A Self-configuration Scheme for Power and

Bandwidth Assignment in Femtocell Networks

Wang ChengLi HongyanLi JiandongMa Yinghong
State Key Laboratory of Integrated Service Networks
Xidian University
Xian P.R.China
chengwang@stu.xidian.edu.cnhyli@xidian.edu.cnjdli@xidian.edu.cnyinghongma@yahoo.com.cn
AbstractFemtocells can offload the traffic of macrocells and
satisfy the big high-data-rate wireless communication
requirement of indoor users. Due to the large number of
fmetocells deployed by users, a significant challenge this new
access technology brings to us is coordinating the cross-tier
interference. Much special attention has been paid to solve the
problem. This paper presents a novel strategy to configure
femtocell users power and bandwidth jointly and an algorithm
to estimate the connection QoS requirement. The strategy can
effectively reduce the cross-tier interference while ensuring high
spectrum efficiency. Also simulations are given to show the effect
of the strategy.
Keywords-Femtocells; formatting; self-configuration; ICIC;
SON; cross-tier interference

I.

INTRODUCTION

Femtocell is a new kind of technology with a smaller

coverage and higher spectrum efficiency for indoor radio
network access. They provide a plug and play service to users
and can be purchased in low cost. Due to these advantages,
femtocells will be potential dominant radio access technology
in future cellular networks.
However these benefits will also bring many challenges to
femtocells[1][2]. One of the key challenges is to coordinate the
interference among the macrocells and a great deal of
femtocells overlapping together. Particularly, in CSG (Closed
Subscriber Group) mode only users registered as the
femtocells CSG user can access it, areas that femtocells
cover will be dead zone to non-CSG users. On the other hand,
since the location and time of femtocells are totally decided by
users, it is difficult for operators to manage and maintain
femtocell like macrocells. For this reason, femtocells must
work in self-optimizing, self-configuration and self-healing
manner. The ICIC strategy is necessary to guarantee users
QoS and works in a SON way.
Many existing papers have proposed some method to solve
the cross-tier interference problem.
In [3][6][7][10], they all paid the key attention on spectrum
sharing to avoid the cross-tier interference. In [2], they focused
on frequency planning to avoid the interference of macrocells
and femtocells through fractional frequency reuse (FFR). In [6],
David Lpez-Prez analyzed the interference based on a
realistic OFDMA macrocell and femtocell scenario.

Power control is considered in [4][8][9] for cross-tier

interference avoidance. In [3], two interference avoidance
strategies (open-loop and the closed-loop control) were
proposed to adjust the transmit power of femtocell users.
Reference [7] provided a utility function based on SINR to
control power at femtocells. In [8], they studied several power
control schemes and the deterioration in macrocell
performance.
In [5], Sundeep Rangan proposed a novel interference
control method in the scenario that macrocell bandwidth is
divided into subbands, and also an adaptive method to allocate
power across the subbands.
In this paper, we proposed a new scheme with the
following characteristic for interference avoidance among
macrocells and femtocells

A novel method to configure power and bandwidth

jointly;

On the basis of users QoS requirement, radio resource is

assigned to both femtocells and femtocell users;

Game theory is incorporated to solve in a distributed way;

Estimation method is used to evaluate the connection QoS

requirement in femtocells.

The remainder of the paper is organized as follows. System

model is introduced in Section IIIn section III, we proposed
a novel strategy to configure the power and bandwidth jointly.
In section IV, an estimation method to evaluate the connection
QoS requirement is proposed. Section V shows our simulation
results, and Section VI concludes this paper.
II.

SYSTEM MODEL

In this paper, we consider the WCDMA system model.

Here each macrocell has three sectors. The entire spectrum is
divided into several subbands. In each sector, some subbands
are already used by macrocells. Femtocells can choose one or
more subbands, but these subbands must be adjacent.
Femtocells are randomly distributed in a macrocell to provide
indoor wireless coverage to femtocell users. Macrocell and
femtocell users are randomly distributed. The CSG mode has
not been considered in this paper.

