nICrICrCncCNunuCmCnIn L1LNP1CmIOcC 5y8ICm8

\8n1rucIOnu1rCqUCncyKCU8C
1aeyeaa,|ee.isaaYeea.aa,|ae|ee.non Member aaci|aeaia.Member, IEEE
School ofInformation and Communication Engineering
Sungkyunkwan University, Suwon, Kyeonggi, Korea
;|ia,earlee.yjseacc.|aaya,ex.j|saia)s||a.eca
Abstract- This paper presents an interference management
scheme for OFDMA Femtocell systems. Femtocell is recently
introduced for indoor coverage extension. However,
interference problem between the femtocell and the macrocell
should be solved in advance. In this paper, we propose an
efcient interference management scheme in the OFDMA
femtocell systems using Fractional Frequency Reuse (FFR) in
order to minimize the interference between both cells. Under the
pre-allocated frequency band within a macrocell through the
FFR optimally, the proposed scheme allocates sub-bands the
femtocells efciently to consider macrocell having a priority
over femtocells and total/edge throughputs. Simulation results
show that proposed scheme enhances the throughputs
performance in overall network, especially for the macrocell
users and cell edge users.
1. INTRODUCTION
O
OTHOGONAL Frequency Division Multiple Access
(OFDMA) is one of major features for next generation
communication standards such as Long Term Evolution (L TE)
and IEEE 802.16 (Worldwide Interoperability for Microwave
Access, WiMAX). The fequency and time resources are
allocated to users in orthogonal manner. A key feature is
fequency reuse factor of 1 for maximal resource use in the
OFDMA system. Recently, femtocell has proposed for indoor
coverage extension. The femtocell is defned as very small
size, low-power home base station (BS) that works in the
licensed fequency bands, and it is connected to broadband
Internet backhaul [1]. The femtocell brings various benefts to
both consumers and operators, such as enlarged indoor
coverage, enhanced system capacity, Quality of Service
(QoS), and reduced capital and operation expense. Due to
increasing indoor phone calls and data services but
insufcient macrocell coverage, the femtocell could be an
attractive solution.
However, interference problem between the femtocell and
the macrocell should be solved in advance, because the
femtocell is deployed over the existing macro network, and it
uses same spectrum with the macrocell. Dedicated channel
approach is one of the easiest ways to solve this problem, but
the fequency resources are not utilized efectively.
Co-channel model is suitable for practical deployment, but
the Co-channel interference (CCI) problem should be solved
[2]. The femtocell is ofen tured on and of, and installed at
unknown location, because it is managed by a personal
customer, not by a network operator. Therefore, many
Manuscript received March 25, 2010. This work was supported by the
National Research Foundation (NRF) grant funded by the Korea goverment
(MEST) (2010-0016896).
parameters should be automatically adjusted to avoid the
CCI.
Fractional Frequency Reuse (FFR) is discussed in the
OFDMA based network to overcome the CCI problems [3-5].
In the FFR, whole fequency band is divided into several
sub-bands, and each sub-band is diferently assigned to center
zone and edge zone of the cell. While reuse factor of the
center zone is one, the edge zone adopts bigger reuse factor.
As a result, intra-cell interference is removed, and inter-cell
interference is substantially reduced. At the same time the
system throughput is enhanced.
Dynamic Frequency Planning (DFP) algorithm was
another proposal for interference avoidance [6]. Afer
dividing the OFDMA network into several sectors, the
amount of sub-channels is estimated considering user
bandwidth demand in each sector. Interference among sectors
are calculated when the sectors transmit the same fequency.
Optimization function is run to minimize the overall network
interference. However, this scheme follows a centralized
network structure, which is not appropriate for femtocell
managed by a personal owner.
Several heuristic algorithms for fequency assignment
were investigated [7]. The recommended one is Least
Interference Power (LIP) algorithm, where a powered-up
femtocell BS chooses a frequency segment that minimizes the
interference level. In tur-on orders algorithm [7], fequency
allocation is conducted according to the order of the femtocell
tured on. However, the femtocells are deployed in the form
of rectangular matrix, not in the random manner. Also,
interference between the macrocell and the femtocell is not
evaluated.
Isolated and coupled model considering user location was
discussed for OFDMA femtocells [8]. In the isolated model,
macrocell and femtocell users are allocated different
resources split by time and fequency slots. The coupled
model reuses macrocell resources for femtocells which are
located in the cell boundary, while the center zone femtocells
