Fractional Frequency Reuse Schemes and Performance Evaluation for OFDMA Multi-hop Cellular Networks Yue Zhao, Xuming Fang, Xiaopeng Hu, Zhengguang Zhao, Yan Long Provincial Key Lab of Information Coding & Transmission Southwest Jiaotong University Chengdu 6131, China E-mail: yuezhao@foxmail.com, xmfang@swjtu.edu.cn 1 Abstract For the next generation wireless communication systems, inter-cell interference (ICI) is the primary cause of performance degradation in cell edge mobile stations of OFDMA multi-hop cellular networks with the frequency reuse pattern (1, 3, 1). This paper introduces a new fractional frequency reuse scheme to assign radio resources in OFDMA multi-hop networks to reduce inter-cell interference and maintain the sector frequency reuse factor at 1. Through the numerical analysis, we find that this scheme can yield higher SINR and improve the system capacity. Furthermore, the results also indicate that fractional frequency reuse relay system with local forwarding scheme can achieve higher throughput than local forwarding scheme in conventional relay system. Keywords- fractional frequency reuse; resource allocation; inter-cell interference (ICI); local forwarding I. INTRODUCTION Future cellular networks are tending broadband access to deploy relay stations (RSs) into the conventional single-hop cellular networks. Compared with base station (BS), RS has a lower cost and does not need to connect to backhaul network via cables. RS can also help improve the performance for mobile stations (MSs) near the edge of the cell and has the potential to solve the coverage problem for high data rates in wide areas [1]. Additionally, orthogonal frequency division multiple access (OFDMA) has been considered as a modulation and multiple access method for 4G wireless networks. Therefore, OFDMA based multi-hop cellular network is one of the most promising solutions for the next generation wireless system. With OFDMA instead of CDMA, the inter-cell interference (ICI) at the cell-edge areas becomes a critical problem in a multi-cell environment. Especially, ICI is more severe in multihop cellular networks with the frequency reuse pattern (1, 3, 1). According to this scheme each sector can use the total system frequency resources. Neglecting the inter-sector interference in the same cell, the reuse pattern of (1, 3, 1) has the potential of providing three times higher capacity than that with sector reuse factor 3. This is due to the fact that the same spectrum is reused three times in one cell. In respect of the conventional 1 This work was supported by NSFC under the Grant 677285. Figure 1. ICI is more severe in multi-hop cellular networks single-hop cellular networks, as shown in Fig.1, inter-cell interference increases when the MS moves to the cell edge. MS is affected by adjacent directional antennas of neighboring cells when MS is in the coverage of these antennas. This produces a low SINR to the MS at the cell edge accordingly. Interference is more complicated in the event of deploying RS. Up to now, some studies have been done for adaptive resource allocation of conventional OFDMA relay systems. In [2], it adopts a frequency reuse pattern (1, 3, 3) in cellular fixed relay networks. According to this scheme the available spectrum is divided among the three sectors in a way that each sector uses one third of the available resources. Sector reuse factor 3 significantly decreases ICI between neighboring cells due to the increased spatial reuse distance with the same frequency, but the spectral efficiency is low. An orthogonal resource allocation algorithm is used to eliminate interferences in [3]. Although this method works quite well in a non-saturated network, it looses its effectiveness as the network load increases. Moreover, the available subcarriers allocated for each MS by BS or RS according to the certain priority and the location of the MS, this increases complexity of BS and RS devices. This paper focus on the downlink of OFDMA multi-hop cellular networks working on time division duplex (TDD)
mode, while introduces a new fractional frequency reuse scheme to assign radio resources effectively in multi-hop networks. The scheme is examined to significantly reduce ICI and maintain the sector frequency reuse factor at 1. Simulations are implemented for the performance analysis of this scheme. The present paper is organized as follows. In section II, the architecture of OFDMA multi-hop cellular networks working on TDD mode is introduced and the special frame structure is described in detail. Section III provides a path selection method and the local forwarding technology in multi-hop networks. Section IV gives out the simulation environment, the numerical results and some discussions. Section V concludes this paper. II. SYSTEM CONFIGURATION AND FRAME STRUCTURE A. Network Architecture The concept for the new system architecture