ENHANCED TECHNIQUES FOR INTERFERENCE MANAGEMENT IN LTE FEMTOCELL NETWORKS

Abstract

Femtocells are low power access points deployed indoor to alleviate indoor coverage problem.
The most severe problem related to femtocells is the interference. Interference can
be generated either between macrocells and femtocells (cross-tier) or among femtocells (cotier).
This thesis addresses the cross-tier downlink interference problem of LTE femtocell
systems. Firstly, the thesis surveys and compares some management schemes proposed in
the literature. This comparative survey shows that management schemes based on efficient
frequency allocation techniques and Fractional Frequency Reuse (FFR) can be a good solution
for managing interference in LTE femtocell systems. Secondly, the thesis describes the
cellular system model including cellular layout, indoor area modeling, propagation Pathloss
(PL) models, and finally the signal to interference plus noise ratio (SINR) and capacity mathematical
analysis used. Thirdly, the thesis proposes some Macro-Femto frequency allocation
schemes based on some well-known reuse schemes and FFR variations to allow coexistence
of both networks (macrocell and femtocell) and allocate resources efficiently between them.
These proposed schemes don’t require any coordination between the two networks. The
well-known schemes exploited are: Reuse-1, Reuse-3, Soft Frequency Reuse (SFR), Partial
Frequency Reuse (PFR), and Soft Fractional Frequency Reuse (SFFR). Fourthly, the evaluation
metrics used to evaluate system performance are described. These evaluation metrics
include average users’ capacity as a measurement of throughput, SINR maps and 10%-tile
SINR as a measurement of coverage, outage probability as a measurement of Quality of Service
(QoS) and finally Jain’s index and fairness ratio as a measurement of fairness. Fifthly,
simulation setup and simulation results are provided. A modified version of Vienna LTE
simulator is used as a powerful MATLAB simulation tool for simulating this thesis work.
Simulation results show that Reuse-1 scheme degrades most of network performance metrics
other than throughput for dense deployment of femtocells like coverage, QoS and fairness.
Reuse-3 and PFR schemes provide good performance in terms of coverage, QoS, and
fairness but poor performance in terms of throughput due to poor spectral efficiency. SFFR
scheme provides higher throughput performance rather than Reuse-3 and PFR but lower in terms of coverage and fairness. SFR scheme provides the best throughput performance in
dense deployment of femtocells and good performance for coverage, QoS and fairness. So
SFR scheme can represent the best tradeoff among different proposed schemes via different
evaluation metrics. Two optimal SFR interior radii are then calculated for two different
femtocell deployment densities using exhaustive search. The two calculated radii are found
to optimize throughput performance without significant degradation in fairness performance
and provide good coverage performance.

Full text

ENHANCED TECHNIQUES FOR INTERFERENCE
MANAGEMENT IN LTE FEMTOCELL NETWORKS
A THESIS
Submitted to the Graduate School of
Electronics, Communications and Computer Engineering ,
Egypt-Japan University of Science and Technology (E-JUST)
In Partial Fulfillment of the Requirements for the Degree
of
Master of Science
in
Electronics and Communications Engineering
by
Mahmoud Mohammed Selim
September 2012

ENHANCED TECHNIQUES FOR INTERFERENCE
MANAGEMENT IN LTE FEMTOCELL NETWORKS
Submitted by
Mahmoud Mohammed Selim
For The Degree of
Master of Science
in
Electronics and Communications Engineering
Supervision Committee: Signature
Prof. Mohamed El-Sharkawy, E-JUST ...........................
Prof. Hossam Shalaby, E-JUST ...........................
Dr. Mostafa El-Khamy, E-JUST ...........................
Prof. Hiroshi Furukawa, Kyushu University, Japan ...........................
Dr. Osamu Muta, Kyushu University, Japan ...........................
Examination Committee: Approved
Prof. Hossam Shalaby
Chairperson, Electronics and Communication Engineering Department,
E-JUST ...........................
Prof. Said El Nouby
Professor Emeritus, Electrical Engineering Department, Alexandria
University ...........................
Prof. Essam Sorour
Professor, Electrical Engineering Department, Alexandria University ...........................
Prof. Zen Kawasaki
Advisor of School of Electronics, Communications and Computer
Engineering, E-JUST ...........................
Dr. Masoud El-Ghoneimy
Associate Professor , Department of Electronics and Communications
Engineering, E-JUST ...........................
Vice President for Education and Academic Affairs
Prof. Ahmed Abo Ismail

SUMMARY
Femtocells are low power access points deployed indoor to alleviate indoor coverage prob-
lem. The most severe problem related to femtocells is the interference. Interference can
be generated either between macrocells and femtocells (cross-tier) or among femtocells (co-
tier). This thesis addresses the cross-tier downlink interference problem of LTE femtocell
systems. Firstly, the thesis surveys and compares some management schemes proposed in
the literature. This comparative survey shows that management schemes based on efficient
frequency allocation techniques and Fractional Frequency Reuse (FFR) can be a good solu-
tion for managing interference in LTE femtocell systems. Secondly, the thesis describes the
cellular system model including cellular layout, indoor area modeling, propagation Pathloss
(PL) models, and finally the signal to interference plus noise ratio (SINR) and capacity math-
ematical analysis used. Thirdly, the thesis proposes some Macro-Femto frequency allocation
schemes based on some well-known reuse schemes and FFR variations to allow coexistence
of both networks (macrocell and femtocell) and allocate resources efficiently between them.
These proposed schemes don’t require any coordination between the two networks. The
well-known schemes exploited are: Reuse-1, Reuse-3, Soft Frequency Reuse (SFR), Partial
Frequency Reuse (PFR), and Soft Fractional Frequency Reuse (SFFR). Fourthly, the evalu-
ation metrics used to evaluate system performance are described. These evaluation metrics
include average users’ capacity as a measurement of throughput, SINR maps and 10%-tile
SINR as a measurement of coverage, outage probability as a measurement of Quality of Ser-
vice (QoS) and finally Jain’s index and fairness ratio as a measurement of fairness. Fifthly,
simulation setup and simulation results are provided. A modified version of Vienna LTE
simulator is used as a powerful MATLAB simulation tool for simulating this thesis work.
Simulation results show that Reuse-1 scheme degrades most of network performance met-
rics other than throughput for dense deployment of femtocells like coverage, QoS and fair-
ness. Reuse-3 and PFR schemes provide good performance in terms of coverage, QoS, and
fairness but poor performance in terms of throughput due to poor spectral efficiency. SFFR
scheme provides higher throughput performance rather than Reuse-3 and PFR but lower in
iii

terms of coverage and fairness. SFR scheme provides the best throughput performance in
dense deployment of femtocells and good performance for coverage, QoS and fairness. So
SFR scheme can represent the best tradeoff among different proposed schemes via differ-
ent evaluation metrics. Two optimal SFR interior radii are then calculated for two different
femtocell deployment densities using exhaustive search. The two calculated radii are found
to optimize throughput performance without significant degradation in fairness performance
and provide good coverage performance.
iv

c
Copyright by Mahmoud Mohammed Selim, 2012.
All rights reserved.

ACKNOWLEDGEMENTS
Firstly, praise is to Allah, the Almighty, on whom ultimately we depend for sustenance and
guidance.
Secondly, I’d like to thank:
•Prof. Mohamed El-Sharkawy, my academic supervisor, for his leading role in con-
ducting research done by students of DSP and Communications Lab.
•Prof. Hossam Shalaby, my academic supervisor, for his leading role in ECE
department.
•Dr. Mostafa El-Khamy, my supervisor and research mentor, for his guidance and his
fruitful academic and research advices.
•Prof. H. Furukawa &Dr. O. Muta, my Japanese co-supervisors, for their hospitality
with me during my visit to their lab at Kyushu university. I really appreciate our video
conferences that provided me with valuable research ideas.
•My colleagues at E-JUST for lovely and cooperative environment we had together.
•The Missions Department of Egyptian Ministry of Higher Education for their financial
support to complete my degree.
vi

To my beloved parents ..
and my dear sister ..
vii

TABLE OF CONTENTS
Summary ...................................... iii
Acknowledgements................................. vi
ListofTables .................................... xi
ListofFigures.................................... xii
ListofAbbreviations ................................ xiv
1 INTRODUCTION 1
1.1 3GPPRelease8(LTE) ............................ 1
1.2 Indoor Coverage Challenge . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 ProblemStatement .............................. 3
1.4 Research Motivations and Opportunities . . . . . . . . . . . . . . . . . . . 5
1.5 Outline of Master Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 FEMTOCELL TECHNOLOGY: AN OVERVIEW 7
2.1 FemtocellDefinition ............................. 7
2.2 FemtocellDeployment ............................ 8
2.3 FemtocellAdvantages............................. 8
2.3.1 Operator’s advantages . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3.2 Subscriber’s advantages . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Femtocell Access Control Strategies . . . . . . . . . . . . . . . . . . . . . 9
2.4.1 ClosedAccessMode......................... 9
2.4.2 OpenAccessMode.......................... 9
2.4.3 HybridAccessMode......................... 9
viii

2.5 Femtocell Network Architecture . . . . . . . . . . . . . . . . . . . . . . . 10
3 INTERFERENCE MANAGEMENT IN OFDMA FEMTOCELL NETWORKS:
A SURVEY 11
3.1 SpectrumSplitting .............................. 12
3.2 Femto-aware spectrum arrangement scheme . . . . . . . . . . . . . . . . . 12
3.3 Clustering of femtocells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.4 PowerControl................................. 13
3.5 CognitiveRadio................................ 13
3.6 Fractional Frequency Reuse (FFR) . . . . . . . . . . . . . . . . . . . . . . 14
3.7 Comparison Evaluation of Surveyed Schemes . . . . . . . . . . . . . . . . 14
4 SYSTEM ANALYSIS 16
4.1 ViennaLTESimulator ............................ 16
4.2 SystemModel................................. 17
4.2.1 CellularLayout............................ 17
4.2.2 Modeling Indoor Area . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.3 Propagation Pathloss (PL) Models . . . . . . . . . . . . . . . . . . 20
4.2.4 AntennaPatterns ........................... 22
4.2.5 ShadowingModels.......................... 23
4.3 SINR and Capacity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.3.1 SINRAnalysis ............................ 23
4.3.2 CapacityAnalysis .......................... 25
5 PROPOSED FREQUENCY ALLOCATION SCHEMES 26
5.1 Reuse-1Scheme................................ 27
5.2 Reuse-3Scheme................................ 28
5.3 Soft Frequency Reuse (SFR) Scheme . . . . . . . . . . . . . . . . . . . . . 29
5.4 Partial Frequency Reuse (PFR) Scheme . . . . . . . . . . . . . . . . . . . 31
5.5 Soft Fractional Frequency Reuse (SFFR) Scheme . . . . . . . . . . . . . . 32
ix

