Interference Manageent in LTE Fetocell Systes Using Fractional Frequency Reuse Poongup Lee and Jitae Shin School of Inforation and Counication Engineering Sungyunwan University, Suwon, 440-746, Korea {poongup, jtshin}@su.edu Abstract Recently, Long Ter Evolution (LTE) has developed a fetocell for indoor coverage extension. However, interference proble between the fetocell and the acrocell should be solved in advance. In this paper, we propose an interference anageent schee in the LTE fetocell systes using Fractional Frequency Reuse (FFR). Under the acrocell allocates frequency band using the FFR, the fetocell chooses sub-bands which are not used in the acrocell sub-area to avoid interference. Siulation results show that proposed schee enhances the throughput and reduces the outage probability in overall networ, especially for the cell edge users. Keywords Fetocell, Interference anageent, Fractional Frequency Reuse (FFR), Long Ter Evolution (LTE), Throughput I. ITRODUCTIO Long Ter Evolution (LTE) is one of the ost proising candidates for next generation counication standard. The LTE is technologically based on Orthogonal Frequency Division Multiple Access (OFDMA) to achieve higher data rates and enhanced spectral efficiency. The frequency and tie resources are allocated to users in orthogonal anner. A ey feature is Frequency Reuse Factor (FRF) 1 for axial resource use. However, sae sub-carriers are used by different users aong adjacent cells, Co-Channel Interference (CCI) proble occurs, especially for cell edge users. Appropriate inter-cell interference coordination technique should be required to enhance the syste capacity [1]. Recently, the LTE has developed a fetocell for indoor coverage extension. The fetocell is defined as very sall size, low-power hoe base station that wors in the licensed frequency bands, and it is connected to broadband Internet bachaul [2]. The fetocell brings various benefits to both consuers and operators, such as enlarged indoor coverage, enhanced syste capacity, Quality of Service (QoS), and reduced capital and operation expense. Due to increasing indoor phone calls and data services, but insufficient acrocell coverage, the fetocell could be an attractive solution. However, interference proble between the fetocell and the acrocell should be solved in advance, because the fetocell is deployed over the existing acro networ, and it uses sae spectru with the acrocell. Dedicated channel approach is one of the easiest ways to solve this proble, but the frequency resources are not utilized effectively. Co-channel odel is suitable for practical deployent, but the CCI proble should be solved [3]. The fetocell is often turned on and off, and installed at unnown location, because it is anaged by a personal custoer, not by a networ operator. Therefore, any paraeters should be autoatically adjusted to avoid the CCI. Fractional Frequency Reuse (FFR) is discussed in the OFDMA based networ such as the LTE, to overcoe the CCI probles [4-6]. In the FFR, whole frequency band is divided into several sub-bands, and each sub-band is differently assigned to center zone and edge region of the cell. While reuse factor of the center zone is one, the edge region adopts bigger reuse factor. As a result, intra-cell interference is reoved, and inter-cell interference is substantially reduced. At the sae tie the syste throughput is enhanced. In this paper, we propose an interference anageent schee in the LTE fetocell systes using the FFR. Under the acrocell allocates frequency band using the FFR, the fetocell chooses sub-bands which are not used in the acrocell sub-area to avoid interference. The reaining part of this paper is coposed as follows. Related wor is explained in Section II. Section III describes the proposed interference anageent schee. The siulation results are shown in section IV. Finally, we conclude our paper in section V. II. RELATED WORK Interference anageent issues for the fetocell systes have been actively discussed both in the LTE and in the Worldwide Interoperability for Microwave Access (WiMAX) [6]. Spectru partitioning and Frequency ALOHA (F-ALOHA) schee was proposed [7]. It reoves interference between acrocells and fetocells by assigning orthogonal spectru. Aong the fetocells, each fetocell uses sub-channels in rando anner. Optial spectru portion for the fetocells