is to sectorize cells and put six fixed relays at the cell edge. Considering one sector with two relays, they are 2/3 cell radius away from the BS. A mobile station can communicate directly with BS over a single-hop, or otherwise, establish a two-hop link via an RS. We define the cell edge area as the relay coverage. The role of the fixed relays is to cover the mobile stations at cell edges. Frequency reuse factor 1 at the sector of the cell is to maximize the network spectral efficiency. One time-frequency resource unit can only be assigned to one transmission link in a sector during the scheduling period. So no intra-sector interference would be considered. However, if time-frequency resource units can be used sufficiently in each sector, inter-cell interference inevitably occurs for edge MSs that can possibly receive interference signals both BS and RS of neighboring cells using the same frequency at the same time. MSs at the cell edge are far away from BS than RSs of neighboring cells. Therefore, the local MS receives a relatively lower interference signal power from BS of neighboring cell as shown in Fig.2. Consequently, fractional frequency bands are used for the RS- MS communication at the relay coverage area, and total frequency bands are used for the BS-MS and BS-RS communication in each sector. The details of the proposed system architecture are described in Fig.3. Each cell is physically divided into three sectors: X, Y, and Z. Relays of each sector can use F1, F2, and F3 frequency bands respectively to RS-MS links communication. With this configuration, the full load frequency reuse is maintained for sector MSs to maximize spectral efficiency and different sets of fractional frequency reuse is implemented for edge MSs to assure edge MSs connection quality and throughput. B. Frame Structure and Resources Allocation Algorithm In Fig.4, a frame structure is presented as a possible solution to enable fractional frequency reuse in an OFDMA multi-hop cellular network working on TDD mode. The frame consists of downlink frame and uplink frame. Each frame contains N slots plus two duplex guard times, in which M slots are for downlink and the rest for uplink. Each frame consists of Figure 2. A possible scenario to interfere with MS at the cell edge Figure 3. Fractional frequency reuse relay system L subchannels, and is also divided into three frequency bands F1, F2, and F3 equally in frequency domain. Each slot is divided into 2 equal time-frequency resource units, namely, OFDMA symbols, and a guard time gap is inserted into each OFDMA symbol. Considering the characteristic of RSs which can not transmit and receive data simultaneously, the index of 1 to m slots are dedicated to reception of MSs from BS or RSs and the index m+1 to M slots are dedicated to BS transmission towards both RSs and MSs. As shown in Fig.4, F1 frequency bands are used for RS-MS communication from 1 to m slots in downlink subframe. Similarly, the other frequency bands are used for BS-MS communication. If MSs are located in sector Y or Z, F2 or F3 frequency bands are used for RS-MS communication from 1 to m slots respectively.
Figure 6. Illustration of local forwarding mode Figure 4. Frame structure (X sector) of fractional frequency reuse scheme III. PATH SELECTION AND LOCAL FORWARDING SCHEME A. Relay Selection In Fig.5, each MS is linked with either the direct BS-MS link or the relayed RS-MS link. In a manner that maximizes the achievable throughput of the cell, data will be delivered by two-hop transmission if forwarding the packets by relay can achieve a higher effective transmission rate. Referring to Fig.5, assume that R bm is the transmission rate achieved in the link between BS and MS; R br is that between BS and RS; and R rm is that between RS and MS. Consider a packet with size P. With two transmissions phases, the harmonic mean of transmission rate R brm for a two-hop communication can be given as R brm P P 1 1 = = = t P P + + R R R R BS RS MS br rm br rm Figure 5. Path Selection where t BS-RS-MS is the total two-phase transmission time. Therefore, the optimal routing path is determined as 1 (1) r = arg max( R, R ) (2) brm bm B. Local Forwarding Mode Local forwarding means MSs belonged to one RS communicate with each other while the data is forwarded only by the RS. Consider a simple local forwarding mode shown in Fig.6, data between MS1 and MS2 is forwarded by RS. Accordingly, BS will not receive or transmit the data but perform control signaling if necessary. It is noticed that local area communication plays a more and more important role with various usage scenarios in daily life. Local forwarding not only decreases the number of hops for communication, but also removes the burden from BS s air interface, which will result in increment of cell capacity from system point of view. The introduction of local forwarding involves problems