A Hybrid Inter-cell Interference Mitigation Scheme for OFDMA System Zhongnian Li1, 2, Yafeng Wang1, and Dacheng Yang1 Wireless Theories and Technologies Lab (WT&T) Beijing University of Posts and Telecommunications, 100876, Beijing, P. R. China1 Electronics and information college, Yangtze University, 434023, Jingzhou, P. R. China2 Email:
[email protected] Frequency hopping scheme is an interference randomization technique. It changes the transmit frequency every time slot to obtain the different experienced channel frequency response and different interference, which can get frequency diversity and whiten the interference of the single UE to improve the link performance. But the frequency hopping scheme is not highly compatible with the frequency-selective scheduling. The latter will schedule the appropriate UE on the time-frequency chunk based on the UE’s reported SINR to get high system throughput.
Abstract—In OFDMA cellular system, the inter-cell interference is the main interference of the system, which greatly degrades the system performance. In order to enhance cell average throughput and the cell-edge user’s peak bit rate, a hybrid inter-cell interference mitigation scheme is proposed. The hybrid scheme combines the fractional frequency reuse scheme and the frequency hopping scheme, in which cell-edge user transmits data through frequency hopping method in a restricted subcarrier group and inner-cell user is scheduled by frequency-selective scheduler in the left subcarriers. The hybrid scheme can improve the peak bit rate of cell-edge user while getting high cell average throughput. According to the different hopping pattern, the hybrid scheme has two manners: mirroring frequency hopping with fractional frequency reuse and multi-cell joint frequency hopping. The simulation result demonstrates that the cell-edge peak bit rate increases 7% with the proposed hybrid method.
For a high performance cellular system, the peak bit rate of cell-edge UE and system throughput are two important criteria, in which the former is more important. A UE, no matter where it is, i.e., regardless in the inner cell or at the cell boundary, has same date rate request. So how to enhance the cell-edge UE’s peak rate while keeping the high system throughput is a problem.
Keywords: Interference Mitigation; Fractional frequency resue; Frequency hopping;
In this paper, a hybrid inter-cell interference Mitigation scheme is proposed, which combines the fractional frequency reuse and frequency hopping scheme. For the UE at the cell boundary, the hybrid scheme not only can reduce the experienced interference from the neighbor cell using fractional frequency reuse, but also enhance the link performance using frequency hopping to improve the peak bit rate. For the UE in inner cell, the interference is lower, the hybrid scheme uses frequency-selective scheduling to achieve high cell throughput.
I. INTRODUCTION Recently, orthogonal frequency division multiplexing (OFDM) has received a great deal of attention for radio transmission technology of the next generation cellular systems, for example: 3GPP Long Time Evolution (LTE), IEEE 802.16, Ultra Mobile Broadband (UMB), due to its many advantages such as high spectral efficiency and anti multi-path capacity. However, in order to get high spectral efficiency, the frequency reuse factor (FRF) of OFDMA cellular system is set as 1, same with CDMA cellular system. The user equipment (UE), which is assigned the same time-frequency resource chunk with other UE in different neighbor cells, will experience a large inter-cell interference (ICI), especially at the cell boundary [1]. It will greatly degrade the system performance, especially for the cell-edge UE. Techniques for mitigating ICI can be mainly classified as interference coordination, interference randomization and interference cancellation [2].
The reminder of this paper is organized as follows. The two interference mitigation schemes are described in Section II. Section III gives the hybrid scheme: mirroring frequency hopping with fractional frequency reuse and multi-cell joint frequency hopping. Section IV gives simulation and performance analysis. Section V contains the conclusions. II. INTERFERENCE MITIGATION SCHEMES A. Fractional Frequency Reuse In [4, 5], fractional frequency reuse scheme is proposed. In this scheme, the users in each cell are divided into two groups based on its path gain (including path loss and shadow fading) to base station (BS): one is called cell-edge user group, in which user’s path gain to BS is above a specific threshold, the other is called inner-cell user group, in which user’s path gain to BS is below that specific threshold. Correspondingly, subcarriers (frequency band) are also divided into two parts in
Fractional (soft) frequency reuse scheme [3] is an interference coordination technique. It restricts the cell resource assignment including the time-frequency chunk and power resource at the cell boundary and coordinates these resource restrictions among the neighbor cells. The scheme can result in low interference in the adjacent cell boundary to enhance system performance.
