An Incremental Frequency Reuse Scheme for an ... - IEEE Xplore

3 downloads 0 Views 351KB Size Report
Abstract—We propose an incremental frequency reuse (IFR) scheme that reuses effectively the radio spectrum through sys- tematic segment allocation over a ...

An Incremental Frequency Reuse Scheme for an OFDMA Cellular System and Its Performance Ki Tae Kim O and Seong Keun Oh School of Electrical and Computer Engineering Ajou University San 5, Wonchon-Dong, Youngtong-Gu, Suwon, 443-749, KOREA E-mails: {erene46O, oskn}@ajou.ac.kr Abstract—We propose an incremental frequency reuse (IFR) scheme that reuses effectively the radio spectrum through systematic segment allocation over a cluster of adjoining cells. It divides the entire frequency spectrum into several spectrum segments. Based on the above segmentation, a set of cell- specific segment allocation sequences is designed for universal frequency reuse. Here, each sequence defines its own base segment and allocation order. The designed sequences are then assigned to respective cells over the cell cluster. In this scheme, the base segments over the cell cluster are mutually non-overlapped and collectively exhausted, and the added segments are interfered with from surrounding cells, but only in an incremental and coordinated manner. Hence, the proposed scheme can provide better reuse efficiency over the conventional ones such as the classical universal frequency reuse scheme and the soft frequency reuse (SFR) scheme. In addition, the simple and flexible IFR scheme can be easily configured as most existing reuse schemes only by redefining the set of segment allocation sequences. A system-level simulator for an orthogonal frequency division multiple access (OFDMA) cellular system covering surrounding cells up to 3rd-tier has been implemented. Simulation results show that the IFR scheme provides quite better reuse efficiency as compared to the classical universal reuse scheme and the SFR scheme, especially at the cell-edge region. Keywords—Cellular System, Frequency Reuse, Incremental Frequency Reuse, OFDMA

I. INTRODUCTION Over last two decades, there has been an upsurge of demands for mobile and wireless communications from several points of view such as new services and the number of subscribers [1]-[2]. In addition, next-generation mobile communication systems should be able to meet high-quality service requirements such as high-quality video and high-speed Internet over wireless networks at the lowest possible price. In particular, tremendous interests in multimedia services are fueling the need for very high data rates in future wireless networks [1]. However, a radio spectrum might be lacking for supporting these demanding services unless an epoch-making improvement in spectrum utilization has been done [1]-[2]. So far, several advanced mechanisms have been developed for better use of a radio spectrum. They include adaptive modulation and coding [3]-[4], hybrid automatic repeat request [5], fast channel-aware scheduling [6]-[7], and multiple-input multiple-output techniques [8]. Despite those efforts, there are still a lot of limiting factors to the system capacity in wireless cellular systems. In particular, inter-cell interference (ICI) from neighboring cells is one of major limiting factors to the achievable signal-to-interference-plusnoise ratio (SINR) and the system capacity, especially at the cell-edge region.