III.
A.

SELF-CONFIGURATION FOR POWER AND BANDWIDTH

Self-configuration Strategy

1) Strategy 1: In order to alleviate the interference between

macrocell and femtocell users, we surely first consider dividing
the spectrum into two orthogonal parts. One is for macrocell
users only, another is for femtocell users. Each femtocell will
randomly choose one or more adjacent subbands as its uplink
or downlink work spectrum:
a) Fixed subbands are allocated for macrocells
b) Subbands are randomly allocated to femtocells based
on the connection QoS requirement
c) Subbands allocated to femtocells are orthogonal to
macrocells
2) Strategy 2: Strategy 1 gives a simple method to allocate
the system spectrum resource between macrocells and
femtocells. However, interference in subbands has not been
considered, and actually, subbands with the lowest interference
should be chosen by femtocells to work
a) Fixed subbands are allocated for macrocells
b) Subbands with the smallest Interference are allocated
to femtocells based on the connection QoS requirement
c) Subbands allocated to femtocell are orthogonal to
macrocells
3) Strategy 3: The method of spectrum orthogonal divided
directly reduces the interference among users in a two-tier
network environment, but it cannot utilize spectrum effectively.
Specifically, in the case of unbalanced traffic networks, this
problem will be more obvious. For this reason, macrocells and
femtocells should be equally permitted to use the same
subbands:
a) Fixed subbands are allocated for macrocells

d) Macrocells and femtocells are permited to use same

subbsnds
5) Strategy 5: As we know, Shannon equation shows the
basic method to calculate a users throughput. According to
this equation, we can find out that if wider bandwidth is used,
more interference will be generated by users. In short, the
power and bandwidth are inter-related and both aspects can be
considered to increase throughput. When a femtocell is close to
the center of macrocell, it can increase transmitting power,
while in the edge, it can choose more bandwidth to reduce the
interference to the poor signal receiving macrocell users.
a) Fixed subbands are allocated for macrocells
b) Femtocells must combine the spectrum and power
assignment in its self-configuration
c)

Femtocells must be configured jointly in a distribute

way
d) Macrocells and femtocells are permited to use same
subbsnds
B. Utility Function
In this section, we propose a novel self-configuration
algorithm in order to jointly allocate spectrum and power
resource. This algorithm is proposed on the basis of utility
function theory.
Here the utility function has two parts: reward and penalty.
Given the configuration method set M , we calculate the
throughput of user j when it works with power Pi and
bandwidth
[9]:

Bi (scheme i) by the modified Shannons equation

ri = Bi log 2 (1 + 0.5

b) Subbands with the smallest Interference are allocated

to femtocells

Pi
)
Ii

(1)
*

Given the users connection QoS requirement r and the

ri , we model the

c) Macrocells and femtocells are permited to use same

subbands
4) Strategy 4: Femtocells may use the subband regardless
of whether it is already used by macrocells. But how much it
will choose must accord with the connection QoS requirement.
In the urban environment with dense apartments, cross-tier
even multi-tier cells overlapping will exist. If a femtocell only
greedily chooses spectrum resource for its own connection QoS
requirement, network interference may be extremely high. In
our opinion, configuration information should be exchanged
between adjacent cells and the information should be
incorporated by a femtocell to complete self-configuration:
a) Fixed subbands are allocated for macrocells

users throughput when taking method-i

reward part as:

When ri r , more reward is given to encourage the user

to pursue its connection QoS requirement;

b) Subbands with the smallest interference are allocated

to femtocells based on the connection QoS requirement

When ri

c)
way

Femtocells must be configured jointly in a distribute

e ( ri ri ) 1
Reward ( Bi , Pi ) =
( r r* )
1 e i i
*

and

ri r *
ri > r *

(2)

are two constants which reflect how the user

*

expect to achieve r .The reward function denotes the payoff a

femtocell user take a certain scheme i:
*

> r * , the reward approximates to a constant.

The penalty function reflects the interference to other users

one user brings when taking method i:

Penalty ( Bi , Pi ) = ri r *
I
i

ri r *
ri > r

H t
f

( ) =

(3)

( Lf )t =

is a constant. It ensures that the user may have fewer

penalties together with scaling the penalty by the users
interference suffered I i .
*

When ri r , no penalty is given to user to encourage

him to pursue his connection QoS requirement;

When ri
user.

Given the femtocell fs user set F, we denote its total utility

value as:

U f = U ij

Femtocells should choose the configuration scheme which

can make U f maximize.
IV.

