Content uploaded by Ibrahim shglouf Shellmani
Author content
All content in this area was uploaded by Ibrahim shglouf Shellmani on May 18, 2017
Content may be subject to copyright.
Efficient Femtocell Deployment Under Macrocell
Coverage in LTE-Advanced System
Ibrahim Shgluof, Mahamod Ismail, and Rosdiadee Nordin
Department of Electrical, Electronic and System Engineering
Universiti Kebangsaan Malaysia
43600 UKM Bangi, Selangor, Malaysia
Email: shellmani@ieee.org, mahamod@eng.ukm.my, adee@eng.ukm.my
Abstract— Due to the increasing demand for higher data rates by
mobile users, the deployment of femtocells under macro coverage
has been adopted by 3GPP LTE-Advanced systems as one of the
main proposals to enhance the network perfor mance in the
respects of coverage and capacity improvement especially for
indoor users. However the random deployment of femtocells
raise the issue of interference between macrocell and femtocells
and its impact on coverage and capacity of the network which
become s a critical challenge for LTE-Advanced system designers.
In this paper we study and analyze the efficiency of different
deployment locations of femtocells. Simulations show that by
deploying femtocells at an appropriate location, the coverage
almost doubled while the data rate received by the end user
enhanced by 28% on average as compared to random
deployment.
Keywords-3GPP, capacity, coverage, femtocell, interference, LTE,
macrocell.
I. INTRODUCTION
Due to the increasing demand for higher data rates and for
the quality of service by home users, it became a necessity for
wireless communication operators today to develop suitable
design criteria to meet the rising expectation by users from
current and future network operators. The coverage and data
rates provided by existing wireless communication systems
have been pushed to the limit especially for indoor users; and
are not appropriate to meet the rising demand for multimedia
services. As the distance between the transmitter and receiver
of a wireless system gets closer; this results in an enhancement
in the capacity of a wireless link and establishes a higher
quality links in terms of data rates and more spatial reuse [1].
The notion of femtocell deployment was presented to Long
Term Evolution (LTE) and currently is under review by the
3rd Generation Partnership Project (3GPP) to improve indoor
coverage and to enhance system capacity [2]. The deployment
of femtocells or home evolved NodeB (HeNBs) as known in
LTE-Advanced under macrocell; also known as evolved
NodeB (eNB) coverage area is a challenging criterion for the
designing and dimensioning of LTE-Advanced networks.
HeNBs have small coverage areas (10-15m) with small
transmit power in the range of (1-100mW) and connect to the
cellular core network via an IP-based backhaul, such as
Digital Subscriber Line (DSL) or cable connection. There are
many challenges that operators must resolve before a
successful deployment of a HeNBs network; like interference
between HeNB networks and macrocell (eNB), also
interference between HeNBs themselves. This interference
could draw back the benefits of HeNBs deployment and this
could cause some impairment to the overall performance of
the LTE-Advanced system. The deployment of large numbers
of HeNBs indoors under eNB coverage and the randomness of
the distribution of their locations make the network
optimization a hard task for designers in terms of interference
and handover processes between eNBs and HeNBs. Hence,
many studies have been carried out analyzing the effect of
HeNBs interference on eNBs and also many efforts have been
made trying to develop the best algorithm for
inbound/outbound handovers.
For multiple deployments of HeNBs under eNB coverage,
the main goal is to reduce the coverage overlaps of adjacent
HeNBs and also gaps between them, as well as balancing the
workload amongst HeNBs. When user equipment (UE), such
as a mobile device, enters the edge of HeNBs network, it
initiates and performs an inbound handover (eNB to HeNB).
This will result in traffic offloading from eNB and core
network. However, if UE is located between HeNBs there may
be a coverage gap where UE is not covered by any HeNB, and
hence an outbound handover (HeNB to eNB) takes place
which in turn increases frequent and unnecessary handovers.
On the other hand UE may be covered by more than one
HeNB (coverage overlaps), which causes an increase in
interference. Thus in designing femtocell networks; both
coverage gaps and overlaps between HeNBs must also be
given an important consideration [3].