can be deterined. However, this schee is based on the dedicated channel approach. Although the spectru is utilized adaptively, the full bands are not serviced for the acrocell. Dynaic Frequency Planning (DFP) algorith was another proposal for interference avoidance [8]. After dividing the OFDMA networ into several sectors, it estiates the aount of sub-channels considering user bandwidth deand in each sectors. Interference aong sectors are calculated when the sectors transit the sae frequency. Optiization function is run to iniize the overall networ interference. However, this schee follows a centralized networ structure, which is not appropriate for fetocell anaged by a personal owner. Several heuristic algoriths for frequency assignent were investigated [9]. The recoended one is Least Interference Power (LIP) algorith, where a powered up feto base station chooses a frequency segent that iniizes the interference level. In Turn-on Orders algorith, frequency allocation is conducted according to the order of the fetocell turned on. Optiu Pattern (OPTM) and Rando Schee are described for coparison. However, the fetocells are deployed in the for of rectangular atrix, not in the rando anner. Also, interference between the acrocell and the fetocell is not evaluated. Isolated and Coupled odel considering user location was discussed for OFDMA fetocells [10]. In the isolated odel, acro and feto users are allocated different resources split by tie and frequency slots. The coupled odel reuses acro resources for fetocells which are located in the cell boundary, while the center zone fetocells use orthogonal resources fro the acrocell. However, these schees do not apply the FFR concept for the OFDMA syste. Frequency reuse and pilot sensing schee was proposed to reduce the CCI [11]. After applying the FRF of 3 or above to the acrocells, the fetocells use the reaining frequency sub-bands. For exaple, if the reuse factor is three and the acrocell uses sub-band I aong the three sub-bands I, II, and III, the fetocell chooses sub-band II and III. However, the acrocell throughput is reduced, even though the overall capacity is enhanced. Also the reuse factor in acrocell is against the OFDMA based networ such as the LTE and the WiMAX, where the target of the reuse factor is one. The FFR is one of the solutions to reduce inter-cell interference in acrocell syste, especially for the cell edge users. Also it is helpful to achieve the reuse factor of one. Under this condition, the interference fro the fetocell deployent should be iniized for the acro users. So, we focus on the interference anageent between the acrocell and the fetocell using the FFR. III. PROPOSED SCHEME I FEMTOCELL A. Illustration of the Schee We allocate the frequency sub-bands into acrocell and fetocell as depicted in Figure 1. The acrocell coverage is divided into center zone and edge region including three sectors per each, denoted by C1, C2, C3, and E1, E2, E3. The whole frequency band is partitioned into two parts, and one part of the is classified into three pieces again, totally denoted by A, B, C, and D. For acrocell, different frequency sub-band is allocated to the each acrocell sub-area according to the FFR. The reuse factor of one is applied in the center zone, while the edge region adopts the factor of three. The sub-band A is used in the center zone (C1, C2, and C3), and sub-band B, C, and D is applied in the E1, E2, and E3 regions, respectively. Under this circustance, the fetocell chooses sub-bands which are not used in the acrocell sub-area. Especially when the fetocell is located in the center zone, the fetocell additionally excludes a sub-band which is used by acrocell in the edge region of current sector. For exaple, when a fetocell is located in region E1, it uses sub-band A, C, and D, while the acrocell uses sub-band B. If a fetocell is positioned in zone C1, sub-band C and D is applied. The fetocell avoids