use orthogonal resources fom the macrocell. However, these
schemes do not apply the FFR concept for the OFDMA
system.
Frequency reuse and pilot sensing scheme was proposed to
reduce the CCI [9]. Afer applying the FRF of 3 or above to
the macrocells, the femtocells use the remaining fequency
sub-bands. For example, if the reuse factor is three and the
macrocell uses sub-band I among the three sub-bands I, II,
and III, the femtocell chooses sub-band II and III. However,
the macrocell throughput is reduced, even though the overall
capacity is enhanced. Also the reuse factor in macro cell is
978-i-4244-822J-8/i0/$2.00 20i0 IEEE
against the OFDMA based network such as the L TE and the
Wi MAX, where the target of the reuse factor is one.
The FFR is one of the solutions to reduce inter-cell
interference in macrocell system, especially for the cell edge
users. Also it is helpfl to achieve the reuse factor of one.
Under this condition, the interference fom the femtocell
deployment should be minimized for the macrocell users. So,
we focus on the interference management between the
macrocell and the femtocell using the FFR.
In this paper, we propose an effcient resource allocation
scheme in the OFDMA femtocell systems using the FFR. We
defne optimization problem on allocating frequency
sub-bands to femtocell in each zone and evaluate an effcient
interference management scheme using sectored-FFR and
fequency band allocation. Under the macrocell allocates
fequency band using the FFR, the femtocell chooses
sub-bands which are not used in the macrocell sub-area to
avoid interference.
II. FFR FEMTOCELL ALLOCA nON PROBLEM
A. OFDMA Femtocell System Model
The frst step in problem formulation in the OFDMA
femtocells is to allocate the fequency sub-bands into a
macrocell based on FFR. The other fequency allocation for
femtocell users will be done afer allocating macrocell users
optimally. This allocation order is acceptable because
macrocell users have a priority over femtocell users. This
order reduces the complexity of the optimal allocation
problem in macro-/femto-cell mixed systems.
The optimal fequency sub-band allocation for macro cell is
assumed by referring previous works [5, 10]. In [5], authors
provide an analysis of the interference coordination technique
in macrocell OFDMA systems. An optimization problem is
formulated and solved via a linear interior point method. The
objective of this problem is to determine the fequency reuse
factor of the cell exterior zone and the percentage of the
overall system bandwidth attributed to either interior or
exterior zones. Simulation results show that the optimal
interior zone radius is 630m (cell radius 1000m) and the
optimal fequency reuse factor of the exterior zone is equal to
3. Ref. [10] shows more than three sectorization can not make
distinct differentiation on performance.
From previous research results for macrocell fequency
allocation, we can set the size of center zone (interior zone) to
0.63 of macrocell coverage and fequency reuse factors of
interior and exterior zone to I and 3 respectively. Also for
macrocell, each cell is partitioned into three sectors and
different fequency sub-band is allocated to the each
macro cell sub-area according to the above FFR method. As
shown in Fig. l(a), the set of interferers for a user in sector El
of macro cell 1 is the adjacent macrocells of cells numbered {2,
7,8,9, 17, 18, 19} but a user in center zone of cellI interfered
fom all 18 macrocells. By applying cell confguration of
Fig.l, edge users can achieve higher throughput due to
reduced number of interferers.
(a)
Fig. !. (a) 19 Multi-cell OFDMA structure, (b) The basic layout and
allocation for macrocell users using sectored-FFR.
Fig. 1 (b) shows basic fequency allocation for macrocell
users when sectored-FFR used.
B. Problem Formulations
CCI between the femtocell and macrocell users is
increased when using same sub-band at same sectored zone.
Thus, the objective of allocating fequency sub-bands for
femtocell is how many sub-bands are allocated to femtocell
users in each zone.
The achievable SINR value of macrocell user m in center
zones can be shown as follows:
P
G
St\'
{Cl C2 C3}
'
M.k m.M.k
'
:
^ ' L. P'.kGm.M'.k ' L.
P
F,kGm.F.k
!1' F
(1)
where PM,k and PM',k is transmit power of serving macrocell M
and neighbor macrocell M' on sub-carrier k, respectively.
G m,M,k is channel gain between macrocell user m and serving
macrocell M on sub-carrier k. Channel gain from neighbor
macrocells are denoted by Gm,M',k. Similarly, PF,k is transmit
power of neighbor femtocell F on sub-carrier k. Gm,}',k is
channel gain between macrocell user m and neighbor
femtocells F on sub-carrier k. No is white noise power spectral
density, and /f is sub-carrier spacing. Similarly, SlR of
macrocell users in exterior zone El on sub-carrier k is
... (2) k
t\