6 EVALUATION METHODOLOGY & SIMULATION RESULTS 34
6.1 EvaluationMetrics .............................. 34
6.1.1 Coverage Performance . . . . . . . . . . . . . . . . . . . . . . . . 34
6.1.2 Throughput Performance . . . . . . . . . . . . . . . . . . . . . . . 35
6.1.3 Quality of Service (QoS) Performance . . . . . . . . . . . . . . . . 36
6.1.4 Fairness Performance . . . . . . . . . . . . . . . . . . . . . . . . . 37
6.2 SimulationSetup ............................... 38
6.3 SimulationResults .............................. 38
6.4 Optimizing Interior Radius of Soft Frequency Resue (SFR) Scheme . . . . 50
7 CONCLUSION & FUTURE WORK 55
7.1 Conclusion .................................. 55
7.2 FutureWork.................................. 57
References 58
x

LIST OF TABLES
3.1 Comparison of Different Interference Management Schemes . . . . . . . . 15
6.1 SimulationParameters ............................ 39
xi

LIST OF FIGURES
1.1 Interference scenarios for both UL and DL in LTE femtocell network . . . . 4
2.1 An Example of a Femtocell Network . . . . . . . . . . . . . . . . . . . . . 7
2.2 Simplified Femtocell Network Architecture . . . . . . . . . . . . . . . . . 10
4.1 CellularLayout ................................ 18
4.2 Example of a house with deployed femto base station . . . . . . . . . . . . 19
4.3 Hybrid Macro-Femto Cellular Network . . . . . . . . . . . . . . . . . . . 19
4.4 Different Connection Links of Hybrid Macro-Femto Network . . . . . . . . 20
5.1 Reuse-1Scheme................................ 27
5.2 Reuse-3Scheme................................ 28
5.3 SFRScheme.................................. 29
5.4 PFRScheme.................................. 31
5.5 SFFRScheme................................. 32
6.1 Macro-Femto SINR coverage map for different allocation schemes (60 de-
ployedfemtocells) .............................. 42
6.2 Macro SINR coverage map for different allocation schemes (60 deployed
femtocells) .................................. 43
6.3 Macro-Femto SINR coverage map for different allocation schemes (150 de-
ployedfemtocells) .............................. 44
6.4 Macro SINR coverage map for different allocation schemes (150 deployed
femtocells) .................................. 45
xii

6.5 10%-tile SINR vs. number of femtocells . . . . . . . . . . . . . . . . . . . 46
6.6 Average users’ capacity vs. number of femtocells . . . . . . . . . . . . . . 47
6.7 Outage Probability vs. SINR threshold . . . . . . . . . . . . . . . . . . . . 48
6.8 Jain’s index vs. number of femtocells . . . . . . . . . . . . . . . . . . . . 49
6.9 Fairness Ratio vs. number of femtocells . . . . . . . . . . . . . . . . . . . 50
6.10 Fairness/Throughput Tradeoff (90 active femtocells) . . . . . . . . . . . . . 50
6.11 Fairness/Throughput Tradeoff (150 active femtocells) . . . . . . . . . . . . 51
6.12 Average users’ capacity vs. SFR interior radius . . . . . . . . . . . . . . . 52
6.13 Fairness Ratio vs. SFR interior radius . . . . . . . . . . . . . . . . . . . . 53
6.14 Macro-Femto SINR coverage map for Reuse-1 and SFR scheme with opti-
mized interior radius (60 deployed femtocells) . . . . . . . . . . . . . . . . 54
xiii

LIST OF ABBREVIATIONS
2G 2nd Generation
3GPP 3rd Generation Partnership Project
AP Access Point
AWGN Additive White Gaussian Noise
BS Base Station
CP Cyclic Prefix
CQI Channel Quality Indicator
CSG Closed Subscriber Group
DAS Distributed Antenna Systems
DL Downlink
DSL Digital subscriber line
EDGE Enhanced Data Rates for GSM Evolution
eNB Evolved NodeB
FAP Femto Access Point
FFR Fractional Frequency Reuse
FFT Fast Fourier Transform
xiv

FSC Femtocell System Controller
FUE Femto User Equipment
GPRS General Packet Radio Service
GSM Global System for Mobile Communications
HeNB Home Evolved NodeB
HSPA High Speed Packet Access
ICI Inter-Cell Interference
IFI Inter-Femto Interference
IP Internet Protocol
ISP Internet Service Provider
LLS Link Level Simulator
LTE Long Term Evolution
MATLAB Matrix Laboratory
MCS Modulation and Coding Schemes
MIMO Multiple-Input Multiple-Output
MUE Macro User Equipment
OFDM Orthogonal Frequency Division Multiplexing
OFDMA Orthogonal Frequency Division Multiple Access
PFR Partial Frequency Reuse
PL Path Loss
xv

PSTN Public Service Telephone Network
QoS Quality of Service
RF Radio Frequency
ROI Region of Interest
Rx Receive
SC-FDMA Single-Carrier Frequency Division Multiple Access
SD Standard Deviation
SFFR Soft Fractional Frequency Reuse
SFR Soft Frequency Reuse
SINR Signal to Interference plus Noise Ratio
SISO Single-Input Single-Output
SLS System Level Simulator
Tx Transmit
UE User Equipment
UL Uplink
UMTS Universal Mobile Telecommunications System
VOIP Voice over Internet Protocol
WiMAX Worldwide Interoperability for Microwave Access
xvi

CHAPTER 1
INTRODUCTION
“The next significant performance leap will come from small cells such as fem-
tocells and picocells, which bring the network closer to users. Femtos cost-
effectively improve coverage and capacity and significantly enhance data rates
for users within the femto coverage area. As femtos offload traffic from the macro
network, users outside femto coverage also benefit [1]”
1.1 3GPP Release 8 (LTE)
3GPP Release 8 known as Long Term Evolution (LTE) is the evolution of the third generation
mobile communications standard, UMTS to the fourth generation technology with increased
capabilities of providing voice and other broadband data services. The requirements and
main targets of LTE technology are presented in [2] and [3]. LTE standard aims to increase
system data rates especially for cell edge zones and improve spectrum efficiency to avoid the
problem of scarcity of resources. LTE also allows spectrum flexibility (1.25, 2.5, 5, 10, 15
and 20 MHz) for flexible radio planning. The reduction of packet latency is a crucial target of
LTE to allow real-time services like VOIP and video calls. LTE has to reduce the radio access
network cost and provides cost-effective migration process from earlier 3GPP releases. LTE
network architecture is assumed to be a flat all-IP packet based network architecture which
highly simplifies the functionalities done by the system.
The target instantaneous downlink (DL) peak data rate of LTE is assumed to be 100
Mbps using 20 MHz spectrum allocation (5 bps/Hz). While the target instantaneous up-
1

link (UL) peak data rate is assumed to be 50 Mbps using 20 MHz spectrum allocation (2.5
bps/Hz). LTE exploits Orthogonal Frequency Division Multiple Access (OFDMA) as an ac-
cess scheme for the downlink, while it exploits Single-Carrier Frequency Division Multiple
Access (SC-FDMA) as an access scheme for the uplink. These target data rates are assumed
to be achieved using a flexible radio interface based on OFDM technology and MIMO an-
tenna processing [4][5].
1.2 Indoor Coverage Challenge
It has been noticed recently that most of cellular traffic occurs indoor. This traffic is not
limited to voice calls but extended to video and high speed data services. Some surveys show
that 45% of households and 30% of businesses experience poor indoor coverage problem [6].
A typical usage of outdoor macrocells for serving indoor users will results in some drawbacks
such as:
•Indoor users will require higher power from Base Stations (BS) to alleviate high pen-
etration loss which usually comes at the expense of power used for other users which
leads to reduced cell throughput.
•The need for large number of outdoor BS sites which represents additive cost for op-
erators especially for densely populated areas.
•The building penetration is a big challenge for networks operating at 2 GHz and above.
•Lower channel conditions will results in lower modulation and coding schemes (MCS)
and hence lower data rates for indoor users especially those at cell edges.
Some indoor solutions such as Distributed Antenna Systems (DAS) [7] and Picocells [8]
have been proposed to solve the problem of poor indoor coverage in environments like office
buildings and shopping centers. These solution are deployed by operators in order to enhance
coverage, offload traffic from ordinary macrocells, increase data rates and provide high qual-
ity services. These solutions are not cost effective for small environments such as home users
2

and home offices. A recent low cost indoor solution which is femtocell technology has been
proposed and can be easily deployed by users without operator intervention.
Femtocells, also known as home base stations, are cellular network access points (AP)
that connect standard mobile devices to a mobile operator’s network using residential DSL,
cable broadband connections, optical fibres or wireless last-mile technologies [9]. A more
details a bout femtocell technology will provided in Chapter 2.
1.3 Problem Statement
The deployment of femtocells results in a change in the topology of the ordinary macro-
cellular network. The new network architecture is composed of two different layers; the
macrocell layer and the femtocell layer. This new network architecture is usually called
the two-tier network. The presence of femtocells within the same geographic location of
macrocells and the possible transmission on the same frequency bands will result in the most
severe problem in any telecommunication system which is interference. Dead zones within
the macrocell may appear if no interference avoidance techniques are exploited in the two-
tier network.
The interference in the two-tier network can be classified into co-layer interference and
cross-layer interference. Co-layer interference occurs when both the aggressor and the vic-
tim belong to the same layer (femtocell layer). An example of this type of interference is a
group of femtocells close to each other such that the transmission of one femto user equip-
ment (FUE) is interfering the reception of a neighbor femtocell (Uplink interference) or the
transmission of a femtocell is interfering the reception of a FUE that belongs to different
femtocell (downlink interference). This type of interference results in inter-carrier interfer-
ence (ICI) at the uplink and femtocell coverage holes at the downlink [10].
The other type is the cross-layer interference. It occurs when the aggressor and the
victim belong to different network layers. An example of this type of interference at the
uplink occurs when a macro user equipment (MUE) is interfering the reception of a neighbor
3

