A Resource Allocation Policy for Macro-Femto Double-layer Network

Transcription

1 A Resorce Allocation Policy or Macro-Femto Doble-layer Network Yeshn He*, Ping D, Wenjing Li College o Inormation Engineering, East China University o Technology, Nanchang, Jiangxi,330013,China Abstract Crrently doble-layer network model based on Macrocell and low power nodes can eectively increase the capacity o the commnication system. So traic congestion at data service can be solved in the model. Bt the mltiple intererence between Macrocell base stations and emtocells will seriosly damage commnication qality and redce system perormance. To solve this problem, we propose a hierarchical schedling resorces- joint intererence sppression algorithm. Firstly the system model which incldes inside and otside layer (I-O model) is constrcted. Then cross-tier and hybrid intererence is analyzed between inside and otside layer. Frthermore two schedling algorithms are presented or inside and otside layer. Allocation strategy o time - reqency resorce schedling strategy based on sb-channel and power level. Finally the whole system is simlated. Simlation reslts show that the proposed algorithm can achieve good reslts at some aspects sch as the spectrm eiciency o cell - edge ser and the average throghpt o system. Keywords the doble-tier and heterogeneos network; I-O model; resorce allocation; intererence sppression I. INTRODUCTION The evoltion o 3GPP organization gradally to 4G mobile commnication system whose representative is LTE-A promotes the development and standardization o home base stations and other major technologies. The homebase station has a smaller transmit power, coverage within 50 m, and data transmission is by mltiplexing reqency resorces o macrocell, ths orming a doble-tier and heterogeneos network in which macrocell and homebase station coexist. Wherein the dierent tiers reer to dierent types o celllar networks, while heterogeneos emphasizes a signiicant dierence in the power transmission and the coverage o a signal. Femtocell technology has become one o an important technology promoted in LTE-Advanced standard R12 version crrently [1], despite the small coverage, bt the FUE (Femtocell User Eqipment) still exist strong mtal intererences with the adjacent MUE (Macrocell User Eqipment). This orm o isomeric intererence has seriosly aected the qality o commnication and impedes a rther enhancing o system capacity and perormance. So how to schedle resorces to achieve a goal o ser signal intererence sppression or coordination is one o the most challenging isses in emtocell deployment nder the coverage o existing macrocell. Athors in [2] propose a method or optimizing homebase station, which aims to limit a downlink intererence o home base station to otdoor macrocell ser. The athors model the optimization problem as a mixed integer programming problem, reslting in Femto s maximm transmission power and operating reqency. However, in the actal FAP (Femtocell Access Point), de to high complexity, taking too long, it is not easible to implement this techniqe. In [3], the athors develop a simple techniqe to avoid the homebase interering the macrocell sers nearby. Bt it is based on the assmption: Femtocell can get inormation abot the MUEs rom the macrocell base station. Once the macrocell sers ser intererence, nearby Femtocell will redce it s transmit power[4]. In this paper, we consider system hierarchy, we simpliy the complex intererence between the LTE-A doble network sers by constrcting inner and oter layer system model, and proposed a way o schedling resorce method to sppress intererence. Finally, system-level simlation platorm have proved that both Macro and Femto layers have been better intererence management and sppression ater sing the algorithm. Research and simlation o this paper can provide a reerence or tre commnication network planning and optimization. II. SYSTEM MODEL AND INTERFERENCE ANALYSIS OF MICRO-FEMTO DOI /IJSSST.a ISSN: x online, print