\
' '....' |..
M'=2,7,,9, l7, l, l9]
The other SlR of macrocell users in E2 and E3 are similar to
Eq. (2) except M' which is different according to sector.
In case of a femtocell user in center zones, it is interfered
fom all 19 macrocells and adjacent femtocells. The received
SlR of a femtocell user f on sub-carrier k in center and edge
zones can be expressed similarly as Eq. (1) and (2).
The channel gain G is dominantly affected by Rayleigh
fading( n, log-normal shadowing( Xu in dB), and
outdoor/indoor path losses. The path loss for outdoor is
modeled as PLoll/door 28+3510gJO(d dB, where d is the
distance fom a BS to a user in meters. Otherwise, indoor
model is PLiooor 38.5+20l0glO(d+Lwols dB, where Lwalls is 7,
10, and 15 dB for light internal, interal, and exteral walls,
respectively [2]. So, the channel gain G of user i on
sub-carrier k can be expressed as
G " 10-(P+xu)IIO IH 12 (3)
I,
k
I,
k

The practical capacity of macroceIl user N or femtoceIl
user f on sub-carrier k can be given by [3]
Cmor/,k =f.j10g2(1+aSINRmor/,k)
(4)
where, ( is a SIR gap for target Bit Error Rate (BER), and
defned by a = l.5 / [ 5BLR) . Here, we set BER to 10.6.
The overaIl throughput of serving macroceIl M and
femtoceIl F can be expressed as,
TM ..`..
(S)
N .
1
T} ..
(6)
/ .
where, _,

, notif the sub-carrier assignment for

macroceIl and femtoceIl users. When _,

, 1, the
sub-carrier k is assigned to macroceIl user N or femtoceIl user
f Otherwise, _,

, O. From the characteristics of the

OFDMA system, each sub-carrier is aIlocated only one
macroceIl user or femtoceIl user in a macroceIl or a femtoceIl
in every time slot. This implies that _, 1 for Vk,
where Nm is the number of macroceIl users in a macrocell.
Similar expression for femtoceIl users related to the practical
capacity and the overaIl throughput is possible except

,,

, 1 on k E I
x
' Where I
x
is available
sub-bands allocated to femtoceIl in a zone X , i.e., X E
[c1, c2, c8, |1, |2, |8].
EventuaIly, we can formulate centralized optimization
problem of allocating sub-bands to femtocells in each zone as
follows.
subject to . 1. _, 1, for Vk
2.