Figure 1.1: Interference scenarios for both UL and DL in LTE femtocell network
femtocell or a FUE is interfering the reception of a neighbor macrocell. It also occurs at the
downlink when a femtocell is interfering the reception of a neighbor MUE. This type of
interference results in inter-carrier interference at the femtocell besides risk of invalidation
of macrocell sub-channels at the uplink. It also results in femtocell coverage holes and inter-
carrier interference at the macrocell users at the downlink [10]. The different interference
scenarios for both Uplink and Downlink are summarized in figure 1.1.
Many interference management schemes have been proposed in the literature which will
be discussed in details in Chapter 3. One famous mitigation method is based on the effi-
cient frequency allocation process between the two network layers. While the co-channel
allocation (sharing the entire system bandwidth between the two network layers) results in
a high spectral efficiency and high total system throughput, the high amount of interference
generated in the network results in many coverage holes and low Quality of Service (QoS).
The orthogonal channel allocation (dividing the system bandwidth orthogonally between the
two network layers) results in limited amount of interference and consequently better cov-
4

erage and QoS, the system still suffers from poor spectral efficiency and waste of valuable
frequency resources. Thus efficient frequency allocation schemes that gather the best of the
two allocation methods are needed for efficient interference mitigation. This will be the core
of this research work and our proposed frequency allocation schemes are described in details
in Chapter 5.
1.4 Research Motivations and Opportunities
Recent research on the topic of femtocells and especially the interference mitigation schemes
in the two-tier networks is assumed to be crucial for the following reasons:
•2/3 of calls and about 90% of data services occur indoor.
•A large deployment of femtocells is expected in the near future [11].
•The capability of providing data services for indoor users using femtocells and conse-
quently more revenues for operators.
•Offloading more traffic from macrocells and consequently less number of macrocell
sites is needed which means lower cost for operators.
•Wide deployment of femtocells should be preceded by efficient resource management
process which is a very hot topic in most recent research work.
1.5 Outline of Master Thesis
This master thesis is divided into 7 chapters:
•Firstly, Chapter 1 provides a brief introduction about 3GPP LTE Technology, small-
cell technologies, problem statement,and research motivations and opportunities.
•Chapter 2 provides an overview about the technology of femtocells and a brief descrip-
tion of the LTE femtocell network.
5

•Chapter 3 provides a literature survey on the management schemes used for interfer-
ence avoidance in the two-tier networks.
•Chapter 4 provides a detailed description of system model assumed in this thesis work.
•Chapter 5 provides a detailed description of proposed frequency allocation schemes
used in the process of frequency resources allocation between macrocells and femto-
cells.
•Chapter 6 provides a detailed description for the evaluation methodology used for eval-
uating proposed frequency allocation schemes, the simulation setup done, and finally
the simulation results and comments on these results.
•Chapter 7 summarizes the concluding remarks of this master thesis work and proposes
the future work to be done in order to continue the investigation of this research topic.
6

CHAPTER 2
FEMTOCELL TECHNOLOGY: AN
OVERVIEW
This chapter gives a brief overview on the femtocell technology in terms of definition, de-
ployment techniques, importance, access strategies and the network architecture.
2.1 Femtocell Definition
Figure 2.1: An Example of a Femtocell Network
As mentioned in the previous chapter, femtocells are cellular network wireless access
points that connect standard mobile devices to the network of a mobile operator via residen-
tial DSL, cable broadband connections or optical fibres as shown in figure 2.1 [9]. The first
product of home base station was firstly announced by Motorola at 2002. The Small Cell
Forum [11] is responsible for the standardization and deployment of femtocells worldwide.
Femtocell Networks are expected to play an important role in the network architectures of
7

upcoming 3GPP releases. Femto access points (FAP) can be found for different cellular
technologies such as 2G technologies (e.g. GSM/GPRS/EDGE), 3GPP technologies (e.g.
UMTS/HSPA/LTE) and non-3GPP technologies such as mobile WiMAX.
2.2 Femtocell Deployment
Femtocells are assumed to be self-deployed by users rather than operators. The femtocells
should have the capability of automatic configuration as they are regarded as consumer elec-
tronics. The automatic configuration is achieved via two processes; the sensing process in
which the femtocells sense the surrounding environment for assessment and the auto-tuning
process in which the femtocell adjusts its configuration parameters like downlink Tx power
and sub-channel allocation. The femtocell deployment can also be classified based on the
capacity into; home femtocells which are capable of supporting 3-5 simultaneous users and
enterprise femtocells which are capable of supporting 8-16 users [12].
2.3 Femtocell Advantages
Femtocells can provide a lot of advantages for both operators and subscribers as follows.
2.3.1 Operator’s advantages
•Offloading traffic from macrocells and hence less macrocell sites are needed.
•Simplifying Radio Frequency (RF) planning process.
•Improving service quality and wideband data services and hence more revenues.
•providing coverage for places where macrocell implementation is very difficult.
8

2.3.2 Subscriber’s advantages
•Gathering all voice, video and high speed data services in one consumer electronics
device.
•Making use of bundled services which are very cost-effective.
•Saving user equipment (UE) power.
2.4 Femtocell Access Control Strategies
There are three access control modes in which a femtocell could be operated:open,closed,
and hybrid [13]. The description of these different access modes is shown below.
2.4.1 Closed Access Mode
It is also called Closed Subscriber Group (CSG) in 3GPP terminology. In this access mode,
only a set of UEs are allowed to use the services of the femtocell based on a list of subscribed
UEs to this femtocell. No other UEs are allowed to use the femtocell except for emergency
calls when no other cell is available.
2.4.2 Open Access Mode
In this access mode, any UE in the coverage area of the femtocell can camp on it without any
preferential access.
2.4.3 Hybrid Access Mode
This access mode is similar to closed access mode except that UEs that are not part of the list
of the subscribed UEs may camp and acquire some level of service but the subscribed UEs
still have preferential access.
9

2.5 Femtocell Network Architecture
Figure 2.2: Simplified Femtocell Network Architecture
Figure 2.2 shows a sample network architecture of the femtocell deployment. The homes
are expected to have a broadband modem connection (i.e. cable, or fiber) through their
internet service provider (ISP) to the internet. The cellular specific data will be transferred
to the femtocell gateway for access back into the cellular network. For the 3GPP network,
the gateway would interface to the core network; this interface is called Iu-h. We have also
shown the cellular core network accessing the public service telephone network (PSTN) for
voice services and a broadband interconnect for data services. As the core network evolves
into a packet network, all traffic will be IP based, thus allowing for a convergence of services
[12].
10

CHAPTER 3
INTERFERENCE MANAGEMENT IN OFDMA
FEMTOCELL NETWORKS: A SURVEY
The interference problem has been discussed briefly in Section 1.3. It has been stated that
the interference in two-tier networks has two types; co-tier interference and cross-tier inter-
ference. Both types of interference may occur either in the uplink or downlink transmission.
Due to the unplanned and random deployment of femtocells, interference can be a severe
problem in two-tier networks if it is not well managed.
Interference is managed via two different approaches which are interference cancelation
or interference avoidance [14]. Interference cancelation aims to reduce interference at the
receiver end by means of signal processing using some prior knowledge about interfering
signal. It has been found that interference cancelation is not preferred practically in two-tier
networks as it mostly requires antenna array systems and signalling overhead which may not
be suitable especially for downlink. Interference avoidance aims to allow intelligent trans-
mission to avoid or reduce interference as much as possible. In this chapter we provide a brief
survey on some of the approaches and schemes used for providing interference avoidance in
two-tier OFDMA femtocell networks.
Some of the approaches proposed for interference avoidance in two-tier OFDMA femto-
cell networks are mentioned below.
11

3.1 Spectrum Splitting
Spectrum splitting has been proposed in [15] to solve the problem of Co-tier interference.
The spectrum is divided into two portions; one of them is assigned to the operation of macro-
cell users while the other one is assigned for the operation of femtocell users. While cross-tier
interference is substantially eliminated, co-tier interference still exists. Spectrum splitting
usually tends to reduced spectral efficiency but it may be needed in some scenarios such
as dense deployment of femtocells with dominant cross-tier interference that is difficult to
manage in any other way.
3.2 Femto-aware spectrum arrangement scheme
This approach has been proposed in [16] to solve the cross-tier interference problem in up-
link. The spectrum is split into two portions; one of them is dedicated for macrocell operation
and the other is shared between macrocells and femtocells such that this allocation process is
configured by the network operator and known for macrocells. The macrocell prepares a list
of MUEs that represent a threat for nearby femtocell so that they will be assigned resources
from the portion dedicated for macrocells to avoid interference.
3.3 Clustering of femtocells
This scheme has been proposed in [17] to avoid both co-tier and cross-tier interference in
downlink transmission. The scheme depends on assigning a central node called Femtocell
System Controller (FSC) which is responsible for gathering all system configuration done by
femtocells and performing computations. Similarly like the previous approach, spectrum is
divided into two parts; one is dedicated for macrocells while the other is shared. A clustering
algorithm is used to allocate femtocells into different frequency reuse clusters and FUEs
of different femtocells in the same cluster use the same sub-channels allocated from the
shared frequency band. Based on the geographical locations of the femtocells, the threshold
12

distance for clustering interference is calculated. If the Euclidean distance between any two
femtocells is less than the threshold distance, then they are assigned to different clusters to
avoid co-tier and cross-tier interferences.
3.4 Power Control
Power control schemes are used to mitigate cross-tier interference by reducing transmission
power of femtocells. The main advantage of power control schemes is the possibility of
using the entire bandwidth by a method of interference coordination. Adjusting femtocell
power can be done either by measurement results of a femtocell a lone or based on coordina-
tion with macrocells. A hybrid method is proposed in [18] where a femtocell switches power
adjustment technique between either a lone or based on coordination depending on the de-
ployment scenario. For a cluster basis scenario, the sensing process required in adjustment
of power control is done either by centralized method by the help of macrocell or distributed
method where each femtocell works in an individual manner. Game theory models can be
also used in power control interference mitigation schemes [19].
3.5 Cognitive Radio
The concept of cognitive radio is used in the process of distributed spectrum sensing in a
downlink co-tier interference management scheme [20]. The co-tier interference is estimated
by each femtocell during the sensing phase using path-loss information. Using the femtocell
gateway, the information obtained by each femtocell about carriers used by neighboring
femtocells are shared among femtocells. Femtocells are then capable of accessing carriers
that are not used by neighbor femtocells to minimize interference as much as possible. If all
carriers are used by neighbor femtocells, then it use the carriers used by the furthest femtocell
to minimize interference as much as possible.
13