1-4244-2424-5/08/$20.00 ©2008 IEEE
656
ICCS 2008
subcarriers
Fig.2 Frequency hopping scheme Fig.1 fractional frequency reuse scheme
each cell. One part, called edge subcarrier group, can be mainly used by cell-edge user group, also can be used by inner-cell group if it is not fully used by cell-edge user group. The other group, called inner subcarrier group, is only used by the inner-cell user group. Every neighbor cell has different subcarriers division to guarantee the neighbor cell’s edge subcarrier groups are orthogonal (i.e. no overlap). A typical 3-sector cell subcarrier division is shown in Fig.1.
The advantages of employing frequency hopping in a cellular system are well known. Frequency diversity is obtained if the channel frequency response varies over the hopping frequency. Moreover, by assigning different patterns to interfering cells, successive data symbols experience different interference conditions. Interference diversity can thus be achieved. A typical frequency hopping scheme is shown in Fig.2. The hopping pattern is N=5 Latin Squares. For UE A1, the transmit frequency is changed every OFDM symbol. Due to different subcarrier frequency response, it can have frequency diversity. Meanwhile, due to Latin Squares hopping pattern, the number of the frequency collision of A1 in cell 1 and A2 in cell 2 is only one. And UE A1’s experienced interference from cell 2 is different every OFDM symbol, it can get interference diversity. With frequency hopping, there is almost 1dB link gain in LTE system [6], but it is sensitive to the traffic load. If the adjacent cell traffic load is high, the experienced interference is high and frequency hopping gain is relatively low.
Cell 1, cell 2 and cell 3 are adjacent. Each cell’s cell-edge group is noted as E1, E2 and E3 respectively, the part in the virtual circle of each cell is inner-cell group. The subcarrier division of each cell is illustrated in the right figure. Each cell’s edge subcarrrier group F1, F2 and F3 is labeled the same color with the cell-edge group. They are orthogonal through inter-cell resource coordination, which is given by
Fi ∩ Fj = ∅ (i ≠ j; i, j = 1, 2,3) 3
* Fi = F ,
(1)
i =1
Where F stands for the whole frequency band. In this scheme, the cell-edge UE should firstly be scheduled followed by the inner cell UE. The UE b in the cell-edge group of cell 1 can only use subcarriers in F1. Correspondingly, the UE c can only use subcarriers in F2. If the bandwidth of F1, F2 and F3 is same, the FRF of cell edge area is 3. But the FRF of whole cell is still 1 since inner UE can use subcarrier in edge subcarrier group. Although UE b and c’s distance is very small, there is no interference between b and c since they use different subcarriers. the UE a and d can use all the subcarriers, and the assigned subcarrier of UE a maybe collides with UE d, but the interference between them is not very high because of long distance fading.
Therefore, frequency hopping scheme improves the performance under the relatively low user load. The frequency hopping scheme always randomly assigns the hopping subcarrier, and it has low average throughput compared to the frequency-selective scheduling. III. HYBRID INTERFERENCE MITIGATION SCHEME Considering these two inter-cell interference migrating techniques and the advantage of these two schemes, we propose a hybrid inter-cell interference mitigation scheme, which can increase the cell-edge user peak bit rate and keep high average cell throughput. In the hybrid scheme, the UE and subcarriers are still divided into two groups, which is same as in the fractional frequency reuse scheme. The cell-edge UE is still restricted in the edge subcarrier group, but it transmits data in frequency hopping ways, hopping in the edge subcarrier group.
Recently, fractional frequency reuse scheme is the most popular ICI mitigation scheme, which can greatly improve the cell-edge throughput. Additonly, this scheme has some derivative on subcarriers and power allocation.
The cell-inner UE is scheduled by the frequency-selective scheduler according to its reported SINR. In this way, for the cell-edge UE, it is not interferenced by the cell-edge UE because of orthogonal subcarriers group, only interferenced by the inner-cell UE from the neighbor cell, which interference is relatively low because of long distance fading.
B. Frequcency Hopping Frequency hopping scheme is an effective method to improve the link performance. In OFDMA system, the subcarrier is used to differentiate different UE, so it is very easy to implement frequency hopping. In the frequency-hopped OFDMA (FH-OFDMA), the subcarriers are assigned to users according to predetermined hopping patterns.
657
cell 1 f
cell 2
cell 3 F1
F1
F1
...
F2 F3 F1'
F1'
slot
F2' F3' (a)
F1'
Scheduling period (b)
Fig.3 mirroring hopping with fractional frequency reuse scheme
Fig.4 Multi-cell joint frequency hopping
Moreover, it has frequency hopping link gain. For the inner-cell UE, it is scheduled by the frequency-selective scheduler on the left subcarriers, which will bring high average throughput.
division of the UE is the same with the former. The division of the subcarriers is not fixed, which changes in every slot. The edge subcarrier group is hopping based on a specific hopping pattern. All the neighbor cells use a same hopping pattern, but the hopping start point is different, and is orthogonal. Due to hopping based on the same pattern, the following hopping subcarriers will not overlap and edge subcarriers group remains orthogonal. The example is shown in Fig.4.