978-1-4244-1645-5/08/$25.00 ©2008 IEEE

In classical cellular systems, frequency reuse mechanisms have been adopted in order to avoid ICI from neighboring cells [9]. Such mechanisms limit the utilization of the available frequency spectrum, of which amount is determined by the frequency reuse factor (FRF) adopted. Hence, the FRF of one or near one is getting one of most desirable features for upcoming systems. However, the classical universal frequency reuse scheme suffers from severe ICI from adjoining cells [9]-[10] because of its tight frequency reuse. Recently, some promising flexible spectrum reuse schemes [10]-[13] have been proposed such as the SFR scheme adopted in the 3GPP-LTE system [10], [12] and the fractional frequency reuse (FFR) scheme [13]. Among them, the SFR scheme achieving FRF-one can overcome severe ICI from adjoining cells at a cell-edge region, by emphasizing a part (called as the primary band) of the available radio spectrum and allocating it preferentially for cell-edge users. However, it still may incur even severer ICI to some of cell-edge users, because the high-powered primary band can accommodate only a pre-defined number of cell-edge users, while the remaining cell-edge users may be allocated to limitedly -powered secondary bands. In this paper, we propose an IFR scheme that reuses effectively the radio spectrum by allocating systematically spectrum segments over a cluster of adjoining cells. It divides the entire frequency spectrum into several segments. A set of cell-specific segment allocation sequences is designed for universal frequency reuse. Here, each sequence defines its base segment and allocation order for additional segments. The designed sequences are assigned to respective cells over the cluster. In this scheme, all the base segments within the cell cluster are mutually non-overlapping and collectively exhausted, and the added segments are interfered with from surrounding cells, but only in an incremental and coordinated manner. In each cell, the base segment is occupied first, and then the remainder of traffic channels is allocated over the added segments. Hence, the IFR scheme can provide better reuse efficiency over the conventional ones. To verify the effectiveness of the proposed scheme, a system-level simulator for an OFDMA cellular system covering surrounding cells up to 3rd-tier is implemented. We use the outage probability, spectral efficiency, and overall cell capacity as performance measures. II. SYSTEM MODEL A. Spectrum Reuse and Inter-cell Interference In this subsection, we deal with a spectrum reuse strategy and an ICI model for a generic cellular system. Fig. 1(a) and (b) show frequency assignments and interfering cells or interfering sectors up to 3rd-tier in a cellular system with FRF=3, for omni-cell and 3-sector cell, respec-

1504

tively. For a notation convenience, we would define the sets of indices of interfering cells according to the loading factors and sectorization types as follows:  if omni-cell and LF ≤ 1/3  J = {8,10,12,14,16,18,19, 22, 25, 28, 31, 34}  ,   if omni-cell and LF > 1/3  J = {1, 2, 3," ,36} 

 if     if  

(1)

trol, we would set P(i , j ) , j = 0,1, ",36 to a constant value.

B. Performance Metrics Now, we define three metrics to evaluate the system performance such as the SINR, the outage probability, and the spectral efficiency. First, we define the SINR of the i -th center user as

3-sector cell and LF ≤ 1/3, J = {4, 5,12,13,14,15,16, 27, 28, 29, 30, 31, 32} 3-sector cell and LF > 1/3,

SINR(i ) =

. (2)

P(i ,0) ⋅ L(i ,0) ⋅ S(i ,0) ⋅ H (i ,0)

∑(

)

β β

β β γ

β γ β

γ β γ

α

α

γ

β γ

β α

α

γ

β

γ β γ

α

α

α

α

α

γ

β

γ β γ

β γ

β γ

β γ

α

α

α

α

α

γ

β

γ β γ

β γ

β

γ β

α

γ

β γ

α

α

β γ

β α

α

α

α

α

γ

β γ

β

γ β γ

β γ

2

.

(5)

P(i , j ) ⋅ L(i , j ) ⋅ S(i , j ) ⋅ H (i , j ) + Ni

j∈J

J = {1, 2, 3," , 36}

2

From the SINR in (5), we define the outage probability as (6) Pout = Pr [ SINR(i ) < η ] , α

α

α

α

α

α

β

γ

β γ

β γ

β

γ β γ

α

α

α

α

α

β

γ

ββ γγ β γ β γ

where Pr [ SINR( i ) < η ] denotes the probability that the α

SINR value goes down below a service outage threshold η . From using the SINR again, the spectral efficiency for the i -th center user is defined as (7) C( i ) = log 2 (1 + SINR( i ) ) bps/Hz.

α

α

α

(a) (b) Fig. 1. Frequency assignments and interfering cells or interfering sectors up to 3rd-tier, with FRF=3: a) Omni-cell; b) 3-sector cell.

To consider OFDMA systems, we would define subchannels each consisting of N OFDM-subcarriers. To evaluate the performance of each user allocated to a subchannel in a frequency-selective fading channel, we use the well-known exponential effective SINR mapping [15]-[16], defined as γ

For ICI modeling, we consider path losses, shadowing factors and Rayleigh fading coefficients from 36 surrounding cells to the desired cell [9],[14]. The effective path loss including the shadowing attenuation between the i -th center user and the j -th interfering base station (BS) is approximated to [9],[14] ξ /10 L( i , j ) = d ( i , j ) − ρ ⋅10 ( i , j ) , (3) where ρ denotes the path loss exponent but can also be differentiated, and d (i , j ) and ξ (i , j ) , respectively, denote the propagation distance and the shadowing factor between the i -th center user and the j -th interfering BS. The shadowing factor ξ (i , j ) is modeled as a real Gaussian random variable with zero mean and the standard deviation of σ p in dB [14]. In what follows, we would set the standard deviation σ p and the path loss exponent ρ to 8 dB and 4, respectively [14]. From (1) or (2) and (3), the received signal at the i -th center user can be expressed as

