and software features –Usage of BS site solutions

5. A! Aalto University Comnet Power Efficiency Model for Mobile Access Networks  This work presents a simple Radio Network Power Usage Model. BS Transmission power –Based on the model, two different network deployment scenarios have been compared from energy consumption perspective.  Aim >> To make visible the impact of different link budget and network parameters to the network energy consumption.

6. A! Aalto University Comnet Power Efficiency Model for Mobile Access Networks  Two comparison scenarios –First scenario: The inter-site distance (ISD) is fixed and focus is on power consumption. –Second scenario: In second scenario the BS transmission power is constant but the ISD is scaled which reflects to the number of BS sites.  First scenario is related to the case where operator is updating BSs in an existing deployment while second scenario considers Greenfield operator that is, building a new network.  Comparison Cases –Comparing power usage in two different networks (network 1 and network 2) by computing the ratio between powers that are needed in all BSs in the networks.

7. A! Aalto University Comnet First Scenario Result Analysis  The impact of cell edge rate requirement (TP min )and outage probability (Pr out )are large.  Cases 1 (Diamond) & 2 (Triangle) –TP min has been doubled w.r.t reference n/w >>Power difference is large in case 2 than in 1 because P out has been decreased in case 2.  Cases 3 (Square) & 4 (Circle) –Improvement in (W eff ) BW efficiency and (Γ eff ) SINR efficiency (reflecting use of multiple antenna)will reduce the effect of Increased rate requirement and decreased outage probability. (W eff ; Γ eff ; TP min ; Pr out ).

8. A! Aalto University Comnet Equations behind results We can compute ע and ratio D1/D2, Where, P N is the noise power I M is the interference margin G is the BS antenna gain α is parameter that includes impact from carrier freq & antenna height β is path loss exponent σ SF is standard deviation for shadow fading W eff is BW efficiency & Γ eff is SINR efficiency TP min is the minimum throughput & Pr out is Outage probability

9. A! Aalto University Comnet Second Scenario Result Analysis  Changes in rate and outage requirement have crucial impact to coverage.  Doubling rate requirement on cell edge will lead to approximately one a half fold increase in BS density.  If also P out is halved then two fold increase in BS density is needed.  This huge cost source can be reduced by introducing multi- antenna sites. Reference parameters: (0.56;2.0;0.5;0.10) New parametersValue of R Case 1: (0.56;2.0;1.0;0.10)1.47 (1.68dB) Case 2: (0.56;2.0;1.0;0.05)2.10 (3.22dB) Case 3: (0.66;1.1;1.0;0.10)0.97 (-0.10dB) Case 4: (0.66;1.1;1.0;0.05)1.39 (1.44dB) (W eff ; Γ eff ; TP min ; Pr out ).

10. A! Aalto University Comnet Impact of Femtocells to the WCDMA Network Energy Efficiency  Investigated the potential energy savings by deploying femtocells into a macrocellular WCDMA network.  Study utilized WCDMA Downlink load equations.  Determined >> How load sharing between femtocells and macrocells will contribute to the overall energy consumption of the network by exploiting the Cell Breathing phenomenon.  Introduced two comparison scenarios >> To estimate how the change in network configurations will effect the energy consumption in the modified network w.r.t the reference network. Cell Size Breathing: At BS noise floor is proportional to load. To maintain specific SINR in case of increased noise floor, mobile terminals have to move closer to the BS site due to their limited power which effectively shrinks the size of the cell. Similarly, when load decreases, the cell expands outwards.

11. A! Aalto University Comnet Energy Model  Dimension >> kWh/km 2  Area covered by a three sector site: A=9/4.R 2 =ISD 2  For accurate calculations we adopt UMTS macrocell BS specific values: P oper =137W and P TX =57W (Ref *)  For femto BS input power we use two values 2W and 5W (Ref **)>> To show the difference in energy savings between both i/p power cases  The daily energy consumption per square kilometer in the network * G. Micallef, P. Mogensen, H. O. Scheck: “Cell Size Breathing and Possibilities to Introduce Cell Sleep Mode”. European Wireless Conference, p. 111-115, 2010. ** Press Release on NEC FP 810 femtocell product, 2010. http://www.slashgear.com/nec-fp810- femtocell-tiny-but-14-45-7mbps-data-rates-1073529/

12. A! Aalto University Comnet Comparison Scenarios First Scenario: Macrocell ISD is fixed, the change in n/w energy consumption is estimated when femtocell penetration rate is increasing. Addition of femtocells is only visible in macrocell load factor. Second Scenario: Macrocell ISD is not fixed, the addition of femtocells to macrocellular system decreases the macrocell BS density due to cell breathing and thus reduces the energy consumption.