CONNECTION QOS REQUIREMENT ESTIMATE

Section III has proposed the joint self-configuration

algorithm. This section mainly focuses on the time when
femotocells do this self-configuration algorithm. In order to
solve the problem, we propose a new algorithm to estimate the
connection QoS requirement in femtocells.
We assume the femtocells will decide whether to do reconfiguration periodically. Now, let

R Hf and R Lf reflect the

high and low connection QoS requirement threshold of

femtocell f separately. To describe how to estimate the
connection QoS requirement, suppose that

( R Hf )treal and

( R Lf )treal denote the current value in time slice t,

( R Hf )tpredict and ( R Lf )tpredict denote the predicted value in time
slice t,

( R Hf )tstatistic and ( R Lf )tstatistic denote the statistic value

in time slice t.
The normalization of predicted error
can be modeled as follows:

( Hf )t and ( Lf )t

( R Lf )tpredict

(6)

(7)

+1
(( R Hf )tstatistic
) and

+1
(( R Lf )tstatistic
) have already been known. The predict value can

be molded as:
+1
+1
( R Hf )tpredict
= (( R Hf )tstatistic
) (1 + ( Hf )t +1 )

(8)

+1
+1
( R Lf )tpredict
= (( R Lf )tstatistic
) (1 + ( Lf )t +1 )

(9)

( Hf )t +1 = ( Hf )t

(10)

( Lf )t +1 = ( Lf )t

(11)

Here in the equation above denotes the degree of

correlation between the predicted value in time slice t and t+1.
Let

(5)

jF

( R Lf )treal ( R fL )tpredict

threshold in time slice t+1, we assume that

After given the reward function and penalty function, we

molded the whole utility function as:
(4)

( R Hf )tpredict

To predict the high and low connection QoS requirement

> r * , some penalty must be given to the greedy

U i j = Reward ( Bi , Pi ) Penalty ( Bi , Pi )

( R Hf )treal ( R fH )tpredict

( R Lf )t denote the current throughput femtocell f

provide to its users, we can propose the following reconfiguration algorithm according to the connection QoS
requirement prediction interval (( R f

L t +1
predict

+1
), (( R Hf )tpredict
) :

+1
+1
) (( R Lf )tpredict
), (( R Hf )tpredict
)

L t

When ( R f

re-configuration is not necessary;

z

+1
+1
) (( R Lf )tpredict
), (( R Hf )tpredict
)

L t

When ( R f

Femtocells need to re-configure the transmit power and

bandwidth.
V.

SIMULATION

In this section, we have conducted simulations to evaluate

the performance of our proposed spectrum configuration and
connection QoS requirement estimated algorithm.
A. Simulation Scenario
An overlapping urban cellular network is considered here. As
shown in figure 1, we consider a cluster of cellular with 19 cells
and three sectors per cell. Femtocells are randomly distributed in a
macrocell to provide indoor wireless coverage to femtocell users.
Each macrocell has a diameter of 1732 m and each femtocell has a
diameter of 20 m. Macrocell and femtocell users are randomly
distributed. Path loss is modeled based on 3GPP simulation
assumptions and parameters for HeNB[11]:

PL ( dB ) = 15.3 + 37.6 log10 R + Low

R is the distance between a user and a macrocell or

femtocell BS. Lo w is the penetration loss of an outdoor wall,

55

which is 10dB in our paper. Total macrocell BS TX power is

46dBm and the Max Tx power in HeNB is 23dBm . In the
simulation, we have totally generated 100 instance and the
results are the average of 100 points

Re-configuration Times in One Period

50

B. Simulation results
In this simulation we compare the performance of the five
strategies proposed in Section II from the perspective of UE
average throughput among macrocells and femtocells. In
strategy 5, we use the algorithm based on the utility proposed
in our paper. Simulation results show that the UE average
throughput drops obviously with the increasing number of UEs
in each femtocell. But the curve labeled strategy 5 is always
above the other four curves. So we can conclude that the joint
spectrum and power assignment perform better than others.