In this paper we focus on analyzing the efficiency of
different locations for HeNBs when they are deployed under
eNB cell coverage; the efficient locations in terms of their
distance from eNB. The efficient location that can meet the
best design criteria in terms of the minimum allowable signal
power received by User Equipment (UE), throughput and the
required coverage for the home user. The rest of the paper is
organized as follows: Section II presents an overview of
978-1-4673-2088-7/13/$31.00 ©2013 IEEE 60
related work. Section III describes LTE femtocell system
architecture. Section IV describes and discusses the simulation
scenarios. Section V presents discussions and simulation
results. And finally our conclusion is summarized in section
VI.
II. RELATED WORK
Prior research on the performance of HeNBs when they are
deployed under eNB coverage area has been carried out by
many authors. Their work has mainly focused on analyzing
interference and its impact on capacity, coverage and
inbound/outbound handover. Oh et al. [4] studied the effect of
interference on inbound handover at a certain deployment
location of a cluster of HeNBs. This work showed that an
inbound handover cannot occur when HeNB1 is located at 225
m from eNB as the UE cannot decode the MIB/SIB1 system
information of that HeNB at this distance. Namgeol et al. [5]
investigates system performance of femtocells based on
different environmental factors like number of walls and their
structure which determines the appropriate distance between
an eNB and a HeNB to guarantee the minimum requirement of
spectral efficiency. The author concluded that the required
separation distance from eNB is 50~200m. Jang et al. [6]
proposed a self-optimization algorithm for a single HeNB
coverage deployed under eNB. His proposal was based on
handover requests from UEs. When an outdoor UE requests
Handover to HeNB; the transmit power of HeNB is reduced
and vice versa when an indoor UE requests handover to eNB.
Thus minimizing the number of unnecessary handovers and
optimizing camp on the area. Jo et al. [7] proposed a coverage
coordination scheme for two tier femtocell networks. In this
scheme the femtocell adjust their transmit powers based on
their measure of down link signal and interference powers.
This proposal showed that it can provide adequate femtocell
indoor coverage and prevents leakage of indoor coverage to
outdoor macocell. Claussen et al. [8] proposed a femtocell
coverage algorithm that can control the transmitted power
based on mobility events for indoor and passing by outdoor
users. It was shown that this algorithm can reduce unnecessary
mobility events and also improve indoor coverage. However
none of those authors addressed or investigated the efficient
location of a cluster of HeNBs when deployed under eNB
coverage. The efficient location that defines the distance
between a cluster of HeNBs and eNB. This separation distance
can translate how severe is the interference between eNB and
HeNBs which in turn can determine the optimized coverage
and capacity of the system.
III. LTE FEMTOCELL SYSTEM ARCHITECTURE
For the E-UTRAN HeNB architecture, the discussions for
the LTE femtocell standards are still undergoing in 3GPP, in
NGMN Alliance and in the Femto Forum. Up to date the final
architecture has not been agreed on yet; but there is a
unanimous decision to keep it in the flattest manner due to the
adoption of ‘all-IP’ networks in the LTE standards. Another
debate is still in discussion regarding the need for signaling
interfacing elements or should femtocells be supported
Fig. 1. Overall E-UTRAN architecture with deployed HeNB GW. [10]
directly by the evolved packet core (EPC) [9]. Fig. 1 shows
the overall E-UTRAN architecture with deployed HeNB GW.
It can be seen that there is a set of S1 interfaces between
HeNBs and EPC. The presence of Home eNB Gateway
(HeNB GW); is for expanding the S1 interface between
HeNBs and core network, to scale up the deployments of a
large number of HeNBs.
It is assumed that HeNB GW worked at (C-Plane),
especially the concentrator of the S1-MME. And The S1-U
interface can be terminated at HeNB GW or by a direct logical
connection between the HeNB and the S-GW via the logical
(U-Plane). Fig. 2 shows the E-UTRAN HeNB Logical
architecture. To enhance the integration between HeNBs and
LTE macrocell networks; the MME should interface with
HeNB GW as an eNB, thus the HeNB GW will appear to the
HeNB as an MME between the HeNB and the EPC. Due to
the expected deployment of thousands of femtocells in an
LTE macrocell, interfacing between HeNB GW and operator’s
O&M system may be required for configuration and control
[10] [11].