Traditionally, the frequency hopping is divided into mirroring hopping and pattern based hopping [7], so we have two different hybrid inter-cell interference mitigation schemes. One is mirroring frequency hopping with fractional frequency reuse, the other is multi-cell joint frequency hopping.
Assuming that the whole frequency band has 6 subcarriers. At the first slot, through inter-cell resource coordination, the edge subcarrier group of cell 1 has 1 and 2 subcarrier, the edge subcarrier group of cell 2 has 3 and 4 subcarrier, and the edge subcarrier group of cell 3 has 5 and 6 subcarrier. The left subcarriers in each cell belong to inner subcarrier group. Obviously, at the first slot, the edge subcarrier groups of neighbor cell of neighbor cells are orthogonal. The subcarriers in the edge subcarriers group change based on the same specific pattern in the following time slot. The hopping pattern is random, but should guarantee that the frequency gap after hopping is larger than the system coherent bandwidth to make sure that the channel response is not coherent. Based on the same hopping pattern, the different subcarriers will not hop onto the same subcarrier.
A. Mirroring Hopping with Fractional Frequency Reuse In the real LTE system, a scheduling period (TTI) is two time slots. For the frequency hopping UE, the assigned subcarrier is unchanged in every time slot, but changed over time slot. So in scheduling period, the number of the frequency hopping is two. The subcarriers can be mirrored at the central subcarrier, as Fig.3. F1 and F1’ subcarriers compose cell 1 edge subcarrier group, while F2 and F2’ subcarriers belong to the cell 2 edge subcarrier group. The frequency hopping of cell-edge UE of cell 1 is shown in the right figure. In the first time slot, cell-edge UE use the upper subcarrier group F1. In the second time slot, UE changes to the lower subcarriers group F1’, and vice versa.
At the second slot, subcarrier 1, 2, 3, 4, 5 and 6 hop onto subcarrier 2, 4, 1, 6, 3 and 5, subcarrier 2 and 4 belong to the cell 1 edge group, subcarrier 1 and 6 belong to the cell 2 edge group, and subcarrier 4 and 5 belong to the cell 3 edge group. The edge subcarriers group still keeps orthogonal, which lead to no interference between the cell-edge UEs in the neighbor cell.
In such a case, the initial subcarrier selected for transmission (in the first slot) can be chosen based on UE channel quality index (CQI) reports while the subsequent subcarrier are chosen from a deterministic hopping pattern (mirroring mechanism).
With the continuous hopping, the edge subcarrier group can experience all the subcarriers, but not restricted in some subcarriers. The number of frequency hopping in one packet transmission is decided by the scheduling period. In multi-cell joint hopping, it is very ease to implement cell handoff. The UE should be informed the hopping start subcarrier when it handoffs to the neighbor cell, and it changes the subcarrier based on the same hopping pattern.
If the scheduling period is larger than two time slots, then we can fold the frequency band several times.
B. Multi-Cell Joint Frequency Hopping In mirroring hopping scheme, the hopping frequency is restricted in the edge subcarrier group instead of the whole frequency band. If the channel frequency response in F1 is similar with channel response in F1’, the hopping gain is not very large. So we propose a multi-cell joint frequency hopping scheme, which allows the cell-edge UE hopping in the whole frequency band to get high hopping gain. In this scheme, the subcarriers and UEs are also divided into two parts, and the
IV. SIMULATION AND ANALYSIS In fact, the proposed scheme is the combination of the fractional frequency reuse and the frequency hopping. We will
658
Tab.1: Simulation parameters Parameters
Values
Channel bandwidth
5MHz
Carrier frequency
2GHz
FFT size
512
Number of subcarrier available
300
Subcarrier spacing
15KHz
Sampling frequency
7.68MHz
White noise power density Cellular layout Hexagonal grid
-174 dBm/Hz 19 sites
Inter-cell distance
2km
BS transmit power
43dBm
Scheduling
Proportional Fair
Slot duration Channel model
Pedestrian B, 30Km/h
Distance-dependent path loss Lognormal Shadowing
Fig.5 Received SINR varying with the distance
0.5ms
L=128.1 + 37.6log10(R) 0 mean, 8dB standard deviation
compare hybrid scheme and traditional fractional frequency reuse scheme from complexity and performance aspects. The complexity difference of two schemes is based on the frequency hopping. In LTE system, frequency hopping is optional for the control channel, so hopping will not bring much complexity for data transmission. In addition, frequency hopping can reduce scheduler signaling overload. For a packet data transmission, the estimation of the channel response may be done the times with the number of frequency hopping. As to the performance, since the inner cell UE is scheduled with same criterion, the performance of inner cell UE is identical. The main difference is based on the hopping gain and the frequency-selective scheduling gain on the cell-edge UE. The hopping gain is based on the channel response difference and the estimation of the channel response. Due to short duration time, the estimation of channel response in one time slot maybe not accurate compared to the two time slot duration. The transmitted data rate of UE is calculated as
R = B ⋅ log 2 (1 + SINR).