13. A! Aalto University Comnet Methodology  To Calculate the daily energy consumption without femtocells  N F Femtocell addition >> Femtocells take certain portion of users (R f x100) % –R f =ratio between femtocell and macrocell connection –Number of macrocell sites decreasing>> reducing energy consumption of network  Assumptions >> Femto BSs are turned on only when there is traffic >> Each femtocell serves single user  Service rate=64kbps  Initial system load=0.9

14. A! Aalto University Comnet First Scenario Result Analysis 100%·R femto 0%25%50%75% (E/A) M [kWh/km 2 ] 24.3522.8821.4019.93 (E/A) Total [kWh/km 2 ], P F = 5W 24.3525.7727.1828.59 (E/A) Total [kWh/km 2 ], P F = 2W 24.3524.0323.7123.39  E/A decreases in Macrocells when Rf increases  5W femto BS power >> Total energy consumption in the network is growing  2W femto BS power >> Total energy consumption is slightly decreasing with additional femtocells

15. A! Aalto University Comnet Second Scenario Result Analysis  Energy consumption by macrocells is rapidly decreasing since lower load allows less dense macrocell grid  Decrease in macro BS density is limited by the non-femtocell users –If network is to be build from scratch then second scenario would be beneficial from energy efficiency perspective  50% femtocell (2W) penetration>> 63 % of energy would be saved  50% femtocell (5W) penetration>> 49 % of energy would be saved

16. A! Aalto University Comnet Case Study: Idle Mode Feature in Femtocell  In second comparison scenario case, significant amount of energy savings was observed. –Thus we continued with this case to discuss the importance of designing a special feature in femtocells >> Power Save Feature (Idle Mode) to achieve significant amount of energy savings in network  If femto BSs are activated all the time >> Femto BS energy consumption easily overtake the achieved savings on the macrocell side. –So femto BS should not remain activated if indoor traffic is non-existent, rather should enter sleep mode. –Femto BS should have the traffic sensing ability.  Addition in Previous methodology >> Assume that when femtocell utilized with PS feature, any fixed N F remain active throughout R f scale.

17. A! Aalto University Comnet Results Analysis  Energy savings for both types of femtocells (with and without power save feature) having input powers 2W and 5W.  Energy Saving is more when deployed femtocell possess power save feature.  When R f ~1, almost same energy is saved in both types of femtocells.This explains: –Femtocells without PS feature are more unfavourable when there is not much indoor traffic.

18. A! Aalto University Comnet  Case 1: N F =N users –With i/p power of 2W >> Network with femtocells that do not possess PS Feature consumes extra energy (No Energy Saving) until R f reaches 0.05 –With i/p power of 5W >> Extra Energy consumed until R f reaches 0.2. However, this extra energy is more than the extra energy consumed in case of 2W Femto BS In case of 5W input power this femtocell PS feature is necessarily required  Case 2: N F = 2.3 x N users –Energy Saving gets reduced. Number of Femtocells should be limited N F (upper bound)=N users >> Maximum gain in energy saving

19. A! Aalto University Comnet Conclusions & Future work  This work discusses the importance of reduction in energy consumption to achieve economic and environmental benefits.  It has addressed the possible ways to minimize the energy consumption on the BS sites.  Power consumption model is developed>> comparisons have been made to show the impact of different link budget and network parameters.  Impact of femtocells to WCDMA Network energy efficiency has been discussed, along with the importance of idle mode feature in femtocell. Future work:  To take into account the impact of daily traffic variations in previous work. For that purpose a new performance metrics will be needed.  To see how the co-ordinated and planned microcellular/picocellular/femtocellular network extensions would effect the energy efficiency.

20. A! Aalto University Comnet THANK YOU