2 YUESHUN HE et al: A RESOURCE ALLOCATION POLICY FOR MACRO-FEMTO DOUBLE-LAYER NETWORK North American markets will be the two key market o emtocell development [7]. The heterogeneos network architectre composed by macrobase station and home base station is shown in Fig. 1. Its characteristic parameters listed in Table I: A System Architectre and Model Homebase station, as a powerl soltion o indoor coverage, was introdced in 3GPP Rel-8 and was in-depth research in 3GPP Rel-12[6], in which added a new paradigm or operators. Till 2015, the shipment o emtocell will reach 54 million, o which China and TABLE I. CHARACTERISTIC PARAMETERS OF THE DOUBLE-TIER AND HETEROGENEOUS NETWORK. Femtocell Macrocell The transmit power o base station 46 dbm 23 dbm The eective coverage 1-3 km 50 m Interace Type S1 IP internet Providing open access to a large For some activation sers in range o sers amily or bsinesses Prposes covering dierent ranges [9]. However, i two or more emtocells sing the same reqency or sbchannel, then it is easy to generate co-channel intererence. Inter-layer intererence reers to that the sorce o intererence and UEs interered with are Femtocell 1 Internet Femtocell service gateway FUE2 FUE1 coming rom dierent layers. For example, MUEs near emtocells receive intererence rom downlink signal transmitted rom the emtocell, and FUEs are interered by macrobase station [10]. In LTE-A system, we sally adapt base adaptive power control technology and intelligent channel allocation techniqes to solve the intererence problem [11]. Figre 2 realistically shows a ew plink-downlink intererence problems broght by emtocells and macrobase stations being combined together to distribte network. MUE3 Femtocell 2 FUE3 Macrocell served MUE2 MUE1 Fig.1 The Heterogeneos Network Architectre o Macro-Femto From igre 1, we can see that the model o the inner and oter layer system, namely I-O model, can be constrcted to allocate hierarchically these resorces. The system is divided into inner and oter layers, and the ormer means the coverage o the main schedler by homebase station, while the later reers to coverage o the macro base station. B Cross-mixing Intererence Analysis The introdction o emtocell makes the network divided into two layers, not only bringing a high edge-coverage, bt also leading two intererences: inner-layer intererence and inter-layer intererence [8]. Inner-layer intererence happens between two or more emtocell intererence. De to the lexibility o emtocell deployment location and geographical proximity between them, reslt in an overlapping coverage when DOI /IJSSST.a Fig.2 Sitation o plink-downlink intererence in Macro-Femto doble network In the case o macrobase station in center, we discss 1.2 ISSN: x online, print

3 the downlink resorce allocation in Macro-Femto network. System bandwidth is divided into N orthogonal RBs [12]. Assme that eective coverage radis o the enteral base station is R m, and the base station shares C sbcarriers with F Femtos, and then th MUE has an eective SINR on sbcarrier c, as ollows: e N ' nn c 1 P P nc,, nc,, 1, n, c, n, c 1 ' Where N means the set o RBs allocated to ser, P is transmission power o ser on sbcarrier c,, n, c nc,, is SINR. So the transmission power o the th ser is: B nc,, ln 5BER Rc, log 21, where is M 1.5 represents the bit error rate nder BER. 1 III.THE ALGORITHM OF RESOURCE ALLOCATION TO SOLVE THE INTERFERENCE PROBLEM This paper presents an inner-oter layer collaboratively schedling algorithms. The inner-layer se time-reqency-domain schedling. Firstly, consider FUE s channel qality inormation (CQI), and determine the set o pre-served sers. Secondly, the schedler set corresponding weights according to the related parameters o each FUEs, sch as channel condition, throghpt, and so on [13]. Then, we need to make a priority qee, according to which to allocate time-domain spectrm in general. I there exist retransmission sers, they wold be the preerred, or ollowing the priority qee ntil the nmber o FUEs reaches a predetermined vale [14]. Finally, we combine these FUEs and corresponding resorce blocks (RBs) with pairs, and sort them in seqence nder a new priority weights and throgh the entire qee to allocate reqency-domain resorces. In the oter-layer, we consider sbchannel and power, and adopt the model o joint distribtion. Speciic steps will be detailed in the ollowing sections. The algorithm process: Fig.3 The lowchart o inner-layer schedling algorithm A User Classiication and Distinction In the initialization phase, all pre-served UEs within a cell are reqest to the macrobase station to allocate downlink resorce, and emtos within the macrocell also reqest resorces to the enb, in order to obtaining its reqency resorce that can be schedling independently. The enb irstly distingishes emtos rom UEs according to certain ields o reqest packets, then calclates path loss vale PLm or PL rom UEs to MeNB and UEs to FeNB by Hata-Okmra model ormla (2) PL lg() lg( Hb) Wh - ahm ( ) ( lg( Hb) lg( d)) (2) ere, is the operating reqency (150~1500MHz); Hb is the eective antenna height o the base station transmitter (30 ~ 200m); Hm is the eective antenna height o the mobile station receiver (1 ~ 10m); d is a distance (1~10km) rom the send antennas to receive antennas. MeNB and FeNB choose the minimm path loss vale o UEs as their sers. For near MUEs, assming the coordinates points o FeNB and MUE respectively is xi, y j and x j, y j, the radis o intererence MUE [6] is: 2 2 B1 B2 4 AC Ir (3) 2 4A DOI /IJSSST.a ISSN: x online, print