The S1 interface between HeNBs and the core network
should be the same whether the HeNB is connected to the EPC
directly or via a HeNB GW. Either option can be chosen for
LTE femtocell system architecture whether based on
deployment of HeNB GW or not. The HeNB GW shall serve
Fig. 2. E-UTRAN HeNB logical architecture [10]
61
in a manner that mobility to other cells would not necessarily
require inter-MME handovers. HeNB GW can also provide
accumulation function for the S1-MME.
The S1-U interface can be a direct tunnel or can be
aggregated by the HeNB GW. On the other hand, the HeNB
GW may provide support for multiplexing of (U-Plane); to
enhance the system in limited bandwidth links. HeNBs should
support the same functions as those supported by an eNB; also
procedures should be the same when they run between HeNBs
and the EPC or between an eNB and the EPC [10].
As it can be seen from the E-UTRAN HeNB architecture,
and HeNB logical architecture; Fig. 1 and Fig. 2, the related
discussion was focusing on the deployment of HeNB GW and
the need for a set of S1 interfaces between HeNBs and EPC.
The debate was on whether HeNBs should be directly
supported by EPC, or should they be aggregated by HeNB
GW via S1 interfaces to utilize the limited bandwidth links.
However The E-UTRAN HeNB architecture and the relevant
discussion did not highlight the deployment location of
HeNBs nor their possible numbers when integrated into the
system. The location of HeNBs and their distance from eNB
can dictate the impact of high interference from eNB signal on
HeNBs, and this interference in turn can define the overall
coverage and capacity of the LTE-Advanced system.
IV. SIMULATION SCENARIOS
In this paper, we assumed that the same carrier frequency
and the same carrier bandwidth are used for both macro eNB
and HeNBs. The rest of our simulation parameters are shown
in Table I for eNB and HeNB [12]. The relationship between
those parameters is as follows:
( )
pathpenetantTX
LLGG +−=
(dB) (1)
Where:
TX
G
is the gain of the transmitted power from eNB
or HeNB to UE.
ant
G
is the antenna gain of eNB or HeNB.
penet
L
is the wall penetration loss and is equal to
wall
LR +×7.0
, and
wall
L
is the penetration loss of an
external wall which has a typical value of 10dB.This
applies only to HeNBs.
path
L
is the path loss from eNB or HeNB to UE.
TXTXRX
GPP +=
(dB) (2)
Where:
RX
P
is the received power from eNB or HeNB by
UE.
TX
P
is the transmitted power of eNB or HeNB.
The measure of the signal quality of the user data in LTE-
Advanced is determined by the ratio of energy per bit to the
spectral noise density (
ob
NE
). In our simulation using
Matlab we assumed a minimum
N
E
o
b
value of 9 dB; which
is a reasonable value for voice calls (speech) which can be
considered to represent the packet, also an UE speed of 1 m/s
was also assumed to simulate the typical pedestrian movement
TABLE I
eNB and HeNB System Parameters
eNB
Parameter
Typical value
Cell Radius
500 m
Carrier Frequency
2000 MHz
Carrier bandwidth
10 MHz
Distance-dependent path loss
15.3+37.6log10R
Antenna Gain after cable loss
14 dBi
eNB Tx power
46 dBm
UE Noise Fi gure
9 dB
User Data Rate
1024 Kbps
HeNB
Parameter
Typical value
Cell Radius
10 m
Carrier Frequency
2000 MHz
Carrier bandwidth
10 MHz
Distance-dependent path loss
38.46 + 20log10R
Antenna Gain after cable loss
3 dBi
HeNB Tx power
20 dBm
Penetration Loss
10 dB
The ratio of energy per bit to the spectral noise density
N
E
o
b
value and the data rate have been calculated using the
following formulas respectively:
( )
( )
oRXu
o
bIPRWNE ×=
(3)
( ) (
)
oobRX
INEPWRb ××=
(4)
where:
W
is the bandwidth of the system and u
Ris the user
data rate.