R( i: j ) =

P( i ,0) ⋅ S( i ,0) ⋅ L( i ,0) ⋅ H ( i ,0) +

∑( j∈ J

)

P( i , j ) ⋅ S( i , j ) ⋅ L( i , j ) ⋅ H ( i , j ) + N i

, (4)

where P(i , j ) , S(i , j ) and H (i , j ) , respectively, denote the transmit power, the transmitted signal, and the Rayleigh fading coefficient from the j -th interfering BS to the i -th center user; P(i ,0) , L(i ,0) , S( i ,0) and H ( i ,0) , respectively, denote the transmit power, the effective path loss, the transmitted signal, and the Rayleigh fading coefficient from the center-cell BS to the i -th center user; and N i is an additive white Gaussian noise at the receiver input of the i -th center user, with zero mean and variance N o . Assuming no downlink power con-

1 N − βi (8) ∑e . N i =1 Here, β is dependent on a modulation and coding scheme used and should be calibrated. For example, β has to be 2 in a case of QPSK signaling [15]-[16].

γ eff = − β ln

III. INCREMENTAL FREQUENCY REUSE SCHEME In this section, we present an IFR scheme that can reuse effectively a given frequency resource in an OFDMA cellular system and describe its operation principle in detail. Finally, we give some design examples.

A.

Concept of the IFR Scheme The IFR scheme first divides the whole frequency spectrum into several segments each consisting of a non-overlapping set of channels. The number of segments might be larger or equal to the number of cells within a cluster of adjoining cells. In addition, a set of segment allocation sequences for a cluster of cells are designed according to a segment allocation rule. Finally, the designed segment allocation sequences are assigned to respective cells within a cluster of adjoining cells again according to a segment allocation rule. We define the first segment in each segment allocation sequence as the base segment of the corresponding cell. Each cell-specific sequence then determines the incremental ordering for reusing additional segments beyond the base segment in the corresponding cell, when the amount of traffic exceeds the capacity of its own base segment. In this paper, we use the segment allocation rule as follows: ƒ The base segments among adjoining cells should not be overlapped each other. ƒ If the number of users exceeds the capacity of the base segment (called as simply the base capacity), additional

1505

ƒ

ƒ

segments assigned to a desired cell encroach partially or fully base segments assigned to adjoining cells. At every incremental stage to define the set of sequences, each segment is assigned only once, but at least once over the whole sequences. A cluster consists of cells which use collectively the complete set of segment allocation sequences.

B.

Operation Principle The details of operation principle are as follows: 1. In each cell within the cluster, the proposed scheme can allocate arbitrarily channels within its own base segment if the amount of traffic does not exceed the base capacity. 2. Beyond the base capacity, it uses additional segments according to the corresponding cell-specific segment allocation sequence. It first completes channel allocation within its own base segment in the corresponding cell and then allocates arbitrarily the remainder over the newly added segments. 3. This process will be continued until all the segments are exhausted, as the amount of traffic increases. Using this principle, we can avoid major ICIs from adjoining cells when the amount of traffic is less than the base capacity, while beyond the base capacity, ICI is averaged over the newly added segments by arbitrary channel allocation. In addition, the IFR principle can be applied to both omni-cell and 3-sector cell systems. The ICI control capability of the IFR scheme could be maximized if the cell-specific subchannel allocation sequences designed according to the above segment allocation rule are used within the cluster of cells. C.