sub-band A which is used by acrocell in zone C1. Also it avoids sub-band B which is used by acrocell in region E1, because the received signal power of sub-band B is relatively strong for that fetocell. Due to the characteristics of the OFDMA, the acrocell is interfered by inter-cell, and that interference is further itigated by the FFR. The fetocell uses different sub-band to avoid interference fro the acrocell. The sub-band is reused in the acrocell coverage as uch as possible, because transit power of the fetocell is very sall. Therefore, the interference between the acrocell and fetocell is greatly avoided. Also ore sub-carriers are allocated to fetocell which is located in the edge region, in order to iprove the perforance of the edge users. E2 Macro: C Feto: A,B,D E1 Macro: B Feto: A,C,D C1 Feto: C,D Macro: A C2 Feto: B,D C3 Feto: B,C E3 Macro: D Feto: A,B,C Figure 1. The proposed interference anageent schee using FFR B. Perforance Forulation We forulate downlin Signal to Interference and oise Ratio (SIR) and syste throughput. The overall networ is coposed of two-tier 19 acrocells, and any fetocells are randoly deployed over the acrocells. A acro user is interfered fro neighboring 18 acrocells and all of the adjacent fetocells. Due to sall transit power, only fetocells which are located in the 1-tier acrocell area give interference to acro user. The received SIR of a acro user on sub-carrier can be expressed as A B C D
SIR, P G M,, M, = Δ f + P G + P G 0 M ',, M ', F,, F, M ' F where, P M, and P M, is transit power of serving acrocell M and neighboring acrocell M on sub-carrier, respectively. G,M, is channel gain between acro user and serving acrocell M on sub-carrier. Channel gain fro neighboring acrocells are denoted by G,M,. Siilarly, P F, is transit power of neighboring fetocell F on sub-carrier. G,F, is channel gain between acro user and neighboring fetocell F on sub-carrier. 0 is white noise spectral density, and Δf is sub-carrier spacing. In case of a feto user, it is interfered fro all 19 acrocells and adjacent fetocells. The received SIR of a feto user f on sub-carrier can be siilarly given by P G F, f, F, SIR = (2) f, Δ f + P G + P G 0 M, f, M, F', f, F', M F' The channel gain G is doinantly affected by path loss, which is different for outdoor and indoor. The path loss for outdoor is odeled as PL outdoor = 28+35log 10 (d) db, where d is the distance fro a base station to a user in eters. Otherwise, indoor odel is PL indoor = 38.5+20log 10 (d)+l walls db, where L walls is 7, 10, and 15 db for light internal, internal, and external walls, respectively [3]. So, the channel gain G can be expressed as PL /10 G = 10 (3) The practical capacity of acro user on sub-carrier can be given by [4] C =Δf log (1 + α SIR ) (4), 2, where, α is a constant for target Bit Error Rate (BER), and defined by α = 1.5 / ln(5 BER). Here, we set BER to 10-6. The overall throughput of serving acrocell M can be expressed as = β C (5) TM,, where, β, notifies the sub-carrier assignent for acro users. When β, = 1, the sub-carrier is assigned to acro user. Otherwise, β, = 0. Fro the characteristics of the OFDMA syste, each sub-carrier is allocated only one acro user in a acrocell in every tie slot. This iplies that β = 1, where = 1, is the nuber of acro users in a acrocell. Siilar expression for feto users related to the practical capacity and the overall throughput is possible except f β f = 1 f, (1) = 3. f is the nuber of feto users in a acrocell. This iplies that the proposed schee reuses the full frequency band three ties in a acrocell. C. Operational Algorith The acro users are allocated the sub-carriers according to the FFR as shown in Figure 1. The center zone and the three sectored edge region use the separated frequency subband. Under this condition, a fetocell senses the neighboring acrocell signals, when it turns on. The Received Signal Strength Indication (RSSI) values are copared for each subband A, B, C, and D. When the RSSI of sub-band A is strong, the fetocell is located in the center zone. In addition, if the sub-band B signal is strong, the fetocell location is zone C1. The fetocell