b
R
is the bit rate received by UE.
o
I
is the sum of the interference powers in Watts of eNB and
all HeNBs (excluding own signal) and is computed as follows:
( )( )
( )
( )
∑
=
+=
5
1i
oRXeNBUEo
NPI
iHeNB
(5)
( )( )( ) ( )( )
( )
( )
( )
∑
≠
=
++=
5
1
iK
KHeNB
K
oRXeNBRX
iHeNBUE
o
NPPI
(6)
Equation (5) defines all interferences from all HeNBs when
UE is attached to eNB and the noise power (
o
N
) of the
system is included. (6) indicates the interference from eNB
and the interference from all other HeNBs except for the
HeNB that UE are served by, where
=i
1 to 5. The noise
power (
o
N
) is also included here.
The percentage of coverage area provided by each HeNB
for different simulation scenarios is calculated as follows:
( )
%100
dim
×= HCC
sen
(7)
where:
sen
C
is the actual coverage that can be provided by
HeNBs for assumed scenario and;
dim
H
is the house
dimension which in our scenario is equal to10 m.
Fig. 3 shows the simulation environment applied in this
paper which follows the work carried out by Oh et al. [4]. The
environment is a typical suburban road. A cluster consisting of
62
5 HeNBs and horizontally were located at different distances
away from the macro eNB starting at 250 m to 330 m for
HeNB1 in steps of 20 m for each move. Every house in the
environment has a 10 m × 10 m in dimensions with every
HeNB deployed at the center of each house. HeNBs were
located every 20 m and the user is moving horizontally away
from eNB and passing by at a constant distance of 1 m in front
of every house.
In our first scenario HeNB1 was located at 250 m away
from eNB and as the distance between every two adjacent
HeNBs is 20m, hence HeNB5 is located at 330 m away from
eNB in the first scenario. As HeNB1 is moved away from
eNB in an increment of 20 m; we investigate the signal power
received by UE from eNB and HeNBs for each scenario. We
analyze the size of coverage area provided by each HeNB and
the average value of
N
E
o
b
which determines the data rates for
UE. It should also be noted that in our first scenario we
deployed a reference HeNB (HeNBref) on the left hand side of
HeNB1 by a distance of 20m, to show the effect of the high
signal power of eNB compared to low powers of HeNBs when
they are deployed too close to eNB.
In our simulations; a great deal of observation was given to
HeNB1 and HeNB5 as the first one suffers the greatest deal of
interference from eNB of all other HeNBs as it is the closest to
eNB location, and the later one is the closest to eNB cell edge
where there can be some coverage gaps around it due to the
weak signal of eNB near it’s cell edge.
V. RESULTS AND DISCUSSION
Table II presents data obtained from our simulation
scenarios. According to this data and as shown in Fig. 4, it is
obvious that HeNB1 at 250 m is too close to eNB in scenario1
and therefore does not provide a significant coverage for UE
(about 3m only); that is from 248 m to 251 m, which is only
30% of the presumed coverage of 10 m. And that is due to the
high transmit power of eNB compared to HeNB1. Although
the average value of
N
E
o
b
for HeNB1 received by UE in
scenario1 is about 9.96 dB which is above our assumption
value of 9 dB; but this only exists at the middle of the house
where the femtocell is placed. As UE moves away from the
centre, the received
N
E
o
b
is well below the threshold value.