3.6 Fractional Frequency Reuse (FFR)
The concept of Fractional Frequency Reuse (FFR) depends on dividing the macrocell into
center region and edge region such that different reuse factors are used at each region mainly
to enhance cell edge performance. FFR has two types; static and dynamic. A static FFR
scheme is used in [21] to mitigate cross-tier interference in co-channel allocation scenario.
Femtocells sense the spectrum during turn on to discard operation on sub-bands with largest
received signal strength to enhance signal-to-interference and noise ratio (SINR) of MUEs
and overall network throughput. Another static FFR scheme is used in [22] to mitigate
cross-tier interference in downlink. This scheme divides macrocell into center zone with
reuse factor of 1 and edge zone with reuse factor of 3. The spectrum is divided into two
parts; one of them is assigned for center region while the other is divided equally between
the three sectors of the macrocell. The femtocells at each sub-area of the macrocell use the
sub-bands that are not used by macrocell operation at this sub-area.
The other type is the dynamic FFR. An adaptive FFR scheme is used in [23] to mitigate
cross-tier interference in downlink. Resource partitioning process is varied in both time and
frequency (dynamic) and the allocation process depends on the density of femtocells and
the location of each one (center or edge). If the femtocells are in a dense scenario in center
region, femtocells use orthogonal sub-channels to minimize interference otherwise they use
arbitrarily sub-channel.
3.7 Comparison Evaluation of Surveyed Schemes
The interference management schemes mentioned above are compared and evaluated via dif-
ferent approaches. These approaches include type of interference considered, transmission
mode , complexity, efficiency, access strategy, coordination between macrocell and femto-
cell(s). This comparison is summarized in Table 3.1.
14

Transmission Access Type
Scheme Mode Cooperation Mode Complexity Efficiency of
Interference
Femto
aware Uplink Required Closed Moderate Low Cross-
spectrum tier
management
Co-tier
Clustering Downlink Required Closed Moderate Moderate and
of Cross-
femtocells tier
Not Closed Cross-
Power Downlink Required and Moderate High tier
control open
Closed Cross-
Cognitive Downlink Required and Moderate Moderate tier
radio open
Fractional Closed, Co-tier
Frequency Downlink Not open Low High and
Reuse Required and Cross-
(FFR) Hybrid tier
Table 3.1: Comparison of Different Interference Management Schemes
In our research work we rely and focus on interference management schemes that de-
pend on frequency allocation and resource partitioning with special focus on those schemes
relying on FFR as will be described later in details in Chapter 5.
15

CHAPTER 4
SYSTEM ANALYSIS
In this chapter we provide a full description and analysis of LTE femtocell system applied
in our research work. We start firstly by giving an overview of the Vienna LTE Simulator
[24] used as a powerful LTE System Level Simulator (SLS). We then describe the LTE
cellular system Layout and how we can model indoor area. We also describe the propagation
pathloss (PL) models used between BSs and UEs in addition to different types of antenna
patterns used. We provide an SINR and Capacity analysis for Downlink (DL) between BSs
and UEs.
4.1 Vienna LTE Simulator
System Level Simulators are very crucial tools to evaluate the performance of new mobile
technologies like LTE. Vienna LTE Simulator is a MATLAB computationally efficient LTE
System Level Simulator (SLS). The simulator can be used to evaluate the performance of
downlink (DL) shared channel for different transmission modes (e.g. SISO, MIMO, and
Transmit diversity with different antenna configurations) [24].
The Vienna LTE Simulator package has two types of simulators: LTE Link Level Simula-
tor (LLS) and LTE System Level Simulator (SLS). The LTE Link Level Simulator is used to
simulate the Physical Layer procedures and analyze the link level related issues like receiver
structures and coding schemes while the LTE System Level Simulator is used to measure the
whole network performance after abstracting the Physical Layer from Link Level results.
16

The core part of the System Level Simulator is divided into link measurement model
and link performance model. Link measurement model is used to abstract the measured
link quality used for link adaptation and resource allocation based on signal to interference
and noise ratio (SINR) as a metric. The aim of this step is to reduce run-time computational
complexity by pregenerating and storing the results of these heavy computations to be reused
during simulation time. While the link performance model is used to determine throughput
and error rates at a reduced complexity. One of the most important purposes of such simu-
lation tool is measuring the quality and performance of new scheduling algorithms in LTE
Systems.
4.2 System Model
4.2.1 Cellular Layout
Our wrap-around LTE cellular layout is composed of 7 macrocells. A base station (BS)
located at the center of each macrocell is used to provide service for macro user equipments
(MUEs) attached to this BS. Base Stations for LTE cellular systems are given the name
evolved NodeB (eNBs) in 3GPP standard [25]. Each macrocell consists of 3 hexagonal-
sector sites such that each hexagonal sector is served by a different directional antenna. The
beam directions of different sector antennas are separated by 120◦from each other.
MUEs are randomly dropped in the macrocell and can be classified into outdoor MUEs
located in free space area and indoor MUEs located inside offices, enterprises, houses ... etc.
Indoor MUEs are served by ordinary eNBs unless they are authorized to access femto BSs
located in their coverage area. The cellular layout of ordinary LTE cellular system is shown
in figure 4.1.
17

Figure 4.1: Cellular Layout
4.2.2 Modeling Indoor Area
Macrocell Coverage area in our system model is classified into outdoor coverage area (e.g.
Free Space) and indoor coverage area. Indoor coverage area is represented by uniformly
distributed randomly dropped square houses of size 15m×15m. One femto BS is randomly
dropped inside each house. Figure 4.2 shows one example of randomly dropped houses
where femto BS could be located in one of the 9 possible locations inside each house.
The femtocell coverage area is defined by a specific radius centered by the location of
the femto BS inside each house. The number of active femto BSs at a time is not set to
a fixed number but set as a variable parameter to study performance in different femtocell
deployment densities. unlike MUEs that can be located anywhere with macrocell coverage
18

Figure 4.2: Example of a house with deployed femto base station
Figure 4.3: Hybrid Macro-Femto Cellular Network
area, FUEs existence is only limited to the predefined femtocell coverage area otherwise
it will handoff to macrocell and be a MUE. Figure 4.3 shows an example of a macrocell
providing access for ordinary MUEs (outdoor and indoor) overlaid by femtocell network
providing access for only limited number of subscribed UEs (FUEs).
19

4.2.3 Propagation Pathloss (PL) Models
Figure 4.4: Different Connection Links of Hybrid Macro-Femto Network
Suburban deployment scenario is assumed in our research work. Propagation pathloss
(PL) models are the reference formulas used to describe the propagation loss encountered
in the downlink between Tx (macro BSs or femto BSs) and Rx (MUEs or FUEs). Pathloss
formulas are valid for Tx-Rx separation larger than 1 m. Pathloss (PL) model formulas can
be summarized as follows [26]:
UE to Macro BS
1. UE is outside:
The following PL model expressed in equation 4.1 is used when the transmitter is a
macro BS and the receiver is either outdoor MUE or a FUE located outside the house
20

such as link 1 and 2 respectively in figure 4.4.
PL(dB) = 15.3+37.6 logR(4.1)
2. UE is inside a house:
The following PL model expressed in equation 4.2 is used when the transmitter is a
macro BS and the receiver is either indoor MUE or a FUE located inside such as links
3 and 4 respectively in figure 4.4.
PL(dB) = 15.3+37.6 logR+Low (4.2)
UE to Femto BS
1. UE is inside the same house as femto BS:
The following PL model expressed in equation 4.3 is used when the transmitter is a
femto BS and the receiver is either indoor MUE or a FUE located inside the house
such as link 5 and 6 respectively in figure 4.4.
PL(dB) = 38.46 +20 logR+0.7d2D,indoor (4.3)
2. UE is outside:
The following PL model expressed in equation 4.4 is used when the transmitter is a
femto BS and the receiver is either outdoor MUE or a FUE located outside the house
such as link 7 and 8 respectively in figure 4.4.
PL(dB) = max(15.3+37.6 logR,38.46 +20 log R) + 0.7d2D,indoor +Low (4.4)
3. UE is inside a different house:
21

The following PL model expressed in equation 4.5 is used when the transmitter is a
femto BS and the receiver is either MUE or FUE located inside a different house other
than that of femto BS such as link 9 in figure 4.4.
PL(dB) = max(15.3+37.6 logR,38.46 +20 log R) + 0.7d2D,indoor +Low,1+Low,2
(4.5)
where
•R is the distance between Transmitter and Receiver.
•Low,Low,1and Low,2are the penetration losses of outdoor walls, which are 10dB.
•d2D,indoor is the distance inside each house in the case of UE located inside the same
house or the total distance inside two houses if it is located in a different house.
4.2.4 Antenna Patterns
As described in subsection 4.2.1 each hexagonal sector is served by a directional antenna
such that the main beams of the three antennas used in each macrocell are separated by 120◦
from each other. Based on [27] the antenna gain encountered at any point in the macrocell
as a function of the angle from the reference main beam direction of this antenna θcan be
formulated in equation 4.6 below.
A(θ).
=−min[12(θ
θ3dB
)2,Am](4.6)
where θ3dB = 65 ◦and Am= 20 dB.
The antenna patterns of both types of UEs and femto BSs are assumed to be omnidirec-
tional.
22

4.2.5 Shadowing Models
Log-normal shadowing is assumed in our research work. The standard deviation for log-
normal shadowing is assumed to be 4 dB for links between femto BSs and its associated
FUEs. Standard deviation of 8 dB is used for all other links including interference links.
4.3 SINR and Capacity Analysis
4.3.1 SINR Analysis
In this part of the chapter we analyze signal to interference and noise ration (SINR) encoun-
tered by different types of UEs (e.g. MUEs and FUEs) in LTE cellular system for any of the
different frequency allocation schemes described in details in Chapter 5.
As described in section 4.2 MUEs are categorized into outdoor and indoor UEs and can
be located anywhere within the coverage area of each macrocell while FUEs can only be
located within coverage area of each femtocell. The analysis used for calculating SINR
encountered by any MUE in the macrocell can be found below.
Each MUE is assigned to a macrocell where its BS is responsible for providing service for
this UE. This UE is interfered by both neighboring macrocells and femtocells operating on
the same frequency sub-bands assigned to its serving BS. While the intracell interference is
almost eliminated due to the characteristics of OFDMA provided by LTE system [28], inter-
cell interference (ICI) still exists and represents an important issue especially for Reuse-1
operation and small-sized macrocells. The downlink signal to interference and noise ratio
(SINR) for any MUE can be formulated as shown below in equation 4.7.
SIN RMU E =PS,B
R
σ2+
7
∑
b=1
3
∑
s=1
s6=S
Ps,b
R+
Nf
∑
f=1
Pf
R
(4.7)
23

where PS,B
Ris the received power from Sector Sof serving macrocell B,Ps,b
Ris the received
power from interfering Sector sassociated with macrocell b,Pf
Ris the received power from
interfering femtocell f using the same sub-bands, Nfis number of femtocells, and σ2is
the thermal noise power in watts. The power received PRcan be directly calculated using
equation 4.8 shown below.
PR(dBm) = PT(dBm)−PT L (dB)(4.8)
where PTis the macrocell transmission power in dBm and PT L is the total loss encountered
by the signal in dB such that
PT L(dB) = PL(dB)−GT(dB)(4.9)
where PLis the macroscopic pathloss encountered by the signal in dB and GTis the transmit-
ting antenna gain in dB. The receiver antenna gain GRis set to 0 dB since the UEs antenna
pattern is set to be omnidirectional.
Each FUE is attached to its associated femtocell which is responsible for providing ser-
vice for this UE. If FUE becomes unauthorized for femto access, it will directly handoff to
nearest macrocell. Similarly like MUE, each FUE is interfered by all neighboring macro-
cells and femtocells operating on the same frequency sub-bands like its serving femto BS.
The downlink SINR for any FUE can be calculated using equation 4.10 below.
SIN RFUE =PF
R
σ2+
7
∑
b=1
3
∑
s=1
Ps,b
R+
Nf
∑
f=1
f6=F
Pf
R
(4.10)
where PF
Ris the received power from serving femtocell F, and Ps,b
Rand Pf
Rare zero if the
corresponding cell is utilizing another sub-band.
24