Design Examples In our work, even if the IFR scheme can be widely applied to cellular systems combined with various multiple access technologies, we demonstrate its operation principle within the viewpoint of an OFDMA cellular system, in which the resource allocation is done on the basis of subchannel [17]. Now, we give two design examples as follows:

ƒ Scheme A for a cellular system based on FRF=3 Step 1. The scheme first divides the whole frequency spectrum into three segments each containing one-third set of total non-overlapping subchannels, such as A-, B-, and C -segments as shown Fig. 2. Step 2. Below the loading factor of 1/3, the scheme allocates arbitrarily subchannels only within only the corresponding base segment for a desired cell to demanding users. Step 3. If the loading factor is greater than 1/3 and smaller than 2/3, the scheme first completes the subchannel allocation within the corresponding base segment for the desired cell, and then allocates arbitrarily the remainder of requested subchannels within the 2nd segment (i.e., the 1st incremental segment) to demanding users. Step 4. Finally, when the loading factor exceeds 2/3, the scheme first completes the subchannel allocation within the corresponding base segment and the 1st incremental segment for the desired cell, and then allocates arbitrarily the remainder of requested sub-

channels within the 3rd segment (i.e., the 2nd incremental segment) for the desired cell to demanding users. ƒ Scheme B for a cellular system based on FRF=4 Step 1. The scheme first divides the whole frequency spectrum into four segments each containing one-fourth set of total non-overlapping subchannels, such as A-, B-, C-, and D -segments as shown Fig. 3. Step 2. If the loading factor is greater than 1/4 and smaller than 2/4, the scheme first completes the subchannel allocation within the corresponding base segment for the desired cell, and then allocates arbitrarily the remainder of requested subchannels within the 2nd segment to demanding users. Step 3. This process will be continued until four segments are exhausted. Power

C0

Power

C1

Power

C2

Fig. 2. The concept of the proposed IFR scheme A in the omni-cell environment for a cellular system based on FRF=3. Power

C0

Power

C1

Power

C2

Power

C3

Fig. 3. The concept of the proposed IFR scheme B in the omni-cell environment for a cellular system based on FRF=4.

Here, we would summarize the two sets of segment allocation sequences as follows: ƒ Using the scheme A -

Sequence-C0 : A-seg.→B-seg.→C-seg., Sequence-C1 : B-seg.→C-seg.→A-seg., Sequence-C2 : C-seg.→A-seg.→B-seg.

ƒ Using the scheme B -

Sequence-C0: A-seg.→B-seg.→C-seg.→D-seg., Sequence-C1: B-seg.→C-seg.→D-seg.→A-seg., Sequence-C2: C-seg.→D-seg.→A-seg.→B-seg., Sequence-C3: D-seg.→A-seg.→B-seg.→C-seg.

IV. SIMULATION RESULTS AND DISCUSSIONS We performed computer simulations to demonstrate the effectiveness of the proposed IFR scheme and also to compare the proposed scheme with both the classical universal frequency reuse scheme with FRF=1 [9] and the SFR scheme [10], [12]. We used three performance measures such as the

1506

outage performance, spectral efficiency, and overall cell capacity. We considered an OFDMA downlink cellular system in an omni-cell case for simulations. We assumed a uniform loading factor over all the cells considered for the purpose of simplicity. We assumed a uniform user distribution within a hexagonal cell region. For all simulations, we used simulation parameters as shown in Table I. The cell-edge performance has been evaluated by averaging over all the users between the radius of 0.9 km and 1.0 km. The overall cell capacity has been obtained by cumulating the respective capacities of all the users within a cell, uniformly distributed within the hexagonal cell region. Subchannel construction has been done based on the partial usage of subcarriers (PUSC) channelization of the IEEE 802.16 standard, each consisting of 24 data subcarriers [18]. We assumed that only one subchannel is allocated to each user and the exponential effective SINR mapping [15]-[16] was used. TABLE I SIMULATION PARAMETERS Parameter No. of total subcarriers No. of data subcarriers No. of subchannels No. of data subcarriers per subchannel No. of guard subcarriers Path loss exponent Shadowing factor variance Spectral efficiency upper limit SINR outage threshold Maximum delay spread Cell radius No. of interfering cells Power amplification factor α (SFR scheme [10])

the whole frequency spectrum. Hence, it is worst among three reuse schemes. From Figs. 4 and 5, we conclude that the scheme A based on FRF=3 performs better than the scheme B based on FRF=4 as the loading factor increases over 25%, due to symmetric interference geometry. Fig. 6 shows the overall cell capacity of three reuse schemes considered here. From the results, we can see that the proposed scheme can provide better overall cell capacity than other two schemes. The improvement becomes more significant as the loading factor increases. With the full loading factor, the overall cell capacity of the IFR scheme is equal to that of the classical universal reuse scheme and is very greater than that of the SFR scheme. This difference is caused by the transmit power unbalance between the primary-band and secondary-band in the SFR scheme. In the SFR scheme, two-thirds of users allocated to the secondary-band achieve relatively lower spectral efficiencies due to the limited transmit power, while only one-third achieving higher spectral efficiencies. V. CONCLUSIONS