,,

, 1, for k E I

3. six six.
,
-.
Thus, the objective of optimization is that sub-bands
aIlocation for femtocell in each zone X should be decided. In
other words to fnd available fequency sub-bands for
femtoceIls in each zone X, that is to fnd optimal subset I
x
in
each zone X, is our optimization goal. In the given centralized
optimization problem, to investigate overaIl possible
combination of subset I
x
takes much time so that can easily
not obtain optimal solution within each time slot. The
thinkable cases in each zone are selecting n sub-bands among
4 sub-bands {A, B, C, D} , i.e.,
,)+ ,+ '+ ',
1
'
cases arise in the zone. Since a macroceIl has 6 zones, the
number of overaIl cases is 1
'
'. Therefore, we proposed
distributed sub-optimal sub-bands aIlocation scheme having
relatively low interference between macroceIl users and
femtoceIl users
C Inuencing Factors Needed To Discuss
How to setup the relative ratio of fequency sub-bands A, B,
C, and D in macroceIl needs to be discussed. In this paper, for
the sake of simplicity, we refer to the results in [5], [12]. But
there exist the difference between macroceIl-only system and
our femtoceIl system. In summary, the ratio of each sub-band
has an effect on throughput lightly.
III. PROPOSED INTERFERENCE MANAGEMENT SCHEME
In order to solve the formulated problem, frst we can
aIlocate fequency sub-bands in practicaIly disjoint ways. So
the femtoceIl chooses sub-bands which are not used in the
macroceIl sub-area.
z
MOC|O` D
|0mlO A,b,L
L
D
Fig. 2. Proposed practical allocation scheme for femtocell users under given
macrocell allocation distribution
We all ocate the frequency sub-bands into macroceIl and
femtoceIl as depicted in Fig. 2. The macroceIl coverage is
divided into center zone and edge region including three
sectors per each, denoted by Cl, C2, C3, and El, E2, E3. The
whole fequency band is partitioned into two parts, and one
part of them is classifed into three pieces again, totaIly
denoted by A, B, C, and D.
For macroceIl, different fequency sub-band is aIlocated to
the each macroceIl 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 (CI , C2, and C3), and sub-band B, C, and D is
applied in the E 1, E2, and E3 regions, respectively.
Under this circumstance, the femtoceIl chooses sub-bands
which are not used in the macroceIl sub-area. EspeciaIly
when the femtoceIl is located in the center zone, the femtoceIl
additionaIly excludes a sub-band which is used by macroceIl
in the edge region of current sector. For example, when a
femtoceIl is located in region E I, it uses sub-band A, C, or D,
while the macro ceIl uses sub-band B. If a femtoceIl is
positioned in zone Cl, sub-band C and D is applied. The
femtoceIl avoids sub-band A which is used by macroceIl in
zone Cl. Also it avoids sub-band B which is used by
macroceIl in region E 1, because the received signal power of
sub-band B is relatively strong for that femtocell.
Due to the characteristics of the OFDMA, the macrocell is
interfered by inter-ceIl, and that interference is frther
mitigated by the FFR. The femtoceIl uses different sub-band
to avoid interference fom the macrocell. The sub-band is
reused in the macroceIl coverage as much as possible,
because transmitted power of the femtoceIl is very small.
Therefore, the interference between the macroceIl and
femtoceIl is greatly avoided. Also more sub-carriers are
aIlocated to femtoceIl which is located in the edge region, in
order to improve the performance of the edge users.