related to signaling overhead such as user registration and handover operation, but its advantage can offset them absolutely while enhancing the system throughput considerably [5]. IV. SIMULATION RESULTS A. Simulation Environments and Parameters In wireless communication environments, radio channels are randomly changing and are typically analyzed in a statistical way. First of all, the received signal power is inversely proportional to the distance between the transmitter and the receiver. Also, there exist variations of shadow fading and multipath fading [6]. Taking all the above into consideration, the received power can be approximately written as 1 Pr = Pt( Lr) XY δ (3) where P t is the transmitted power and L r represents the path loss. Xσ is a Gaussian distributed random variable with zero mean and standard deviation σ,and Y represents the multipath fading. The received signal to noise ratio (SNR) is defined as SNR = (4) N where N is the noise power. Hence, the signal to interference plus noise ratio (SINR) can be calculated as SINR P r = = = N + Pi ISF i i 1+ 1+ N N P r Pr N SNR P ISF P ISF (5)
where P i is the interference power and ISF is the interference suppression factor [7]. Its value ranges form to 1. The parameter ISF indicates the extent of how much an interference cancellation scheme could reduce the interference. We consider the seven modulation coding schemes as shown in Table I [8]. It lists the data rate and the required SINR. With our path selection algorithm, each MS is first attached to either BS or the RS through calculation of the received SNR. The data rate is renewed because SNR is replaced by SINR. TABLE I. THE MODE OF ADAPTIVE MODULATION AND CODING MCS Modulation Data bits Data bits Data /symbol /slot rate/slot SNR 1 QPSK(1/2) 1 48 233.24 5. 2 QPSK(3/4) 1.5 72 349.86 8. 3 16QAM(1/2) 2 96 466.48 1.5 4 16QAM(3/4) 3 144 699.71 14. 5 64QAM(2/3) 4 192 932.95 18. 6 64QAM(3/4) 4.5 216 149.57 2. Prob(SINR Abscissa) 1.9.8.7.6.5.4.3.2.1 TABLE II. SIMULATION PARAMETERS Parameter name Value Number of 3-sector cells 19 Cell radius 175m Carrier frequency 2.5GHz Channel bandwidth 1MHz FFT size 124 Frame duration 5ms Useful symbol time 91.4us Guard time 11.4us OFDMA symbol duration 12.9us BS transmit power 46dBm RS transmit power 38dBm BS antenna height 25m RS antenna height 15m MS antenna height 1.5m Path-loss type Terrain Type C Multipath fading model SUI-2 channel Thermal noise spectral density -174dBm/Hz Interference suppression factor.5 The CDF of SINR for MSs in the center cell Traditional Relayless System with ISF =.2 with ISF=.2 with ISF=.2 Traditional Relayless System with ISF =.5 with ISF=.5 with ISF=.5 Traditional Relayless System with ISF =.8 with ISF=.8 with ISF=.8 5 1 15 2 25 3 SINR [db] Figure 7. The CDF of SINR for MSs in the center cell with respect to interference suppression factor 6 5 4 3 2 1 Traditional Relayless System Center Cell Throughput 1 2 3 4 5 6 7 8 9 1 User Number of the Center Cell Figure 8. Comparison of the center cell throughput with different systems The location of each MS is assumed to be random. Without losing generality, data streaming is assumed as the only source traffic, and full buffer service model is used to allow MSs to have enough packets to transmit at all times. It is assumed that the ICI of any MS in the center cell affected by the cells from the third tier is very small and can be ignored. Therefore, only the measurements of MSs at the center cell are plotted in order to show performance improvement by fractional frequency reuse scheme for OFDMA multi-hop cellular networks. B. Numerical Results and Discussions The simulation parameters are listed in Table II [9], and the channel models are taken from [6]. Fig. 7 presents the cumulative distribution function (CDF) of the received SINR for the center cell MSs to show the improvements of performance in the traditional system, conventional relay system, and fractional frequency reuse relay system respectively. There are 5 active MSs in each cell. From the results, the fractional frequency reuse relay system achieves the best SINR performances. For example, at ISF=.5, the probability of the SINR above 5dB within a cell is about 96%. In other words, only 4% of MSs in a cell cannot get the service when the system uses the fractional frequency reuse scheme. The conventional relay system has higher SINR than the traditional system. Thus forwarding data by relay can improve the received SINR. Fig. 8 illustrates the proposed fractional frequency reuse relay system also achieves the highest throughput due to the best SINR performances. Compared with the traditional systems, the conventional relay system has better throughput performance. Furthermore, the more throughput gain can be achieved by introducing the fractional frequency reuse scheme. Specifically, when there are 7 MSs in a cell, the center cell throughput of the traditional system, conventional relay system, and fractional frequency reuse relay system are approximately 34, 39 and 42, respectively. It means that the amounts of additional throughput are approximately 8% and 23% compared to the conventional relay system and the traditional system. From the results, when the number of MSs using relays increases, the effect of fractional frequency reuse scheme is