4.3.2 Capacity Analysis
We assume in our work that one MUE is connected to each macro BS at a time such that
this MUE is capable of using all available frequency resources assigned to this macro BS.
The same is assumed for femtocells, one FUE is connected to each femtocell at a time such
that this FUE is capable of using all available frequency resources assigned to its associated
femto BS. Our methodology depends on changing the location of both MUE and FUE con-
tinuously in the x-y domain of our assumed cellular network in order to cover all possible
locations where MUEs and FUEs can be located within coverage area of macrocells and
femtocells respectively. One other important assumption in our work that all neighboring
BSs are assumed to be always transmitting with full power over all their available frequency
sub-bands so worst-case interference scenario can be simulated and studied.
The theoretical user capacity (bps) can be calculated for any UE (macro or femto) using
equation 4.11 below (assuming a static additive white gaussian noise AWGN scenario).
CMUE /FU E =Wlog2(1+SINRMUE /FU E )(4.11)
where Wis the total bandwidth of the sub-bands available for this UE in Hz. This procedure
is used to evaluate SINR and theoretical user capacity for both MUEs and FUEs at all pos-
sible locations in our cellular network which will be used later in Chapter 6 for calculating
our evaluation metrics for measuring system performance.
25

CHAPTER 5
PROPOSED FREQUENCY ALLOCATION
SCHEMES
In this chapter we describe in details our proposed frequency allocation schemes in LTE
femtocell system. Frequency allocation schemes mean the procedures and algorithms used
to allocate the limited frequency resources for both macrocells and femtocells in LTE fem-
tocell system. The aim of any frequency allocation scheme is to enhance the overall system
performance and increase the spectral efficiency of the limited frequency bandwidth.
In each scheme we exploit a well-known frequency allocation method for macrocell op-
eration with focus on those depending on the concept of Fractional Frequency Reuse (FFR)
that was firstly proposed in [29] for GSM networks as it proves to solve the problem of poor
coverage at edge zone that represents a big issue in any mobile cellular system [30]. We then
propose techniques for allocating frequency resources for femtocells such that these tech-
niques cope with those of macrocell operation to enhance the overall system performance.
We ensure that all of our proposed schemes are static and don’t require any signalling be-
tween macrocells and femtocells. The well-known frequency allocation schemes exploited
in our research work are: Reuse-1, Reuse-3, Soft Frequency Reuse (SFR), Partial Frequency
Reuse (PFR), and Soft Fractional Frequency Reuse (SFFR). All of these schemes are de-
scribed in details in the remaining part of this chapter.
26

Figure 5.1: Reuse-1 Scheme
5.1 Reuse-1 Scheme
We study the Reuse-1 scheme in our research work for comparison purpose with other
schemes. The universal reuse scheme (or Reuse-1) assigns the entire frequency resources
to be reused by all macrocells and femtocells existing in the system. The main advantage
of Reuse-1 scheme is the possibility of using all available frequency resources and hence
increasing the spectral efficiency of scarce bandwidth. This usually comes at the expense of
the amount of interference generated in the system. Since all macrocells and femtocells share
the same bandwidth at the same time to provide service for their attached UEs, the amount
of inter-cell interference ICI becomes very high especially for small-sized macrocells. The
Reuse-1 scheme also results in a coverage problem due to poor SINR for those MUEs far
from their serving BSs at the edge of macrocell due to the interfering transmission of nearby
macrocells [31]. Reuse-1 scheme also results in a severe problem for indoor MUEs that are
very near to active transmitting femto BSs.
In figure 5.1 we describe the operation of Reuse-1 scheme where all macrocells use the
entire frequency bandwidth at the same time slots with reference transmission power PM. The
femtocells also applies the concept of Reuse-1 such that they use the same entire frequency
bandwidth simultaneously with macrocells but with limited transmission power PF.
27

Figure 5.2: Reuse-3 Scheme
5.2 Reuse-3 Scheme
Higher reuse factors than Reuse-1 have been proposed in order to solve the problem of poor
coverage at edge zone [32]. Reuse-3 has proved that it can provide the best performance
among different reuse factors as shown in [33]. In Reuse-3 scheme the entire frequency
band is divided equally into three sub-bands such that each cell is assigned only one of the
available sub-bands and the frequency bandwidth is reused every three cells rather than every
cell like Reuse-1 case. We can notice that the amount of inter-cell interference ICI has been
highly decreased due to limiting most of dominant interferers by operating on a different
sub-bands while at the same time the spectral efficiency of the limited frequency resources
also decreased by wasting most of them to avoid interference.
In figure 5.2 we describe the operation of Reuse-3 scheme. Each macrocell is divided into
three different sectors and each sector is served by a different directional antenna. The three
sectors are assigned the three different sub-bands such that the interference is only limited to
one sector (instead of 3 in Reuse-1 case) that operates on the same sub-bands in neighboring
macrocells. The transmission power level in each sector is set as 3P where Pis the reference
power level used in Reuse-1 transmission [34]. For femtocell operation we propose that
femtocells at each sector operate on the two remaining sub-bands not used by macrocell. This
procedure provides almost complete frequency separation between macrocell and femtocell
networks but at the expense of spectral efficiency.
28

5.3 Soft Frequency Reuse (SFR) Scheme
It can be noticed that Reuse-1 alone results in a big amount of interference and Reuse-3 alone
results in poor spectral efficiency and waste of frequency resources. A mix of Reuse-1 and
Reuse-3 has been proposed in [35] such that the composite scheme is a compromise between
the two different schemes. This scheme is called Soft Frequency Reuse (SFR) and can be
described as follows.
Figure 5.3: SFR Scheme
Figure 5.3 describes the frequency allocation process for both macrocells and femtocells
applied in SFR scheme. SFR scheme divides the coverage area of macrocell into interior
(center) region and edge region as shown in figure 5.3. The definition of center region is
defined by the radius from the center of macrocell where macro BS is located. The optimal
radius that maximizes system throughput has been calculated in [36] and found to be 63%
of cell radius. Based on dividing the macrocell into center region and edge region, conse-
quently MUEs are classified into center MUEs and edge MUEs based on their location in the
macrocell.
The LTE frame is divided into two time slots [37]. The first time slot in SFR scheme is
reserved for access of center MUEs where they can use freely the entire frequency bandwidth
during this time slot (Reuse-1 operation). The second time slot is reserved for access of edge
MUEs. The entire frequency bandwidth in the second time slot is divided equally into three
sub-bands such that edge MUEs at each sector of each macrocell can access only one of
29

the three available frequency sub-bands (Reuse-3 operation) as shown in figure 5.3. The
transmission power level of the edge region is set to be three times the transmission power
level of center region [38].
We then propose an allocation scheme for femtocells that copes with macrocell allocation
procedure that was described in previous paragraph and doesn’t require neither synchroniza-
tion nor signalling between the two types of networks. Femtocells will be also categorized
into center femtocells and edge femtocells according to their location in the macrocell as
shown in figure 5.3. Let the entire frequency bandwidth is divided equally into three sub-
bands A, B, and C. Center femtocells at each sector will operate only on one sub-band such
that this sub-band is not accessed by edge MUEs at this sector and different from the fre-
quency sub-bands used by center femtocells in neighboring sectors on the same macrocell.
Edge femtocells at each sector can use the two sub-bands not accessed by edge MUEs at
this sector. In general the sub-bands accessed by edge MUEs at each sector is prohibited for
access by femtocells to avoid strong interfering transmission. For example, if macrocell uses
sub-band A to serve edge region at sector 1, then center femtocells will use either sub-band
B or C and edge femtocells will use both sub-bands B and C.
We assume that this procedure for both macrocell and femtocell allocation provides an
efficient mutual Macro-Femto interference management scheme due to the following rea-
sons:
1. The Macro-to-Femto interference in the center region only exists during the first time
slot.
2. Femto-to-Macro interference in center region is attenuated to 1/3 as femtocells only
transmit on 1/3 of the allocated Bandwidth and hence can be tolerated by center MUEs
as they have a high received power from their serving macro BS.
3. Macro-to-Femto interference over FUEs associated with edge femtocells is minimized
due to relatively large distance from the center interfering macro BS.
30

4. Femto-to-Macro interference occurred by edge femtocells is only limited to center
MUEs and is negligible due to low femtocell power.
Figure 5.4: PFR Scheme
5.4 Partial Frequency Reuse (PFR) Scheme
One variation of the concept of Fractional Frequency Reuse (FFR) is called Partial Fre-
quency Reuse (PFR). PFR has been proposed in [39] as a technique for for resource alloca-
tion in macro-cellular networks. The concept of PFR is somehow similar to SFR as it also
divides the macrocell coverage area into center region and edge region based on the same
optimal radius used in section 5.3.
Figure 5.4 describes the macrocell frequency allocation process in PFR scheme. The
entire system bandwidth is divided into 6 sub-bands as shown in figure 5.4. Three of these
sub-bands (called common sub-bands) are reserved for access of center MUEs at any sector.
The three remaining sub-bands are reserved for access of edge MUEs of the three different
sectors such that each sector is assigned on sub-band for its UEs’ access. Based on [40],
the transmission power level over the sub-bands used by edge MUEs is 2/3 of the total
transmission power assigned for macro BS while the transmission power level over sub-
bands used by center MUEs is 1/3 of total transmission power.
We propose a femtocell allocation process as described in figure 5.4. Center femtocells at
each sector will operate on sub-bands not used neither by center MUEs nor by edge MUEs.
31