Value 2048 1680 60 24 Left(184), Right(183) 4 8 dB 4.5 bps/Hz 3 dB 81 n s 1 km (at a hexagonal vertex) 36 (up to 3rd tier) 1.5, 2

Figs. 4 and 5, respectively, show the outage probabilities and spectral efficiencies of the IFR schemes and two conventional schemes. From both figures, the proposed scheme can reduce effectively the ICI, especially when the loading factor is less than the base capacity (e.g., 1/3 and 1/4 for cellular systems based on FRF=3 and FRF=4, respectively.). The service outage performance of the IFR scheme A becomes approximately the same as that of the SFR scheme, but outperforms that of the classical universal reuse scheme. This improvement in the scheme A results from the following: At the cell-edge region, center-cell users might be heavily interfered with by signals from adjoining cells, since signals transmitted from the center BS might be maximally attenuated at cell-edge region; Below the base capacity, that is, the loading factor of 1/3, the scheme A can avoid completely ICI from adjoining cells, thus minimizing the ICI level. Beyond the base capacity, both the incremental property and distributive property of the segment allocation sequences make interfering cells evenly distributed and also the ICI level gradually increased, thus keeping the ICI level lower. Moreover, in the SFR scheme, cell-edge users might occupy the power-emphasized primary band, thus cell-edge users being able to avoid severe ICI from adjoining cells. Therefore, the SFR scheme performs slightly better than or comparable to the IFR scheme, depending on the power amplification factor value [10]. However, the ICI control mechanism of the classical universal reuse scheme is to only average ICI over

In this paper, we have proposed an IFR scheme that can reuse effectively a given radio spectrum in an OFDMA cellular system, by defining a set of systematic segment allocation sequences over a cluster of adjoining cells. Below the base capacity, the scheme can avoid completely ICI from adjoining cells, thus minimizing the ICI level. Beyond the base capacity, interfering cells are evenly distributed and also the ICI level is gradually increased, thus keeping the ICI level lower. Hence, the proposed scheme provides better reuse efficiency over the conventional schemes such as the classical universal frequency reuse scheme and the SFR scheme. In addition, we have verified the effectiveness of the proposed scheme as compared with that of the classical universal reuse scheme and the SFR scheme through extensive simulations. From simulation results, we see that the IFR scheme performs better over the classical universal reuse scheme does and is comparable to the SFR scheme in terms of the average outage probability and spectral efficiency. The scheme A performs better than the scheme B when the loading factor is over 25%, due to symmetric interference geometry. The proposed scheme also can provide better overall cell capacity than other two schemes. The improvement becomes more significant as the loading factor increases. In addition, the IFR scheme can be easily configured as most existing reuse schemes only by redefining the set of segment allocation sequences. ACKNOWLEDGEMENT This research was supported by the MIC (Ministry of Information and Communication), Korea, under the ITRC (Information Technology Research Center) support program supervised by the IITA (Institute of Information Technology Advancement) (IITA-2008-(C1090-0801-0003)) REFERENCES [1] [2]

1507

K. Letaief and Y. Zhang, “Dynamic multiuser resource allocation and adaptation for wireless systems,” IEEE Comm. Mag., vol. 13, pp. 681-697, Aug. 2006. A. Jamalipour, T. Wada, and T. Yamazato, “A tutorial on multiple access technologies for beyond 3G mobile networks,” IEEE Comm. Mag., vol. 43, pp. 110-117, Feb. 2005.