chooses sub-carriers fro the sub-band C and D. It excludes the sub-band A and B, whose signal power is strong because these are used by the acro users. It is siilar for zone C2 and C3. On the other hand, if the RSSI of sub-band A is wea, the fetocell is positioned in the edge region. Moreover, if the subband B signal is strong, the fetocell position is region E1. The sub-carriers fro the sub-band A, C, and D are selected by the fetocell. The sub-band B is excluded because its RSSI is high. Siilar schee is possible for the region E2 and E3. IV. SIMULATIO RESULTS A. Scenarios and Environents We evaluate the proposed schee in ters of throughput and outage probability using SIR threshold in the range of 0 to 30 db. We also concentrate on the perforance of the cell edge users. The outage is deterined when the SIR level of a sub-carrier do not exceed the designated threshold. The outage probability p out is given by u δ SIR u, u, p = (6) out β SIR u u, u, where, δ u, indicates failed sub-carrier assignent for user u on sub-carrier. If δ u, =1, the SIR of that sub-carrier is under the SIR threshold (SIR u, < SIR threshold ). So, the ratio between the nuber of sub-carriers under the SIR threshold and the nuber of total sub-carriers is the outage probability. The proposed schee is copared with several coparison schees as follows. In schee, FFR is applied to the acro users, and feto users are randoly assigned fro full frequency band. On the other hand, FFR is not adopted for acro users in o schee. The aount of total available sub-carriers for feto users is three ties of the full band, as equal to the proposed schee. These features are suarized in Table 1. The Aount colun iplies the aount of used sub-carriers denoted by the sub-carrier assignent paraeter β. Table 1. Suary of and Coparison Schees Schees Macro user Feto user Frequency Aount Frequency Aount FFR Divide center and edges 1 (Figure 1) 3 FFR Rando o Rando Rando ote: Aount colun iplies value of and = 1, f β f = 1 f, for acro and feto users, respectively.
Table 2. Siulation Paraeters Paraeters Values Macro Feto uber of cells 19 (two-tier) 30-180/acro Cell coverage 280 30 BS transit power FFR: 15, 22 W w/o FFR: 20 W 20 W uber of users in one acro cell coverage 180 180 Channel Bandwidth 5 MHz FFT size 512 uber of occupied subcarriers 300 Sub-carrier spacing 15 Hz White noise spectral density -174 db/hz Size of center zone 0.63 of acro coverage PL outdoor = 28+35log 10 (d) PL Channel odel indoor = 38.5+20log 10 (d)+l walls L (Path loss, PL) walls = 7 db, if d in ( 0,10] L walls = 10 db, if d in (10,20] L walls = 15 db, if d in (20,30] The siulation paraeters are listed in Table 2. The overall networ is coposed of two-tier 19 acrocells, and fetocells are randoly deployed over the acrocells. All the base stations are operated by the OFDMA technology. We vary the nuber of fetocells fro 30 to 180 in one acrocell coverage to change siulation environent. The acro and feto users are randoly distributed in the overall networ. The acro and feto users are randoly allocated the available sub-carriers fro the designated range in each schee. When the FFR is applied to acro user, a half of the full band is assigned for center zone, the other for edge region. An oni-directional antenna and three sector antennas are installed at a acro base station, for center zone with transit power of 15 W and for edge region with 22 W, respectively. On the other hand, if the FFR is not used by acro user, the transit power of a acro base station is 20 W. All the fetocells use 20 W as the transit power. The different channel odel is used for indoor and outdoor, where the path loss is doinant factor. Then, we find out the downlin received SIR values for each user and each subcarrier. Using this value, the throughput and the outage probability are calculated via users located in the central serving acrocell of 19 cells. B. uerical Results and Coparison Figure 2 shows the throughput of acro users located in the acrocell coverage as the nuber of fetocells varies. In proposed schee, the feto users can get the sub-carriers which are not used by acro users at each location. So, the