4 Where, A P P B1 2Px i 2 Pxj B2 2Pyi 2 P yj C P x P y P x P y j j i i P means the transmit power o emtocell, while represents the transmit power o mobile terminal. The intererence radis o MUE depends on the actal physical location o UE and emtocell and their transmit power. Measring the distance D between MUE and emtocell, when it meets D Ir, the MUE is set to be an "intererence with amily", and be added to a "blacklist", which is willing to be sent to macrocell. Macrobase station conts all disabled sbcarrier inormation according to the list, and then eedback to the emtocell to allocate additional resorces. B The Allocation O Sbchannel And Power For Macrocell, we conslt a method that described in [7], and directly se part o the power control algorithm. Deine the th MUE has a transmission power as ollows: MUE max M 0 MUE P min P, P 10 lg( RB ) L (4) Wherein, P 0 is MUE s special parameters, indicates the nmber o RBs assigned to MUE's, is P RB MUE special path loss compensation actor or the cell, we se 0.6 here, L is expressed as a downlink path loss measrement vale or the th MUE. Since emtocells are deployed indoors and the distance o its service sers-fues and emtocell is very short, so yo can assign its power averagely. C Compting the Cell Throghpt Calclate the inner and oter layers network throghpt. Inner C1, oter C2, the ormla is as ollows: m Nsc M B 1 i i1 sc m1 Nsc N B 1 j j 1 N sc N n1 T1 x lb 1 m (5) N M T2 y lb 1 n (6) sb-carriers in RBs set o Microcell and Femtocell; x i and y j are pre-served MUEs and FUEs to be allocated sb-carrier i and sb-carrier j; B indicates the system bandwidth; Nsc is the nmber o system sb-carrier; 1 M m M 1 and N n respectively represent the m1 N n1 average SINR o all MUEs and FUEs; m and n show the signal to intererence noise ratio o inner and oter layer o the network. IV.THE COMPARISON AND ANALYSIS In order to obtain the simlation reslts, MUEs and FUEs se two kinds o modlation (4QAM, 16QAM), and three kinds o modlation (4QAM, 16QAM, 64QAM). A. The simlation parameters are shown in Table II: B. The perormance analysis o algorithm simlation Intererence coordination algorithm based on power control, the algorithm o allocating resorce randomly and the algorithm proposed in this article will be compared and analyzed in this section. Take three aspects in cont: average throghpt o the system ser, the total intererence power o network, and packet loss rate o real-time services. TABLE II. THE SIMULATION PARAMETERS OF SYSTEM Parameters Setting / Description System architectre 7 cells,7*3 sectors The distance between the base stations 500 m Cell coverage radis 500 m/10 m User maximm transmit power 23 db Carrier reqency 2 GHz System bandwidth 20 MHz Total nmber o available RBs 100 The nmber o sb-carriers in each RB 12 Transer mode Open-loop airspace The nmber o antennas 2 2 Channel model SCM Shadow ading standard deviation 8 db/10 db System throghpt is T=T1+T2, the above two eqations are transormed rom Shannon ormla. Wherein N m sc and N sc respectively are the nmber o DOI /IJSSST.a ISSN: x online, print

5 three algorithms are enhancing smoothly. Wherein, the algorithm o allocating resorce randomly does not solve the sbstance intererence problem, ths the liting o the system spectrm eiciency is very limited. Intererence coordination algorithm based on power control ocses on solving complex intererence, leading to a case that signaling interaction is bond to aect the spectrm tilization. The proposed algorithm has broght an increase in average ser throghpt, thereby increasing the overall system throghpt. However, becase the system bandwidth is certain, the improving o the overall system throghpt will bring a rising in spectrm tilization. Fig.4 The average throghpt o system sers in dierent algorithms As can be seen rom Fig.4, with a increasing in the nmber o sers, the average ser throghpt o the system is decreased obviosly. Hierarchical schedling resorce the algorithm proposed in this article can achieve higher average ser throghpt o the system than the algorithm o allocating resorce randomly, becase the ormer not only can niormly schedle resorces and UEs, and choose the most appropriate sb-carriers, bt also redce the serios inter layer intererence in doble layer network. In the case o the nmber o MUEs gradally increasing, the throghpt o FUEs decreases more signiicantly. De to the increasing ser, in order to achieve a predetermined target rate, the system shold improve the transmit power o the ser, ths adding to the inter-cell intererence and inter-layer intererence, reslting in a loss o throghpt. Fig.6 The loss rate o real-time service in dierent algorithms In Fig.6, the new algorithm can sppress the packet loss rate o real-time service to minimize, and the maximm is only reached abot Its advantage is ar more than the other two algorithms. Frther, with the increasing nmber o MUEs, the total minimm spectral eiciency o emtos ollows, which is de to the macro layer can provide a better diversity gain. V. CONCLSION Fig.5 The system spectral eiciency in dierent algorithms It is known rom Fig.5, the system spectral eiciency o This paper takes Macro-Femto doble heterogeneos network as an example, and analyzes a variety o intererence mechanisms in the network, ths proposing a resorce allocation algorithm to solve the problem o inter-cell intererence. The algorithm takes the complex inter-cell intererence into accont very lly, and applies the way o hierarchical schedling - synergistic inhibition, redcing the system signaling overhead, operational complexity and eects o signaling delay on the overall perormance. In addition, the algorithm also ensres airness among FUEs and QoS between all MUEs. Simlation reslts show that the algorithm has signiicantly DOI /IJSSST.a ISSN: x online, print