[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]

Cell-boundary (0.9-1km), SINRth=3dB

4

avg. spectral efficiency (bps/Hz)

[4]

J. Hayes, “Adaptive feedback communications,” IEEE Trans. Comm., vol. 16, pp. 29-34, Feb. 1968. A. Goldsmith and S.-G. Chua, “Variable-rate variable-power MQAM for fading channels,” IEEE Trans. Comm., vol. 45, pp. 1218-1230, Oct. 1997. R. A. Comore and D. J. Costello Jr., “ARQ schemes for data transmission in mobile radio systems,” IEEE J. Select. Areas Comm., vol. 2, pp. 472-481, Jul. 1984. T. Kolding, “Link and system performance aspects of proportional fair scheduling in WCDMA/HSDPA,” in Proc. IEEE VTC2003-Fall, vol. 3, pp. 1717-1722, Oct. 2003. M. C. Necker, “A comparison of scheduling mechanisms for service class differentiation in HSDPA networks,” Int. J. Elect. Comm., vol. 60, pp. 136-141, Feb. 2006. B. Vucetic and J. Yuan, Space-Time Coding, John Wiley, 2003. T. S. Rappaport, Wireless Communications: Principles and Practice, Prentice Hall PTR, 2nd ed., 2002. Y. Xiang and J. Luo, “Inter-cell interference mitigation through flexible resource reuse in OFDMA based communication networks,” in Proc. European Wireless’2007, pp. 1-7, Apr. 2007. M. Sternad, T. Ottosson, A. Ahl´en, and A. Svensson, “Attaining both Coverageand High Spectral Efficiency with Adaptive OFDM Downlinks,” in Proc. IEEE VTC2003-Fall, vol. 4, pp. 6-9,Oct. 2003. “Soft frequency reuse scheme for UTRAN LTE,” Huawei, 3GPP R1-050507, TSG RAN WG1 Meeting #41, Athens, Greece, May. 2005. J. Tomcik, Qualcomm, “MBFDD and MBTDD wideband mode,” IEEE 802.20-05/68r1, Jan. 2006. S. Plass, S. Sand, and G. Auer, “Modeling and analysis of a cellular MC-CDMA downlink system,” in Proc. PIMRC2004, vol. 1, pp. 160-164, Sep. 2005. “Effective SIR computation for OFDM system-level simulations,” Nortel, TSG-RAN WG1 #35, R03-1370, Nov. 2003. R. Yaniv, D. Stopler, T. Kaitz, and K. Blum, ”CINR measurements using the EESM method,” Alvarion Ltd., 2005 Z. Wang and R. A. Stiring-Gallacher, “Frequency reuse scheme for cellular OFDM systems,” Electron. Lett., vol. 38, no. 8, pp. 387-388, Apr. 2002. IEEE 802.16-2004, “IEEE standard for local and metropolitan area networks-Part 16: Air interface for fixed broadband wireless access systems,” Oct. 1, 2004.

Classical(FRF=1) IFR scheme Scheme A(based on FRF=3) Scheme B(based on FRF=4) SFR scheme (Primary-band : Secondary-band) Power ratio 2:1 (α=1.5) Power ratio 4:1 (α=2)

3

2

1

0 0.0

0.2

0.4

0.6

0.8

1.0

loading factor

Fig. 5. Cell-edge average spectral efficiency as a function of the loading factor in the same environment as in Fig. 4

120

100

overall cell capacity (bps/Hz)

[3]

80

60

SINRth=3dB Classical(FRF=1) IFR scheme Scheme A(based on FRF=3) Scheme B(based on FRF=4) SFR scheme (Primary-band : Secondary-band) Power ratio 2:1 (α=1.5) Power ratio 4:1 (α=2)

40

20

0 0.0

0.2

0.4

0.6

0.8

1.0

loading factor

avg. outage probability

0.8

Fig. 6. Overall cell capacity as a function of the loading factor in the same environment as in Fig. 4

0.6

0.4

Cell-boundary (0.9-1km), SINRth=3dB

0.2

Classical(FRF=1) IFR scheme Scheme A(based on FRF=3) Scheme B(based on FRF=4) SFR scheme (Primary-band : Secondary-band) Power ratio 2:1 (α=1.5) Power ratio 4:1 (α=2)

0.0 0.0

0.2

0.4

0.6

0.8

1.0

loading factor

Fig. 4. Cell-edge average outage probability of three frequency reuse schemes as a function of the loading factor, assuming uniform loading factor over all the cells.

1508

Suggest Documents