35

30

25

20

15

10

x 10

10

15

20

Figure 2. Performance of the proposed methods using the connection QoS

requirement

Strategy2
Strategy3

Time in One Day

Strategy1

Strategy4

UE Average Throughput(bps)

40

3.5

45

Strategy5

CONCLUSION

2.5

In this paper, we consider the scenario of macrocells and

femtocells overlapping together. We proposed a novel strategy
to jointly configure femtocell users power and bandwidth, as
well as an algorithm to estimate the connection QoS
requirement. With the simulation conducted, we have shown
that our mechanisms can significantly improve the system
efficiency.

1.5

ACKNOWLEDGMENT
0.5

10

UEs per HeNB

Figure 1. performance of the five strategies proposed in Section II

Also we want to evaluate the performance of the

proposed methods using the connection QoS requirement
estimated from the perspective of the re-configuration times.
We simulate the arrival rate of the traffic in 24 hours. More
specifically, the indoor traffic is high between 7:00 a.m. and
9:00 a.m., 11:00 a.m. and1:00 p.m., 6:00 p.m. and11:00p.m..
Assuming that there are 50 femtocells in one cell and the reconfiguration period is 1 hour. In other words, without the
connection QoS estimation, the re-configuration times will be
50 in every period. But with the connection QoS estimation, as
simulation result in figure2 shows, only in the busy time,
femtocells re-configure frequently.

This work is supported by the National Science Foundation

(60972047), National S&T Major Project (2010ZX03005-003,
2011ZX03005-004-02, 2011ZX03005-003-03, 2011ZX03004003), National Science Fund for Distinguished Young Scholars
(60725105), National Basic Research Program of China
(No.2009CB320404), Program for Changjiang Scholars and
Innovative Research Team in University (IRT0852), the 111
Project (B08038) and State Key Laboratory Foundation
(ISN1002005).
REFERENCES
[1]

[2]

[3]

[4]

V. Chandrasekhar, J. G. Andrews, and A. Gatherer, Femtocell

networks: a survey, IEEE Comm. Magazine, vol. 46, no. 9, pp. 5967,
Sept. 2008.
P. Kulkarni, W. H. Chin, and T. Farnham, Radio resource management
considerations for lte femto cells, SIGCOMM Computing
Communication Review, vol. 40, no. 1, pp. 2630, 2010.
C. Y. Oh, M. Y. Chung, H. Choo, and T.-J. Lee, A Novel Frequency
Planning for Femtocells in OFDMA-Based Cellular Networks Using
Fractional Frequency Reuse, in Computational Science and Its
Applications ICCSA 2010, Fukuoka, Japan, Mar. 2010, pp. 96106.
H.-S. Jo, C. Mu, J. Moon, and J.-G. Yook, Interference mitigation
using uplink power control for two-tier femtocell networks, IEEE Trans.
Wireless Commun., vol. 8, no. 10, pp. 49064910, Oct. 2009.

[5]

[6]

[7]

[8]

S. Rangan, Femto-Macro Cellular Interference Control with Sub-band

Scheduling and Interference Cancelation, Available at arXiv:
1007.0507.
R. Y. Chang, Z. Tao, J. Zhang and C.-C. Kuo, A graph approach to
dynamic fractional frequency reuse (FFR) in multi-cell OFDMA
networks, in Proc. IEEE ICC 09, pp.16, June 2009.
D. Lopez-Perez, A. Valcarce, G. de la Roche, and J. Zhang, OFDMA
femtocells: A roadmap on interference avoidance, IEEE Commun.
Mag., vol. 47, no. 9, pp. 41 48, September 2009.
V. Chandrasekhar, J. G. Andrews, T. Muharemovic, Z. Shen, and A.
Gatherer, Power control in two-tier femtocell networks, IEEE Trans.
Wireless Commun., vol. 8, no. 8, pp. 4316 4328, August 2009.

[9]

N. Arulselvan, V. Ramachandran, and S. Kalyanasundaram, Distributed

power control mechanism for HSDPA femtocells, in Proc. IEEE VTC,
2009.
[10] M. Andrews, V. Capdevielle, A. Feki, and P. Gupta, Autonomous
spectrum sharing for mixed LTE femto and macro cells deployments,
in Proceedings of IEEE Infocom, 2010.
[11] Alcatel-Lucent, picoChip Designs, Vodafone, R4-092042: Simulation
assumptions and parameters for FDD HeNB RF requirements in 3GPP
RAN 4, May 2009.