Edge femtocells will operate on the same sub-bands like center femtocells in addition to
common sub-bands. Since edge femtocells are far enough from center macro BS, the amount
of interference power level received becomes very limited.
If the entire system bandwidth is divided into 6 sub-bands A, B, C, D, E, and F. Com-
mon sub-bands A, B, and C are reserved for center MUEs while sub-bands D, E, and F are
reserved for edge MUEs located at the three different sectors. If we assign sub-band D for
edge MUEs at sector 1, then sub-bands E and F will be used by center femtocells. Edge
femtocells will use the same two sub-bands E and F in addition to common sub-bands (A,
B, and C). Like Reuse-3 scheme, PFR can almost provide complete separation in frequency
between the two different networks of LTE femtocell system.
5.5 Soft Fractional Frequency Reuse (SFFR) Scheme
Soft Fractional Frequency Reuse (SFFR) is another variation of FFR proposed firstly in
[41]. Similar to SFR and PFR, SFFR also divides the macrocell coverage area two distinct
regions;center region and edge region. The optimal radius of center region is also set to 63%
of cell radius as calculated in [36]. The concept of SFFR is very similar to PFR except for
one variation as will be described below.
Figure 5.5: SFFR Scheme
Figure 5.5 describes the macrocell frequency allocation process used in SFFR scheme.
Similar to PFR, the entire system bandwidth is also divided into 6 sub-bands. Three of them
32

are reserved for center MUEs access (common sub-bands) and the remaining three sub-
bands are distributed over the three different sector for edge MUEs access (one sub-band
each). The only difference between PFR and SFFR is the capability of center MUEs at any
sector to access two additional sub-bands assigned for edge MUEs access of the two other
sectors but with limited transmission power to minimize interference as much as possible.
This enhancement is assumed to increase spectral efficiency performance over that of PFR.
For example, if sub-band D is reserved for edge MUEs access at sector 1, center MUEs at
sector 1 can access common sub-bands A, B, and C with a specific power level while they can
also access sub-bands E and F but with limited power level to minimize interference level.
A quarter of the total transmission power assigned to macro BS is reserved for transmission
over common sub-bands. The ratio of transmission power level over edge region sub-bands
to the transmission power level over the two additional sub-bands of center region is set to
be [10:1] [42].
Figure 5.5 shows that the femtocell allocation process in SFFR scheme is the same like
that of PFR scheme without any change. In general, we can see that PFR scheme is a spe-
cialized case from the general SFFR scheme by setting the power of additional sub-bands of
center region to zero power.
33

CHAPTER 6
EVALUATION METHODOLOGY &
SIMULATION RESULTS
In this chapter we describe the evaluation metrics used in evaluating the performance of our
LTE femtocell system and our proposed frequency allocation schemes described in details in
Chapter 5. These evaluation metrics are targeting the evaluation of different aspects of the
network such as throughput performance, coverage performance, Quality of Service (QoS)
performance and finally the fairness performance of LTE femtocell system. The simulation
setup and system parameters are described in details in this chapter. Then we show the
simulation results of our research work and give comments and conclusions based on these
results. The second part of this chapter provides an optimization analysis for one of our
proposed frequency allocation schemes which is Soft Frequency Reuse (SFR) in order to
enhance the overall system performance of both network types. Simulation results of this
optimization analysis are then provided and comments are given on these results. We start
with describing evaluation metrics used as will be shown below.
6.1 Evaluation Metrics
6.1.1 Coverage Performance
The coverage performance in LTE femtocell network is measured via two main metrics.
1. SINR maps
34

The region of interest (ROI) of our LTE femtocell network is divided into small-
squared pixels such that the area of each pixel is 5m×5m. Each pixel is localized
by cartesian coordinates in ordinary x-y domain. The x-y coordinates help to calculate
distance from different pixels to either fixed-located macro BSs or randomly dropped
femto BSs in the network. Each pixel is assumed to represent the location of a FUE if
it is located within the coverage are of a femtocell, the location of indoor MUE if it is
located inside a house, or the location of an outdoor MUE elsewhere.
Using the SINR equations stated in section 4.3.1 and for the different frequency allo-
cation schemes described in Chapter 5, we can easily calculate SINR values for both
MUEs and FUEs at all pixels of ROI based on the location of the pixel and the allo-
cation process used. We can easily translate these calculated values into SINR maps
plotted in x-y domain that describes the distribution of SINR values for both MUEs
and FUEs for each frequency allocation scheme.
2. 10%-tile SINR
Coverage performance is usually a crucial issue for UEs suffering from poor SINR.
For LTE femtocell networks, those UEs with poor SINR values could be indoor MUEs,
MUEs located at cell edges, FUEs far from their serving femto BS, or MUEs/FUEs
subjected to strong interfering transmission. We use the 10%-tile SINR as metric for
measuring coverage performance of all UEs available in the network. The 10%-tile
SINR is calculated for different frequency allocation schemes and for different active
femtocell deployment densities.
6.1.2 Throughput Performance
The second aspect evaluated in our research work is the system throughput. Our methodol-
ogy for evaluating system throughput depends on measuring users capacity. Using equation
4.11 in section 4.3.2 we can easily evaluate the user capacity for any UE (MUE or FUE) in
our LTE femtocell network. The calculated user capacity for any UE represents the maxi-
35

mum data rate that this UE can achieve based on the assumption that one UE at a time can
use all the available resources assigned for its serving BS. By calculating the users capacity
achieved by all UEs available in the network, we can easily calculate two average values
which are; average value of users capacities of MUEs and average value of users capacities
of FUEs. Finally we can express the overall average user capacity of all UEs available in the
network by using weighted average of those two values using equation 6.1.
Coverall =NMUE .CM
avg +NFU E .CF
avg
NMUE +NFU E
(6.1)
where NMUE and NFU E are the number of MUEs and FUEs available in the network
respectively. CM
avg and CF
avg are average user capacities of MUEs and FUEs respectively. We
assume that this overall average user capacity can express average throughput performance
of the whole network. We calculate this overall average user capacity for different frequency
allocation schemes and for different active femtocell deployment densities.
6.1.3 Quality of Service (QoS) Performance
The quality of service (QoS) performance of our LTE femtocell network is measured via
outage probability Poutage. Outage Probability Poutage is defined as the probability of all UEs
(MUEs + FUEs) having SINR values below a predefined SINR threshold. Suppose γis a
vector containing all calculated SINR values for both MUEs and FUEs based on analysis in
section 4.3.1 such that γ= [γ1γ2......γMγM+1γM+2......γM+F] where M and F are number of
MUEs and FUEs in the network respectively. The outage probability can then be expressed
mathematically as follows.
Poutage =Pr(γ<γth)(6.2)
where γth is the threshold SINR value.
The explanation of using SINR values to evaluate QoS comes from that SINR values
usually reflect channel quality indicator (CQI) values. CQI values are sent as feedback by
36

UEs to the serving base station (BS) to determine the modulation and coding scheme (MCS)
used in LTE network [43]. The MCS determines the transmission rate and hence QoS in the
network. The outage probability is calculated for a range of SINR threshold values and for
different proposed frequency allocation schemes.
6.1.4 Fairness Performance
Fairness is a very crucial requirement for any cellular network. It is not sufficient to have
a system with high throughput with no guarantee of fair distribution of system resources.
Fairness performance of our LTE femtocell network is evaluated via two main metrics.
1. Jain’s index
Jain’s index is a very famous parameter used for measuring fairness [44]. Suppose
we have a vector Ccontaining all user capacity values calculated based on analysis
of section 4.3.2 for both MUEs and FUEs available in the network such that C =
[C1C2......CMCM+1 CM+2......CM+F] where M and F are numbers of MUEs and FUEs
respectively. The jain’s index calculated for this user capacity vector can be expressed
in equation 6.3 below.
J(C1,C2,...,CM,CM+1,...,CM+F) = M+F
∑
i=1
Ci2
(M+F).M+F
∑
i=1
C2
i
(6.3)
Jain’s index is calculated for different proposed frequency allocation schemes and for
different active femtocell deployment densities.
2. Fairness Ratio
We define another metric for measuring fairness performance of LTE femtocell net-
work. This metric is called Fairness Ratio [45]. Fairness Ratio is defined as the ratio
of 5%-tile capacity calculated from vector C used in Jain’s index calculation to the
37

overall average user capacity calculated in section 6.1.2. This definition can be ex-
pressed mathematically in equation 6.4 below.
Fairness Ratio =5%-tile Capacity
Coverall
(6.4)
Fairness Ratio is calculated for different proposed frequency allocation schemes and
for different active femtocell densities.
6.2 Simulation Setup
Our LTE cellular network is composed of 7 macrocells, and femtocells are randomly dropped
over the macrocells. The number of active femtocells is varied from 30 to 180 to study the
effect of varying femtocell deployment density on the whole network performance. MUEs
and FUEs are located within all possible locations of macrocells and femtocells coverage
areas respectively. Either MUE or FUE at a time is allocated all available resources assigned
to its serving macro or femto BS respectively. Three sector antennas are installed at each
macro BS to serve the three different sectors such that the maximum allowed transmission
power for each sector antenna is 20 W. All femtocells are assumed to use a limited transmis-
sion power of only 20 mW. AWGN channel model is assumed and the pathloss models are
calculated as explained in section 4.2.3. SINR and capacity values are then calculated for all
UEs and hence used for calculating evaluation metrics explained in section 6.1 above. The
simulation system parameters are summarized in Table 6.1.
6.3 Simulation Results
We plot some SINR maps as explained in Section 6.1.1 to evaluate coverage performance of
the whole network.
Figure 6.1 shows Macro-Femto SINR coverage map with focus only on the target center
macrocell for different proposed frequency allocation schemes with 60 deployed femtocells.
38