IV. SIMULATION RESULTS
A. Scenarios and Environments
In In this paper, we evaluate the proposed scheme in terms
of throughput. We also focus on the performance of the
macrocell users and the cell edge users because macrocell
users have a priority over femtocell users.
We evaluated 100 realizations and compared the proposed
scheme with several comparison schemes: Femtocell 3-sector,
FF-3, and NF-3. Femtocell 3-sector scheme is similar
to the proposed method except sub-band A is excluded in
edge zones. In FFR-3 scheme, FFR is applied to the
macrocell users, and femtocell users are randomly assigned
from fll fequency band. On the other hand, FFR is not
adopted for macrocell users in NoF-3 scheme. In Femto
3-sector, the amount of sub-carriers allocated to femtocell
users is three times in {B, C, D} . Meanwhile, for FFR-3 and
NoF-3, the amount of total available sub-carriers for
femtocell users is three times of the fll band, as equal to total
numbers of sub-bands like the proposed scheme for fair
comparison. These features are summarized in Table 1. The
Amount column implies the amount of used sub-carriers
denoted by the sub-carrier assignment parameter [.
The simulation parameters are listed in Table II. The
overall network is composed of two-tier 19 macrocells, and
femtocells are randomly deployed over the macrocells. All
the BSs are operated by the OFDMA technology. We vary the
number of femtocells fom 30 to 180 in one macrocell
coverage to change simulation environment. The macrocell
TABLE I SUMMARY OF PROPOSED AND COMPARISON SCHEMES
Macrocell user F emtocell user
Schemes
Frequency Amount Frequency Amount
Proposed FFR
Divide center and
edges (Fig. 3)
Femto
Exclude
3-sector
FFR I sub-band A in 3
edge
F-3 FFR Random
NoFFR-3 Random Random
Note: Amount column implies value of
L
", P,k and
LxLr (n_ Pr,k for
macrocell and femtocell users, respectively.
TABLE II SIMULATION PARAMETERS
Parameters
Macrocell Femtocell
Number of cells 19 (two-tier)
30-180
Imacrocell
Cell coverage 280m 30m
BS transmit power
FFR: 15,22 W
20mW
w/o FFR: 20 W
Number of users in one macro cell
180 180
coverage
Channel Bandwidth 5 MHz
FFT size 512
Number of occupied sub-carriers 300
Sub-carrier spacing 15 kHz
White noise power density -174 dBmlHz
Size of center zone 0.63 of macro cell coverage
PLoutdoo, 28+35IoglO(d
Channel model Rayleigh fading,
PLjndoo, 38.5+201og'0(d)+Lwall
path loss, shadow fading(8dB)
Lwall 7 dB, if din ( 0, I 0]
LwolI' 10 dB, i f din (10,20]
Lwol" 15 dB, if din (20,30]
and femtocell users are randomly distributed in the overall
network. The macrocell and femtocell users are randomly
allocated the available sub-carriers fom the designated range
in each scheme. When the FFR is applied to macrocell user, a
half of the fll band is assigned for center zone, the other for
edge zone. An omni-directional antenna and three sector
antennas are installed at a macrocell BS, for center zone with
transmit power of 15 W and for edge zone with 22 W,
respectively.
On the other hand, if the FFR is not used by macrocell user,
the transmit power of a macro BS is 20 W. All the femtocells
use 20 mW as the transmit power. The different channel
model is used for indoor and outdoor, where the path loss is a
dominant factor. Then, we fnd out the downlink received
SINR values for each user and each sub-carrier. Using this
value, the throughput is calculated via users located in the
central serving macrocell of 19 cells.
B. Numerical Results and Comparison
Fig. 3 shows the throughput of macrocell users located in
the macrocell coverage as the number of femtocells varies. In
proposed scheme, the femtocell users can get the sub-bands
which are not used by macrocell users at each zone. So, the
interference between macrocell users and femtocells is
greatly avoided. The femtocells interfere with macrocell users
less than FFR-3 and NoFFR-3. This result is proper for
assumption, that is, macro cell users have a priority over
femtocell users. However, Femto 3-sector has slightly higher
throughput than proposed scheme in macrocell users'
throughput, due to sub-band A was not used in edge regions
for femtocells. This means that macrocell users suffer less
interference fom femtocell users.
In Fig. 4, it describes the throughput of total users located
at the edge zone only. In the OFDMA cellular network,
currently, the performance of the edge zone is poor due to the
inter-cell interference. In the proposed scheme, allocating the
more sub-carriers to femtocells which are located in the edge
zone, the throughput of the edge zone is improved. The
reason that Femto 3-sector has low throughput in edge zones
is due to the exclusion of sub-band A in the available
sub-bands Ix' To conclude, proposed scheme can efectively
reduce interference in the edge zones and can utilize more
sub-bands for femtocell users.
Fig. 5 depicts total throughput of total users in both the
macrocell and femtocells. Even though proposed scheme has
slightly lower throughput for macrocell users than Femto 3-
sector in Fig. 3, it has the much higher total throughput.
FFR-3 shows similar performance, however, throughput of
macrocell users is relatively low due to interference fom
femtocell as depicted in Fig. 3.
As shown in these results, our proposed scheme enhances
the overall throughput with considering macrocell and
femtocells. Especially, the throughput of cell edge users and
macrocell users are much enhanced in proposed scheme.
V. CONCLUSIONS
Femtocell systems are promlsmg to provide higher
throughput service with low cost in wireless cellular systems.
V
However, there is interference problem how to allocate
fequency bands to femtocell because of co-channel operation
between macrocell users and femtocell users. In this paper,
we defne optimal sub-band allocation problem and propose
an effcient resource allocation scheme in the OFMDA
femtocell systems using the 3-sectored FFR. The proposed
scheme enhances overall throughputs, especially for both
macro cell user and cell edge users. It is benefcial for the
OFDMA network such as the L TE, where the 3-sectored FFR
is applied. As a result, in case of using 3-sectored FFR for
macrocell, the premise which allocates fequency sub-bands
in practically disjoint ways is quite usefl for femtocell
networks.
,

... . . . . . . . . . . ..................... .. ... .. ... .. . .... . . . . . . . ....... .............
. .
. . .
. . .
. . .
. .
g ..........................................
. .
............... ................,- ...............,...............
.
.
g ...............................