Also HeNB2 and HeNB3 can only provide 60% and 80% of
Direction of UE Movement
Fig. 3. Simulation Environment
TABLE II
Scenario
1
2 3 4 5
(distance from eNB)
(250m) (270m) (290m) (310m) (330m)
HeNB1
C
30%
60%
80%
100%
100%
N
E
o
b
9.96 10.83 11.62 12.25 12.75
b
R (Mbps) 1.15 1.23 1.26 1.34 1.45
HeNB2
C
60% 80% 100% 100% 100%
N
E
o
b
10.80 11.56 12.16 12.63 12.97
b
R
(Mbps) 1.00 1.11 1.20 1.33 1.51
HeNB3
C
80% 100% 100% 100% 100%
N
E
o
b
11.56 12.16 12.63 12.97 13.51
b
R
(Mbps) 0.94 1.05 1.18 1.37 1.22
HeNB4
C
100% 100% 100% 100% 100%
N
E
o
b
12.17 12.63 12.97 13.50 13.94
b
R
(Mbps) 0.89 1.03 1.22 1.11 1.42
HeNB5
C
100% 100% 100% 100% 100%
N
E
o
b
12.79 13.19 13.75 14.31 14.46
b
R
(Mbps) 0.79 0.903 0.81 0.94 1.05
full coverage respectively. Only HeNB4 and HeNB5 can
provide 100% coverage in this scenario. Also by looking at
Fig. 4, we can see that the signal power from eNB is much
greater than the signal power of HeNBref which is located on
the left hand side of HeNB1 at 230m. This indicates that the
UE cannot decode the signal from HeNBref at this location;
and hence no inbound handover procedure can be executed in
this area. Therefore any displacement of the HeNB cluster to
the left hand side towards eNB location from this point will
severely degrades their performance in terms of capacity,
coverage and handovers.
As the distance between eNB and HeNB1 is increased to
270 m, the coverage of the HeNB1 has improved to 6 m (from
267 m to 273 m) which is out of presumed coverage area of 10
m, and the average value of
N
E
o
b
is at 10.83 dB. Therefore
the location of HeNB1 is still too close to eNB to make any
useful coverage. As 40% of the area of the house still has no
63
Fig. 4. HeNB1at 250 m from eNB location
coverage from HeNB1; thereby this may result in frequent and
unnecessary handovers. Also it can be noticed that the
coverage of HeNB2 and HeNB3 has improved to 80% and
100% respectively as shown in Table II. When HeNB1 was
moved away from eNB even further to 290m and then to 310
m, the data obtained showed that the size of the HeNB1
coverage area has improved to 8 m (from 286 to 294 m) and to
10 m (from 305 m to 315 m) respectively; which means that
HeNB1 can provide 100% coverage or in other words full
coverage for the house dimension at this location as shown in
Fig. 5. Also an improvement was observed in the average
N
E
o
b
value to approximately 12.25 dB which is more than
enough for voice calls. Also we can mention that the average
data rate that can be received by UE from HeNB1 coverage
area is estimated to be around 1.34 Mbps as shown in Table II.
Fig. 5. HeNB1at 310 m from eNB location
Fig. 4. HeNB1at 250 m from eNB location
Fig. 5. HeNB1at 310 m from eNB location
The improvement of coverage area of HeNB1 and the
enhancement of data rate obtained by UE is due to the fact that
the interference experienced by HeNBs; especially the closest
one from eNB is decreasing as the separation distance
increased between eNB and HeNBs. If HeNB1 is moved
further away from the eNB location to be set at 330m; while
HeNB5 is located at 410 m; Fig. 6 and data in Table II shows
that the coverage area of HeNB1, data rate obtained and
average
N
E
o
b
value have all improved. But we can also
observe that the
N
E
o
b
value received by UE from eNB in the
area between HeNB4 and HeNB5 which is the distance from
398 m to 402 m is below the minimum required value of 9 dB.
The user is supposed to be handed over to eNB as he/she
leaves the coverage area of HeNB4. The value of
N
E
o
b
received from eNB by UE in this area is greater than either of
the values received from HeNB4 and HeNB5. But on the other
hand the value of received
N
E
o
b
from simulation results was
found to be 8.69 dB which is less than the assumed threshold
value of 9 dB as shown in Fig 6. This is due to the fact that the
eNB signal near the cell edge is weaker and is under the
impact of HeNBs interference. Hence in this case the UE will
not be able to decode the system information from the
received signal. Therefore as a result we will have a gap in the
coverage of the system between HeNB4 and HeNB5 and the
call will be dropped as the UE reaches 398 m away from eNB
location.