Parameter Value
LTE System Parameters
Inter-site Distance 1732 m
Propagation Model Sub-urban
Carrier Frequency 2 GHz
Bandwidth 5 MHz
FFT Size 512
Sub-carrier Spacing 15 KHz
CP Type Normal
Thermal Noise -174 dBm/Hz
Shadow Fading SD 8 dB
eNB Parameters
Number of eNBs 7
Number of Sectors per eNB 3
Max Tx Power 20 W
Max Sector Antenna Gain 15 dB
Antenna Pattern A(θ)=-min[12( θ
65◦)2,20]
Minimum Coupling Loss 70 dB
HeNB Parameters
Number of HeNBs per macrocell 30 ∼180
Femtocell Radius 20 m
HeNB Tx Power 20 mW
Antenna Gain 0 dB
Antenna Pattern Omni-directional
Minimum Coupling Loss 40 dB
UE Parameters
Antenna Gain 0 dB
Antenna Pattern Omni-directional
Rx Noise Figure 9 dB
Table 6.1: Simulation Parameters
39

The entire map represents the SINR performance of MUEs at all possible locations except
the locations that belong to femtocells’ coverage areas that represent SINR performance of
FUEs. Figure 6.1 addresses the different proposed frequency allocation schemes in addition
to the case when all femto BSs are turned OFF and Reuse-1 is operated. Figure 6.1(a) shows
that Reuse-1 scheme coverage performance is suitable for those MUEs that are close to cell
center as they receive good signal strength from their associated macro BS however there
is a coverage hole generated at the cell edge encountered by edge MUEs due to poor signal
strength and high level of ICI. Figure 6.1(b) shows that the presence of femtocells enhanced
the coverage performance of FUEs close to their associated femto BSs however the cover-
age hole encountered by edge MUEs still exists and new coverage holes are generated at
the edge of femtocells due to unmanaged interference between macrocells and femtocells,
interfering transmission of nearby femtocells, and poor signal strength received from femto
BSs at the cell edge of femtocells. Figure 6.1(c) and (e) show the SINR coverage perfor-
mance of Reuse-3 and PFR schemes respectively. Both schemes substantially eliminate the
coverage problem at both macrocell edge and femtocell edge due to complete frequency sep-
aration between macrocell and femtocell networks. Figure 6.1 (d) and (f) show the SINR
coverage performance of SFR and SFFR schemes respectively. Both schemes also eliminate
macro coverage hole at cell edge and coverage holes generated at cell edge of edge femto-
cells. However SINR performance at cell edge of center femtocells is somehow lower due to
sharing some sub-channels with macrocell transmission for center MUEs.
Figure 6.2 shows Macro SINR coverage map for different proposed frequency allocation
schemes with 60 deployed femtocells. The entire map here represents the all possible loca-
tions for MUEs only and the locations within the coverage area of femtocells are assumed
to be indoor MUEs rather than FUEs like the previous case. We can notice that the perfor-
mance of indoor MUEs is highly degraded with Reuse-1 scheme as shown in Figure 6.2(b)
due to large amount of ICI and interfering transmission of very close femto BSs while the
performance is the best for Reuse-3, and PFR schemes. The performance of center indoor
40

MUEs related to SFR and SFFR schemes are somehow lower ;but still very good compared
to Reuse-1 scheme; due to sharing some sub-channels with femtocell operation.
We address the SINR coverage map for another femtocell deployment density. Figure
6.3 shows Macro-Femto SINR coverage map for different proposed frequency allocation
schemes with 150 deployed femtocells in addition to the case when all femto BSs are turned
OFF with Reuse-1 operation. With dense deployment of femtocells, the interference prob-
lem becomes very severe as shown in Figure 6.3(b) as the co-layer interference between
femtocells becomes very significant. We can notice that the proposed frequency allocation
schemes have treated the cross-layer interference issue and still provide acceptable levels of
SINR coverage performance however there are some scenarios of very dense femtocell de-
ployment in a small geographic area where the co-layer interference becomes the dominant
type of interference which results in coverage holes at the edge of femtocells. The co-layer
interference problem needs to be addressed with different management schemes.
Similarly we address the indoor MUEs performance for the 150-femtocell deployment
scenario in Figure 6.4. Figure 6.3(b) shows the very poor performance of indoor MUEs
to very high interfering transmission of nearby femtocells. The indoor performance of
MUEs have been highly enhanced by different proposed schemes especially Reuse-3 and
PFR schemes. The performance of SFR and SFFR schemes are lower due to sharing some
sub-channels with femtocell operation and large amount of Inter-Femto Interference (IFI)
generated at this dense deployment scenario.
41

(a) Femtos are OFF (b) Reuse-1
(c) Reuse-3 (d) SFR
(e) PFR (f) SFFR
Figure 6.1: Macro-Femto SINR coverage map for different allocation schemes (60 deployed
femtocells)
42

(a) Femtos are OFF (b) Reuse-1
(c) Reuse-3 (d) SFR
(e) PFR (f) SFFR
Figure 6.2: Macro SINR coverage map for different allocation schemes (60 deployed fem-
tocells)
43

(a) Femtos are OFF (b) Reuse-1
(c) Reuse-3 (d) SFR
(e) PFR (f) SFFR
Figure 6.3: Macro-Femto SINR coverage map for different allocation schemes (150 de-
ployed femtocells)
44

(a) Femtos are OFF (b) Reuse-1
(c) Reuse-3 (d) SFR
(e) PFR (f) SFFR
Figure 6.4: Macro SINR coverage map for different allocation schemes (150 deployed fem-
tocells)
45

Figure 6.5: 10%-tile SINR vs. number of femtocells
The other metric used for measuring coverage performance as explained in Section 6.1.1
is 10%-tile SINR. Figure 6.5 shows the 10%-tile SINR (dB) of all network UEs (MUEs +
FUEs) against number of active femtocells deployed in the network for the different proposed
frequency allocation schemes. Reuse-1 scheme provides very poor 10%-tile SINR values
and hence very poor coverage especially for large number of active femtocells where the
performance dramatically decreases as shown. Other proposed frequency allocation schemes
like Reuse-3 and PFR can highly enhance the coverage performance and provide almost
stable performance as the number of active femtocells increases. This can be explained
as these schemes (e.g. Reuse-3 and PFR) provide almost complete frequency separation
between Macro and Femto networks. The two other proposed frequency allocation schemes;
SFR and SFFR also provide acceptable level of 10%-tile SINR coverage performance but
lower than Reuse-3 and PFR.
The evaluation metric used for measuring throughput performance is the overall average
user capacity as explained in Section 6.1.2. Figure 6.6 shows the overall average user ca-
pacity (Mbps) against number of active femtocells deployed in the network for the different
proposed frequency allocation schemes. Reuse-1 scheme can outperform all other proposed
46

Figure 6.6: Average users’ capacity vs. number of femtocells
schemes for only a small number of active deployed femtocells. This can be explained as
the amount of interference power is still limited and can be tolerated by different types of
UEs. SFR scheme provides the best performance as the number of active femtocells be-
comes much higher as shown. Due to the poor spectral efficiency, other proposed schemes
like Reuse-3, PFR, and SFFR provide lower throughput performance than Reuse-1 and SFR
schemes. Figure 6.6 shows that while the total system throughput increases by deploying
more femtocells, average user capacity may decrease with large number of femtocells like
Reuse-1 and SFR schemes.
The evaluation metric used for measuring Quality of Service (QoS) performance is the
outage probability as explained in Section 6.1.3. Figure 6.7 shows the outage probability
against a predefined SINR threshold. The number of active deployed femtocells is set to 90.
The predefined SINR threshold is varied from -5 dB to 30 dB by a step size of 5 dB. The
Reuse-1 scheme has very high outage probability compared to the other proposed frequency
allocation schemes. The Reuse-3 scheme, as expected, the best in term of QoS performance
due to the complete frequency separation between the Macro and Femto Networks and hence
most of the cross-layer interference that represents the dominant factor of interference is
47

Figure 6.7: Outage Probability vs. SINR threshold
eliminated. The other proposed frequency allocation schemes (e.g. SFR, PFR, and SFFR)
provide almost the same level of outage probability that is highly reduced from that of Reuse-
1 and near to that of Reuse-3 for small SINR threshold values up to 5 dB.
The first metric used for measuring Fairness is the Jain’s index of users capacity values
as explained in Section 6.1.4. Figure 6.8 shows the Jain’s index against number of active
femtocells deployed for different proposed frequency allocation schemes. Proposed Reuse-3
and SFR schemes are the best in terms of fairness performance while both of them slightly
decrease as number of active femtocells increases. The other proposed schemes (e.g. Reuse-
1, PFR, and SFFR) have lower fairness performance as shown in Figure 6.8. It can be
also noticed that the rate of Jain’s index decrease of Reuse-1 scheme as number of active
femtocells increasing is much higher than the rate of decrease of the other two schemes.
The second metric used for measuring Fairness is the Fairness ratio as explained in Sec-
tion 6.1.4. Figure 6.9 shows the fairness ratio against number of active femtocells deployed
for the different frequency allocation schemes. Similarly like the results of Jain’s index per-
formance, Reuse-3 and SFR are the best in terms of fairness performance with almost flat
performance as number of active femtocells increases. PFR and SFFR schemes also provide
48

Figure 6.8: Jain’s index vs. number of femtocells
flat performance against number of active femtocells but with lower fairness ratio values than
those of Reuse-3 and SFR schemes. Finally as expected, the fairness of the Reuse-1 scheme
is highly degraded at higher femtocell densities due to the unmanaged interference.
In order to choose the best proposed frequency allocation scheme from those mentioned
above, we depend on a tradeoff between fairness performance and throughput performance
based on calculated evaluation metrics. Figure 6.10 and 6.11 show this tradeoff between
fairness and throughput for two femtocell deployment scenarios. The medium femtocell
deployment scenario is shown in figure 6.10 where 90 active femtocells are deployed within
each macrocell. The heavy femtocell deployment scenario is shown in figure 6.11 where
150 active femtocells are deployed within each macrocell. Figure 6.10 and 6.11 show that
Reuse-3 and PFR schemes can provide high level of fairness at the expense of throughput.
Reuse-1 scheme can provide high level of throughput at the expense of very poor fairness
and QoS. We can notice that SFR scheme can provide the best throughput performance with
an acceptable level of fairness which is not too much far from those of Reuse-3 and PFR
schemes.
49

Figure 6.9: Fairness Ratio vs. number of femtocells
Figure 6.10: Fairness/Throughput Tradeoff (90 active femtocells)
6.4 Optimizing Interior Radius of Soft Frequency Resue
(SFR) Scheme
We have concluded at the end of Section 6.3 that our proposed SFR scheme can provide a
good tradeoff between fairness and throughput performance in out LTE femtocell network.
50