................................

.. . .
. . .
. . .
. . .
. . .
. .
.
.
,
.
"
.
- - - - - .=P - ro - po-s -d -- - I
.
.

. .
Femto 3-sector
-FFR-3
-- NoFFR-3

_
Number of Femtocells
Fig. 3. Total throughput of macrocell users located in center zones and
edge zones as the number of femtocells varies caption.

_ - - - - - - - - - - - - - - - -...............:.- - - - - - - - - - - - - - - ---------------- --.......... ..

, ...............

.................

.................................

........
.

. .
.
.
'

............... L..............................


i



...........................
.... .
,
................ ..... .. ........


.
........ ................. ................J........
. .
.
.



Proposed
Femto 3-sector
-
6FFR-3
NoFFR-3

_
Number of Femtocells
Fig. 4. Throughput of macro and femtocell users located only in the edge
zones as the number of femtocells varies.
.............. ............... ........... .... ............... ............ ...
.............. ........ .......


. .....,..... . .~........ . .......... ........... .
.



. .
......., . .... ............................ ...................... ........... ....



Femto 3-sector
, ....... . ........... ........... .

..................... ---
-
6
-
FFR
-
3
1

,
Number of Femtocells
NoFFR-3

_
_

Fig. 5. Total throughput of macrocell and femtocell users in an entire

macrocell as the number of femtocells varies.
REFERENCES
[ I ] V. Chandrasekhar, 1. Andrews, "Femtocell Netwoks: A Survey," IEEE
Commun. Mag., vol. 46, no. 9, pp. 59-67, Sept. 2008.
[2] L. Ho, H. Claussen, "Effects of User-Deployed, Co-Channel
Femtocells on the Call Drop Probability in a Residential Scenario,"
IEEE Interational Symposium on Personal, Indoor and Mobile Radio
Communications (PIMRC), Sept. 2007.
[3] H. Lei, L. Zhang, X. Zhang, and D. Yang, "A Novel Multi-cell
OFDMA System Structure Using Fractional Frequency Reuse," IEEE
Interational Symposium on Personal, Indoor and Mobile Radio
Communications (PIMRC), Sept. 2007.
[4] M. Assaad, "Optimal Fractional Frequency Reuse (FFR) in
Multicellular OFDMA System," IEEE Vehicular Technology
Conference (VTC), Sept. 2008.
[5] S. Yeh et aI., "WiMAX Femtocells: A Perspective on Network
Architecture, Capacity, and Coverage," IEEE Commnu. Mag., vol. 46,
no. 10, pp. 58-65, Oct 2008.
[6] D. Lopez-Perez et aI., "Interference Avoidance and Dynamic
Frequency Planning for WiMAX Femtocells Networks," IEEE
Interational Conference on Communication Systems (ICCS), Nov.
2008.
[7] H. Zeng, C Zhu, and W. Chen, "System Performace of Self-Organizing
Network Algorithm in WiMAX Femtocells," ACM Interational
Conference on Wireless Interet (W CON), Nov. 2008.
[8] K. Sundaresan, S. Rangarajan, "Effcient Resource Management in
OFDMA Femto Cells," ACM Interational Symposium on Mobile Ad
Hoc Networking and Computing (MobiHoc), May 2009.
[9] T. Kim, T. Lee, "Throughput Enhancement of Macro and Femto
Networks By Frequency Reuse and Pilot Sensing," IEEE Interational
Performance, Computing and Communications Conference (IPCCC),
Dec. 2008.
[10] Lee, D., Xu, C, Mayekar, U, Mohile, M., "Frequency reuse factor vs.
pathloss exponent and sectorization," Wireless Applications Digest,
1997., IEEE MTT-S Symposium on Technologies
U