We must also mention that HeNB1 can be located at 320
m away from eNB and still provide full coverage with high
data rate but again simulation results showed that the value of
N
E
o
b
received by UE from eNB between HeNB4 and
HeNB5 (388m to 392m) is just above the threshold value with
a small margin at 9.10 dB. Hence any variation in the signal in
this area will result in a dropped call. We can also notice in
Fig. 4, Fig. 5 and Fig 6 that the eNB signal rises up after the
location of HeNB5, and this is due to the disappearing of the
Fig. 6. HeNB1at 330 m from eNB location
64
impact of HeNBs interference on eNB signal as UE moves
away from HeNBs locations. Table II shows data obtained for
each HeNB of the assumed cluster of 5 HeNBs for each
simulation scenario.
VI. CONCLUSION
Deploying femtocells under eNB coverage area can solve
the problem of limited coverage indoors; thus enhancing the
data rates for end users. However, interference between
femtocells and macrocell arises when femtocells are deployed
randomly and this can in turn degrade the overall performance
of the LTE-Advanced system. It is shown that by deploying
femtocells at appropriate locations, this can mitigate
femtocell/macrocell interference, hence enhancing the system
capacity.
ACKNOWLE DGMENT
This research was supported by Universiti Kebangsaan
Malaysia under Grant UKM-OUP-2012-182.
REFERENCES
[1] Vikram Chandrasekhar and Jeffrey G. Andrews, “Communications
Magazine, IEEE , vol.46, no.9, pp.59-67, September 2008.
[2] 3GPP TS 36.300, “Evolved Universal Terrestrial Radio Access
(EUTRA) and Evolved Universal Terrestrial Radio Access Network
(EUTRAN) Overall description; Stage 2,” Rel. 9, v9.2.0, Dec. 2009.
[3] T. Ma, P, Pietzuch, “Femtocell Coverage Optimisation Using
Statistical Verification”. NETWORKING 2011 - 10th International IFIP
TC 6 Networking Conference, Valencia, Spain, May 9-13, 2011.
[4] S. Oh, H. Kim, B. Ryu, and N. Park, "Inbound Mobility Management on
LTE-Advanced Femtocell Topology Using X2 Interface," Computer
Communications and Networks (ICCCN), 2011 Proceedings of 20th
International Conference on , vol., no., pp.1-5, July 31 2011-Aug. 4
2011.
[5] O. Namgeol, S. Han, and H. Kim, "System Capacity and Coverage
Analysis of Femtocell Networks," Wireless Communications and
Networking Conference (WCNC), 2010 IEEE , vol., no., pp.1-5, 18-21
April 2010.
[6] S. Jang, Y. Lee, J. Lim, D. Hong, "Self-optimization of single femto-cell
coverage using handover events in LTE systems," Communications
(APCC), 2011 17th Asia-Pacific Conference on , vol., no., pp.28-32, 2-5
Oct. 2011.
[7] H. Jo, C. Mun, J. Moon, J.Yook, "Self-Optimized Coverage
Coordination in Femtocell Networks," Wireless Communications, IEEE
Transactions on , vol.9, no.10, pp.2977-2982, October 2010.
[8] H. Claussen, L.T.W. Ho, L.G. Samuel, "Self-optimization of coverage
for femtocell deployments," Wireless Telecommunications Symposium,
2008. WTS 2008 , vol., no., pp.278-285, 24-26 April 2008.
[9] S. Saunders, S. Carlaw, A. Giustina, R. Bhat, V. Rao and R. Siegberg J.
Holtzman, A. Sampath, Femtocells, Opportunities and Challenges for
Business and Technology, John Wiley & Sons Ltd, 2009 , pp. 68–73.
[10] 3GPP TS 36.300, “E-UTRA and E-UTRAN Overall Description.”
[11] L. Wang, Y. Zhang, and Z. Wei, "Mobility Management Schemes at
Radio Network Layer for LTE Femtocells," Vehicular Technology
Conference, 2009. VTC Spring 2009. IEEE 69th , vol., no., pp.1-5, 26-29
April 2009.
[12] 3GPP TSG RAN. R4-092042, “Simulation assumptions and parameters
for FDD HeNB RF requirements”, May 2009.
65