6 improved in the average spectral eiciency and ser throghpt. In the tre, the continos development and demand or data services reqire a doble-layer, or even mlti-layer,with a low-power nodes (sch as Femto, Pico) network has a high-speed, low intererence characteristics, and this is the next step to be stdied. ACKNOWLEDGEMENTS This work was spported by the grants rom Jiangxi Provincial Edcation Development inancial aid or The visalization o pblic logistics service platorm constrction based on Internet o things [No.KJLD12032], National Natral Science Fondation o China [No ] and [No ], Key Projects in the Science and Technology Pillar Program o Jiangxi Province o China nder Grant No BBF60099, and The Open Project o Shanghai Key Laboratory o Trstworthy Compting (No. 07DZ ). REFERENCES [1] Waleed Ejaz,Najam l Hasan,Hyng Seok Kim, Distribted cooperative spectrm sensing in cognitive radio or ad hoc networks, J. Compter Commnications. (2013) [2] Selami Citci,Mrat Torlak,A Comparison o Energy Detectability Models or Cognitive Radios in Fading Environments, J. Wireless Personal Commnications. (2013) [3] Abdallah K. Farraj,Eman M. Hammad, Perormance o Primary Users in Spectrm Sharing Cognitive Radio Environment, J. Wireless Personal Commnications.(2013) [4] Pablo Serrano,Antonio de la Oliva,Pal Patras,Vincenzo Mancso,Albert Banchs,Greening wireless commnications: Stats and tre directions, J. Compter Commnications. (2012) [5] Olyemi Falade, Clive Parini,Larie Cthbert, Mtal Copling Eects on Pattern Diversity Antennas or MIMO Femtocells, J. Antennas and Propagation.(2010) [6] Kai Ma,Xin-Ping Gan, Jan Wang, Xiao-Min Wei, Mlti-ser Symmetric Cooperation Based on Nash Bargaining Soltion in Cooperative Relay Networks, J. Atomation & Compting.(2011) [7] Kai MA, Xinping GUAN, Bin ZHAO, Jan WANG,A cooperation strategy based on bargaining soltion in wireless sensor networks, J. Control Theory and Applications. (2011) [8] Wang Dian-hong,Fei OU,Yan Y-jie, Adaptive comptational resorce allocation or sensor networks,j. Harbin Institte o Technology. (2009) [9] Jiang Ming li and Li Zhi jing,optimal distribted resorce allocation in a wireless sensor network or control systems, J. Zhejiang University Science A(Science in Engineer). (2012) [10] Kimra and H. Shibasaki, Membrane-inspired qantm shled rog leaping algorithm or spectrm allocation, J. Systems Engineering and Electronics. (2012) [11] YUAN De-y,TENG Ying-lei,SONG Mei,WU Jn-, Stackelberg game or backhal resorce allocation in the two-tier LTE emtocell networks, J. China Universities o Posts and Telecommnications.(2014) [12] Wei FENG,Si-li FENG,Ye-ha DING,Xin HUANG,Cross-layer resorce allocation in wireless mlti-hop networks with otdated channel state inormation, J. Zhejiang University-Science C(Compters & Electronics). (2014) [13] Yee-Loo Foo,Perormance o Cooperative Spectrm Sensing Under Rician and Nakagami Fading, J. Wireless Personal Commnications.(2013) [14] Zaki Brahmi,Mohamed Mohsen Gammodi, Semantic shared space-based complex tasks allocation method or massive MAS[A]. Proceedings o nd IEEE International Conerence on Compter Science and Inormation Technology Vol.2[C]. 2009; Singapore [15] T. Zahir, K. Arshad, Y. Ko, and K. Moessner, A downlink power control scheme or intererence avoidance in emtocells," 7th International Wireless Commnications and Mobile Compting Conerence (IWCMC), 24-8 Jly; Caliornia, USA, DOI /IJSSST.a ISSN: x online, print