Energy Efficient Load Sharing in LTE-A HetNets Petri Luoto, Pekka Pirinen and Matti Latva-aho Centre for Wireless Communications P.O. Box 4500, FI-90014 University of Oulu, Finland Email: {petri.luoto, pekka.pirinen, matti.latva-aho}@ee.oulu.fi Abstract—This paper addresses energy efficiency and through- put performance of joint LTE-A macrocell and femtocell deploy- ment. We consider two techniques discontinuous transmission (DTX) and offloading of users from macrocells to lower power femtocells. When the DTX is applied base stations delay trans- missions until sufficient user data is available to transmit using 40 percent or more of the system bandwidth. This allows the base stations to remain inactive for larger periods of time, and therefore reducing energy consumption. The target is to reduce the overall network energy consumption per area with minimal impact to the throughput. The numerical results show that when the majority of macrocell users can be offloaded to be served by femtocells the mean power per area consumption will decrease by up to 34.7 percent and at the same time a larger portion of users are able to meet the intended throughput target. When DTX is applied energy consumption is reduced further still to 45.3 percent. I. I NTRODUCTION Ever increasing capacity demands in densely populated areas can no longer be satisfied solely by large coverage area macro- and microcells. Smaller pico- and femtocells provide additional capacity and coverage where needed, specifically indoors. Thereby, Heterogeneous Networks (HetNets) [1], [2], being mixes of various cell shapes and sizes, can be seen as a cost-effective evolution step in the development of modern cellular network architectures. Small cells provide additional capacity but at the same time they bring new challenges to re- source and interference management. However, low femtocell transmission power and strong structural isolation due to walls relieves the interference problem between outdoor macro and indoor femto layer and also between different femtos. Another important aspect we would like to tackle is to investigate whether the densified network structure can also be energy efficient. This paper compares power per area power consumption in pure macrocellular deployment to two offloading scenarios where the same total load is shared between macrocells (mostly outdoor users) and indoor femtocells in relative per- centual proportions 50/50 or 20/80. Daily variations in the network load have been adopted from [3]. The traffic inten- sity profile has been extracted from operator measurements to reflect typical European country statistics. The energy consumption metric is chosen to be [W/m 2 ] or equivalently [kW/km 2 ] that allows for a fair comparison of different HetNet approaches to provide service over a predefined geographical area. The reminder of the paper is organized as follows. Section II describes the considered system model including network lay- out, power consumption model and discontinuous transmission enabled scheduler discussions. Section III gives information about the system simulator model with related key parameters and assumptions. Selected numerical performance examples are shown in Section IV. Finally, the concluding remarks are drawn in Section V. II. SYSTEM MODEL This section highlights the main elements of the evaluation framework, i.e., describes wireless network layout, indicates how power consumption is accounted for and how additional power saving is acquired through DTX operation. Detailed explanation for system model is provided in [4]. A. Network Layout The considered wireless network is composed of 19 macro- cell base stations (MBSs) each of them located in the corner of three hexagonal sectors. Hence, the overall hexagonal grid includes 57 sector cells. Overlaid to the hexagonal layout, dual-stripe buildings having indoor femtocells are dropped one per macrocell sector. In the dual-stripe layout there are 40 blocks of size 10 m x 10 m per floor. Attenuation due to internal and external walls is accounted for in the dual-stripe layout path-loss model. Femtocell access points (FAPs) are dropped in the blocks by using a predefined probability. It is also assumed that if there is a FAP there is always a femto user equipment (FUE). A three-sector snapshot of the network layout is shown in Fig. 1. B. Power Consumption Modeling A high level energy efficiency evaluation framework (E 3 F) was developed in EARTH project [5]. One integral part of that framework is the power model [6], [7] that calculates base station power consumption for various cell types and sizes (macro, micro, pico, femto) and network loads. This paper relies on the comprehensive work conducted there to evaluate the MBS and FAP power consumption. The model reveals that in MBS the power amplifier (PA) dominates the total power usage whereas contributions from baseband processing, RF components and overhead are less profound. However, in the case of smaller cells such as pico and femto, the baseband may have even larger share of power consumption than PA. The MBS PA power consumption is also strongly dependent on the actual output RF power level used in the transmission whereas FAP PA does not have such characteristic. EARTH power model combines all these components and it has been 1st International Workshop on GReen Optimized Wireless Networks (GROWN'13) 978-1-4799-0428-0/13/$31.00 ©2013 IEEE 119