Joint Dynamic Energy-efﬁcient Spectrum Allocation and Routing in Two-tiered 4G Cellular Systems Wahyu Pramudito and Emad Alsusa School of Electrical and Electronic Engineering, University of Manchester, United Kingdom Wahyu.Pramudito-2@postgrad.manchester.ac.uk and E.Alsusa@manchester.ac.uk Abstract—This paper proposes a new resource management algorithm that utilizes the self-organizing network functionality for optimising spectrum usage and minimising cross-tier interfer- ence between macrocell and femtocells and co-tier interference between femtocells. To achieve this, the proposed algorithm dynamically maximizes usage of all the available subcarriers by exploiting knowledge of the interference power at each user. Tak- ing into account the associated overhead and power consumption it will be shown that the proposed algorithm improves the average data rate per user as well as the energy efﬁciency of femtocells networks in all femtocell density scenarios, and, is also superior to existing techniques. The fact that the proposed RRM minimizes the energy efﬁciency and maximizes the average data rate makes it an ideal contender for adoption in future green heterogeneous wireless networks. I. I NTRODUCTION It is undoubted that femtocell technology provides a low- cost solution for green heterogeneous cellular network systems in comparison to picocell and distributed relay antennas. This however comes at the cost of increasing interference in the form of macrocell-to-femtocell (cross-tier) and femtocell-to- femtocell (co-tier) interference due to the lack of site planning [1]. In 4G cellular network that employs orthogonal frequency division multiple access (OFDMA) in downlink, downlink interference are practically reduced using radio resource man- agement (RRM), which includes frequency spectrum allo- cation and power control, in contrast to high complexity interference cancellation [2]. OFDMA RRM can be classiﬁed into three categories, which are distributed, centralized and self-organizing network (SON) RRM [3], [2]. Centralized RRM works by adding a central node in a high density femtocell network, where the signal from a femto access point or Home Node BS (HNB) can be strongly received by all user equipments (UEs) in the same shadowed area [2]. The purpose of the central node is to compute which HNBs and UEs gives the highest received signal-to- noise and interference ratio (SINR) at a given subcarrier since in this enviroment, one subcarrier will be allocated to one HNB only. This is achieved by periodic monitoring of the channel state information (CSI) and path loss between HNBs and UEs. Obviously, the main problem of this RRM besides high complexity in order to monitor CSI over short period is low spectrum efﬁciency when the interference from one HNB may not affect all femtocell UEs. For this reason, this RRM will not be considered further in this paper. On the other hand, each HNB allocates its UEs’ subcarriers independently from neighbour HNBs such that the interfer- ence can be minimized in distributed RRM [4]. Using this approach, UEs continuously monitor the received interference from surrounding environment and report it back to the serving HNBs. The subcarrier allocation then can be altered in order to minimize the interference power at the receiver. Each femtocell achieves this by allocating their UEs with half of the available subcarriers only. However, self-organizing RRM may require several iterations before the optimum subcarrier conﬁguration that results in minimum interference can be obtained. Adding the fact that only half of the subcarriers will be used regardless the interference at each UE, this RRM can not be implemented efﬁciently. Adaptation with the environment can be solved by incor- porating SON functionality, which includes self-conﬁguration, self-optimization and self-healing [5]. This allows the HNBs to continuously communicate with neighbour HNBs and constant signal sources monitoring at UEs from different base stations (BSs) [6]. Hence, RRM utilizing this functionality, which is called SON RRM, can be implemented efﬁciently. A ﬁx frequency pattern allocation between adjacent femtocells ob- tained from SON is described in [3]. The frequency allocation stays ﬁx until new neighbour HNB is detected. If a UE receives high interference from neighbour HNBs, the serving HNB asks its neighbour HNBs to conﬁgures the transmitted power or the subcarriers allocation [3], [6]. However, ﬁx frequency alloca- tion may waste frequency spectrum when neighbour HNBs do not perform transmission. Furthermore, simple request to conﬁgure neighbour HNBs’ frequency spectrum and power transmission might not be applicable in practice because the HNBs have interest to serve their UEs as much as possible. This will result in an unsolvable conﬂict between femtocells. So, these two techniques will not be considered further in this paper. Inspired by the ability of SON functionality in heteroge- neous cellular network to monitor the surrounding environ- ment and to allow BSs communications, this paper proposes an adaptive SON RRM through routing algorithm for femtocell technology. This algorithm allows Femto Gateway (FGW), which connects the femtocells with the macrocell networks, to act as a central node for all conﬂicting BSs and each user will be assigned with subcarriers fairly depending on their environment in a similar manner with distance vector routing protocols (DVRP) in data networking [7]. The energy efﬁciency of the proposed technique will be evaluated under co- and cross-tier effect in this paper. In the case of co- Globecom 2013 - Symposium on Selected Areas in Communications 978-1-4799-1353-4/13/$31.00 ©2013 IEEE 2555