interference between acro users and fetocells is greatly avoided. The fetocells affect to acro users less than the coparison schees. The gap of throughput between the proposed and the other schees is bigger when ore fetocells are deployed. In Figure 3, it describes the throughput of total users located at the edge region only. In the OFDMA cellular networ, the perforance of the edge region is poor due to the inter-cell interference. Allocating the ore sub-carriers to fetocells which are located in the edge region, the throughput of the edge region is iproved. The gains are 27% and 47% in average, when copared with the and o schees, respectively. Figure 4 depicts the outage probability of total users according to the SIR threshold, when 30 fetocells are deployed in the acrocell coverage. In a given SIR threshold, the proposed schee indicates lower outage probability than other schees. Also we decrease the outage probability ore at low SIR threshold. This iplies that the proposed schee effectively supports ore users, even though the interference is severe. As shown in these results, our proposed schee enhances the overall throughput and reduces the outage probability, especially for the cell edge users. Throughput (Mbps) 12 11 10 9 8 7 6 5 4 Throughput of Macro Users (Center+Edge) o 3 30 60 90 120 150 180 uber of Fetocells Figure 2. Throughput of acro users located in center zone and edge region as the nuber of fetocells varies Throughput (Mbps) 110 100 90 80 70 60 50 40 Throughput of Macro+Feto Users (Edge) 30 20 o 10 30 60 90 120 150 180 uber of Fetocells Figure 3. Throughput of acro and feto users located only in the edge region as the nuber of fetocells varies
Outage Probability 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 Outage Prob. of Macro+Feto Users (Center+Edge) 0.2 0.1 o 0 0 5 10 15 20 25 30 SIR Threshold (db) Figure 4. Outage probability of acro and feto users according to SIR threshold varies, when 30 fetocells are deployed V. COCLUSIOS In this paper, we propose an interference anageent schee in the LTE fetocell systes using the FFR. The proposed schee enhances throughput and reduces outage probability, especially for the cell edge users. It is beneficial for the OFDMA networ such as the LTE, where the FFR is applied. REFERECES [1] G. Boudreau et al., "Interference Coordination and Cancellation for 4G etwors," IEEE Coun. Mag., vol. 47, no. 4, pp. 74-81, Apr. 2009. [2] V. Chandrasehar, J. Andrews, "Fetocell etwos: A Survey," IEEE Coun. Mag., vol. 46, no. 9, pp. 59-67, Sept. 2008. [3] L. Ho, H. Claussen, "Effects of User-Deployed, Co-Channel Fetocells on the Call Drop Probability in a Residential Scenario," IEEE International Syposiu on Personal, Indoor and Mobile Radio Counications (PIMRC), Sept. 2007. [4] H. Lei, L. Zhang, X. Zhang, and D. Yang, "A ovel Multi-cell OFDMA Syste Structure Using Fractional Frequency Reuse," IEEE International Syposiu on Personal, Indoor and Mobile Radio Counications (PIMRC), Sept. 2007. [5] M. Assaad, "Optial Fractional Frequency Reuse (FFR) in Multicellular OFDMA Syste," IEEE Vehicular Technology Conference (VTC), Sept. 2008. [6] S. Yeh et al., "WiMAX Fetocells: A Perspective on etwor Architecture, Capacity, and Coverage," IEEE Coun. Mag., vol. 46, no. 10, pp. 58-65, Oct. 2008. [7] V. Chandrasehar, J. Andrews, "Spectru Allocation in Two-Tier etwors," IEEE Asiloar Conference on Signals, Systes and Coputers, Oct. 2008. [8] D. Lopez-Perez et al., "Interference Avoidance and Dynaic Frequency Planning for WiMAX Fetocells etwors," IEEE International Conference on Counication Systes (ICCS), ov. 2008. [9] H. Zeng, C. Zhu, and W. Chen, "Syste Perforace of Self-Organizing etwor Algorith in WiMAX Fetocells," ACM International Conference on Wireless Internet (WICO), ov. 2008. [10] K. Sundaresan, S. Rangarajan, "Efficient Resource Manageent in OFDMA Feto Cells," ACM International Syposiu on Mobile Ad Hoc etworing and Coputing (MobiHoc), May 2009. [11] T. Ki, T. Lee, "Throughput Enhanceent of Macro and Feto etwors By Frequency Reuse and Pilot Sensing," IEEE International Perforance, Coputing and Counications Conference (IPCCC), Dec. 2008.