Fig.6 Cell edge throughput
We denote these two schemes as scheme A and scheme B, respectively. We assume that all the available subcarriers are transmitted with equivalent power. The detailed simulation parameters are listed in Tab.1.
(2)
We set SINR threshold as 0 dB to decide whether UE is in cell-edge area or not. The UE in the cell is uniformly distributed. There are 18 users and 6 users in inner and edge region, respectively. We divide all the 300 available subcarriers into two groups. The edge subcarrier group has 100 subcarriers, the left subcarriers belong to inner subcarrier group.
Because log 2 (1 + x) function is nonlinear function and a little increment on x when x is small will have relatively large output increment. If the cell-edge UE’s SINR is relatively small, the hopping gain over SINR will have good performance. As to the frequency-selective scheduling, it will have high cell throughput, but will not enhance the single UE’ peak rate. As to the frequency hopping, it can enhance the single UE’s peak rate.
Fig.5 shows the received SINR varying with the distance between UE and BS, and with the increase of the distance, the received SINR gradually descends because the path loss and interference from neighbor cell become gradually large. The received SINR curve is identical when SINR is above 0 dB because the inner-cell user subcarriers assignment is the same. When the SINR is below 0 dB, the SINR of both scheme A and B get large and the SINR of scheme B is larger than scheme A.
In this section, the performance of the proposed scheme is evaluated by system level simulation. For similarity, we only evaluate normal fractional frequency reuse on the uplink and the mirroring hopping with frequency coordination.
Since cell-edge subcarriers are orthogonal, the interference from neighbor edge UE is zero. Due to the dominant 659
interference, the SINR will get large suddenly. It almost has 10dB gain and the frequency hopping will bring extra 1dB gain. Fig.6 shows the cell edge throughput. The horizontal axis represents simulation time (in slot) and the vertical axis represents throughput. Since the scheduler has no idea of interference from neighbor cell, the curve of throughput has fluctuation. From the Fig.6, we can see the cell edge throughput of scheme B has almost 7% gains above the throughput of scheme A. V. CONCLUSIONS In this paper, a hybrid inter-cell interference mitigation scheme is proposed, which combines the fractional frequency reuse and frequency hopping schemes. In this scheme, cell-edge UE has no interference of adjacent cell-edge UE from neighbor cells and transmits data in frequency hopping to get frequency diversity to enhance the peak data rate. According to the different hopping ways, it has mirroring hopping with fractional frequency reuse and multi-cell joint hopping scheme. The later has more flexibility of frequency hopping. From the analysis and simulation, the hybrid scheme can improve the peak bit rate of cell-edge user while getting high average throughput. REFERENCES [1]. Plass, Simon, Doukopoulos, Xenofon G, Legouable, Rodolphe, “Investigations on Link-Level Inter-Cell Interference in OFDMA Systems”, Proceedings 13th Annual Symposium on Communications and Vehicular Technology in the Benelux (SCVT 2006), pp. 49-52, Nov. 2006. [2]. 3GPP TR 25.814, “3rd Generation Partnership Project; Technical Specification Group Radio Access Network; Physical layer aspects for evolved Universal Terrestrial Radio Access (UTRA) (Release 7)”, June 2006. [3]. Haipeng LEI, Lei ZHANG, Xin ZHANG and Dacheng YANG, “A Novel Multi-Cell OFDMA System Structure using Fractional Frequency Reuse”, Proc. IEEE Int. Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2007), September 2007. [4]. 3GPP R1-050507, Soft frequency reuse scheme for UTRAN LTE, May 2005. [5]. 3GPP R1-050841, Further analysis of soft frequency reuse scheme, Aug 2005. [6]. 3GPP R1-062497, Link Performance of Frequency Hopping in LTE Uplink Localized Transmission, Oct 2006 [7]. 3GPP R1- R1-075086, Way Forward for UL Hopping, Nov 2007
660