more significant than the case when the number of MSs using relays is small. 4 38 36 34 32 3 28 26 24 22 Center Cell Throughput with Probability of Local Forwarding from to 8% Traditional Relayless System with Local forwarding Scheme with Local forwarding Scheme 2 1 2 3 4 5 6 7 8 Probability of Local Forwarding(%) Figure 9. Comparison of the center cell throughput with probability of local forwarding from 1% to 8% 7 6 5 4 3 2 1 Center Cell Throughput with Local Forwarding Probability P=.3 with Local Forwarding Probability P=.5 with Local Forwarding Probability P=.3 with Local Forwarding Probability P=.5 1 2 3 4 5 6 7 8 9 1 User Number of the Center Cell Figure 1. Comparison of the center cell throughput for two different systems with local forwarding scheme Fig. 9, where the number of active MSs per cell is fixed to 5, shows the center cell throughput enhancements with local forwarding scheme. It can be observed that downlink throughput increases with probability of local forwarding from to 8%. For instance, when p is 5%, almost above 1% throughput gain can be achieved. And the gain is especially rich when most of the transmission happens inside the same relay coverage area. A noteworthy fact is that the fractional frequency reuse relay system can improve the throughput efficiently. It is because that the balance of two-hop links is considered, and the fractional frequency reuse scheme can make full use of the savable resources with local forwarding scheme, whereas the data rate of the RS-MS link is severely limited in the conventional relay system because of ICI. Hence, with the local forwarding scheme, the former system has higher throughput than the latter one In Fig.1, performance comparison between two kinds of relay systems with local forwarding scheme versus MS number is presented. As shown in the figure, the latter has lower throughput than the former with the number of MSs per cell range from to 1. In summary, the OFDMA multi-hop cellular networks only achieve its best throughput when the proposed fractional frequency reuse scheme is employed cooperatively. V. CONCLUSIONS The fractional frequency reuse scheme for OFDMA multihop cellular networks has been studied in this paper. Based on the proposed resource allocation algorithm and fame structure in TDD mode, the scheme has been proved to be achievable. Numerical results show that the ICI can be mitigated effectively through this scheme. Moreover, the proposed scheme with local forwarding scheme can achieve more system throughput gain compared with local forwarding scheme in conventional relay system. References [1] Jaewon Cho, and Zygmunt J. Haas, On the throughput enhancement of the downstream channel in cellular radio networks through multihop relaying, IEEE Journal on Selected Areas in Communications, vol. 22, no. 7, September 24, pp.126 1219. [2] O. Mubarek, H. Yanikomeroglu, and S. Periyalwar, Dynamic Frequency Hopping in Cellular Fixed Relay Networks, IEEE Vehicular Technology Conference, VTC 25, pp.3112 3116. [3] Woonsik Lee, Minh-Viet Nguyen, and Jeonghan Jeong, An orthogonal resource allocation algorithm to improve the performance of OFDMAbased cellular wireless systems using relays, IEEE Consumer Communications and Networking Conference, CCNC 28, pp.917 921. [4] Li-Chun Wang, Wen-Shan Su, Jane-Hwa Huang and et al, Optimal relay location in multi-hop cellular systems, IEEE Wireless Communications and Networking Conference, WCNC 28, pp.136 131. [5] Yang Liu, Yuqin Chen, and et al, Comments on Data Forwarding in Project 82.16m System Description Document, IEEE C82.16m- 8/131, October 28. [6] G. Senarath, W. Tong, P. Zhu, and et al, Multi-hop relay system evaluation methodology (channel model and performance metric), IEEE 82.16j-6/13r3, 27. [7] Huining Hu and B.Eng, Performance Analysis of Cellular Networks with Digital Fixed Relays, A thesis of Master Degree, Ottawa-Carleton Institute for Electrical and Computer Engineering, Carleton University, September 23. [8] The Relay Task Group of IEEE 82.16, IEEE Standard for Local and metropolitan area networks Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems Amendment 2:Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands and Corrigendum 1. IEEE Std 82.16e - 25 and IEEE Std 82.16-24/Cor 1-25. [9] Mobile wimax part I a technical overview and performance evaluation, WiMAX Forum, March 26. [1] Roshni Srinivasan,Jeff Zhuang, and et al, IEEE 82.16m Evaluation Methodology Document (EMD), IEEE 82.16m-8/4r3, October 28.