Figure 6.11: Fairness/Throughput Tradeoff (150 active femtocells)
Thus we try at this section to optimize the interior region radius of the SFR scheme using
exhaustive search method for the best possible performance as shown below.
The impact of changing the SFR interior region radius on the overall average user capac-
ity is shown in Figure 6.12. We set the SFR interior radius as a variable parameter during
simulation changing from 50% to 90% of cell radius with step size of 5%. Two femtocell
deployment scenarios are simulated; the relatively low deployment scenario represented by
60 active deployed femtocells and the relatively high deployment scenario represented by
150 active deployed femtocells. Figure 6.12 shows that the optimal interior radius that max-
imizes overall average users’ capacity and hence average throughput performance is found
to be 76% of cell radius for the relatively low deployment scenario. The same figure also
shows that the optimal interior radius that maximizes overall average users’ capacity and
hence average throughput performance is found to be 65% of cell radius for the relatively
high deployment scenario. We can observe that the optimal interior region radius is a factor
of the deployment density and decreases as number of active femtocells increases. As the
SFR interior radius decreases, more of the active femtocells are then considered Edge fem-
51

Figure 6.12: Average users’ capacity vs. SFR interior radius
tocells that can use more frequency sub-bands than Center femtocells and thus have higher
throughput.
We also study the impact of changing SFR interior radius on the fairness performance.
Using the same two deployment scenarios mentioned above and over the same range of SFR
interior radii, figure 6.13 shows the fairness ratio of the two deployment scenarios against
the SFR interior radius. Figure 6.13 shows that for the relatively low deployment scenario
the fairness ratio doesn’t decrease significantly as the SFR interior radius increases. On
the other hand, the same figure shows that the fairness ratio deceases dramatically for the
relatively high deployment scenario due to the increased amount of interference. We can
also notice from this figure that at the two optimal radii points of 76% and 65% calculated
in the previous paragraph, there is a very slight decrease in the fairness ratio by only 5% and
15% for the two deployment scenarios respectively. These results give a conclusion that the
two calculated optimal SFR interior radii of 76% and 65% of cell radius can provide a good
performance for both throughput and fairness.
We finalize with a comparison of SINR map distribution in figure 6.14 between the
Reuse-1 scheme with 60 active femtocells deployed and the proposed SFR scheme with
52

Figure 6.13: Fairness Ratio vs. SFR interior radius
also 60 active femtocells deployed and using the calculated optimal SFR interior radius. Fig-
ure 6.14(a) shows that the Reuse-1 scheme suffers from coverage holes at the edge regions
of both macrocells and femtocells besides coverage holes due to overlapping femtocell cov-
erage regions. Figure 6.14(b) shows that the proposed SFR scheme with optimized interior
radius has solved the problem of coverage holes at edge regions of both macrocells and fem-
tocells. The problem of coverage holes due to neighbor femtocells still exists. However, it
can be alleviated by deploying inter-femto interference (IFI) mitigation schemes.
53

Figure 6.14: Macro-Femto SINR coverage map for Reuse-1 and SFR scheme with optimized
interior radius (60 deployed femtocells)
54

CHAPTER 7
CONCLUSION & FUTURE WORK
In this chapter we summarize the concluding remarks of our research work in this thesis and
the future work to be done.
7.1 Conclusion
This thesis addressed a very important research topic which is interference management in
LTE femtocell networks. The importance of this research work comes from the importance
of the femtocell technology itself as it is expected to be a very crucial part of the network
topology of modern mobile cellular networks such as LTE-Advanced. A brief overview
of the femtocell technology and femtocell network architecture was provided in this thesis
work. Wide deployment of femtocells will not be available until the severe problem of
mutual interference between the macrocell layer and femtocell layer is well managed.
The interference problem addressed by our research work has been defined and interfer-
ence scenarios have been analyzed into two main categories; cross-layer interference and
co-layer interference. The focus of this research work was oriented to the first type which
is cross-layer interference. A survey has been done on some of the pre-prosed interference
management techniques in the literature used for OFDMA femtocell networks. We analyzed
these schemes based on type of interference addressing, complexity, efficiency, coordination
required and transmission mode. We came to a conclusion that interference management
schemes based on efficient frequency allocation techniques especially those exploiting the
55

concept of Fractional Frequency Reuse (FFR) can be the best choice for interference man-
agement process.
We built our own system model by modifying the Vienna LTE Simulator to accept the
presence of the new network element which is the femtocell. We used the standardized
parameters for LTE cellular system, macro BS, femto BS, MUEs and FUEs. We provided a
mathematical analysis for both MUEs and FUEs performance in the network based on SINR
analysis and users capacity analysis.
We then proposed some frequency allocation schemes used for allocating resources for
both macrocells and femtocell to allow coexistence of both networks and minimize inter-
ference as much as possible. These allocation schemes exploit some of the well-know al-
location schemes used for macrocell operation such as Reuse-1, Reuse-3, SFR, PFR and
SFFR schemes. The main advantages of theses proposed schemes are being very simple for
practical implantation and nonnecessity of coordination between the two network types. We
evaluated the different proposed schemes via different metrics such as throughput perfor-
mance, coverage performance, QoS performance and fairness performance.
Simulation results showed that Reuse-1 scheme degrades network performance in terms
of coverage, QoS and fairness especially for large number of deployed femtocells. Reuse-
3 scheme is the best in terms of coverage, QoS and fairness, but throughput performance
is very poor due to poor spectral efficiency. PFR scheme is very good in terms of cover-
age and provided acceptable performance levels in terms of QoS and fairness. Similar to
Reuse-3 scheme, PFR provided very poor performance level in terms of throughput due to
poor spectral efficiency. SFFR scheme increased spectral efficiency rather than PFR scheme
and hence higher throughput performance but lower performance in terms of coverage and
fairness. SFR scheme increased spectral efficiency and hence provided very high throughput
performance that exceeded Reuse-1 in dense deployment of femtocells. It also provided very
good performance level in terms of fairness and acceptable levels in terms of coverage and
QoS. Thus we concluded that SFR scheme can be the best tradeoff among different proposed
schemes via different evaluation metrics. We also optimized SFR interior radius using ex-
56

haustive search for both optimal throughput and fairness performance for two deployment
scenarios. The two calculated SFR interior radii have been found to be the optimal in terms
of throughput and fairness. They also preserved the good coverage performance obtained by
the SFR scheme.
7.2 Future Work
A number of items have been proposed as a future work to complete the investigation of this
important research work as follows.
•Evaluating the system via other important metrics such as total system throughput and
packet latency.
•Investigating the impact of changing traffic type and load size on network performance.
•Investigating the impact of using famous types of scheduling algorithms such as round
robin, proportional fair, max CQI,...etc.
•Implementing new scheduling algorithms and evaluating them in our cellular system
model.
•Investigating the co-layer interference problem or so called inter-femto interference
(IFI) problem especially by management schemes depending on femto power control.
•Combining co-layer interference management schemes and cross-layer interference
management schemes and evaluating the whole network performance.
•Studying dynamic FFR schemes and proposing new allocation schemes based on dy-
namic FFR as it is very crucial in some deployment scenarios.
57

REFERENCES
[1] http://www.qualcomm.com/solutions/wireless-networks/technologies/
femtocells.
[2] 3GPP, “Requirements for Evolved UTRA (E-UTRA) and Evolved UTRAN (E-
UTRAN),” TR 25.913, 3rd Generation Partnership Project (3GPP), Mar. 2006.
[3] 3GPP, “UTRA-UTRAN Long Term Evolution (LTE) and 3GPP System Architecture
Evolution (SAE).” http://www.3gpp.org/Highlights/LTE/LTE.htm.
[4] 3GPP, “Physical layer aspect for evolved Universal Terrestrial Radio Access (UTRA),”
TR 25.814, 3rd Generation Partnership Project (3GPP), Oct. 2006.
[5] 3GPP, “Feasibility study for evolved Universal Terrestrial Radio Access (UTRA) and
Universal Terrestrial Radio Access Network (UTRAN),” TR 25.912, 3rd Generation
Partnership Project (3GPP), Aug. 2007.
[6] D. Chambers, “Crystal frequency oscillators in femtocells.” http://www.
thinkfemtocell.com.
[7] A. Saleh, A. Rustako, and R. Roman, “Distributed antennas for indoor radio commu-
nications,” Communications, IEEE Transactions on, vol. 35, no. 12, pp. 1245–1251,
1987.
[8] M. Smith, A. Bush, P. Gwynn, and S. Amos, “Microcell and picocell base station inter-
nal antennas,” in Wireless Communications and Networking Conference, 1999. WCNC.
1999 IEEE, pp. 708–711, IEEE, 1999.
[9] V. Chandrasekhar, J. Andrews, and A. Gatherer, “Femtocell networks: a survey,” Com-
munications Magazine, IEEE, vol. 46, no. 9, pp. 59–67, 2008.
[10] J. Zhang, G. De la Roche, and MyiLibrary, Femtocells: Technologies and Deployment.
Wiley Online Library, 2010.
[11] http://www.smallcellforum.org/.
[12] J. Boccuzzi and M. Ruggiero, Femtocells: design & application. McGraw-Hill Profes-
sional, 2010.
[13] A. Golaup, M. Mustapha, and L. Patanapongpibul, “Femtocell access control strategy
in umts and lte,” Communications Magazine, IEEE, vol. 47, no. 9, pp. 117–123, 2009.
58

[14] G. Cormode and M. Thottan, Algorithms for next generation networks. Springer-Verlag
New York Inc, 2010.
[15] R. Karlsson, “Radio resource sharing and capacity of some multiple access methods
in hierarchical cell structures,” in Vehicular Technology Conference, 1999. VTC 1999-
Fall. IEEE VTS 50th, vol. 5, pp. 2825–2829, IEEE, 1999.
[16] Y. Wu, D. Zhang, H. Jiang, and Y. Wu, “A novel spectrum arrangement scheme for
femto cell deployment in lte macro cells,” in Personal, Indoor and Mobile Radio Com-
munications, 2009 IEEE 20th International Symposium on, pp. 6–11, IEEE, 2009.
[17] H. Li, X. Xu, D. Hu, X. Qu, X. Tao, and P. Zhang, “Graph method based clustering
strategy for femtocell interference management and spectrum efficiency improvement,”
in Wireless Communications Networking and Mobile Computing (WiCOM), 2010 6th
International Conference on, pp. 1–5, IEEE, 2010.
[18] 3GPP, “Further Discussion on HeNB Downlink Power Setting in HetNet,” R1 -105238,
3GPP RAN1 Meeting, Xian, China, Oct. 2010.
[19] S. Guruacharya, D. Niyato, E. Hossain, and D. Kim, “Hierarchical competition in
femtocell-based cellular networks,” in GLOBECOM 2010, 2010 IEEE Global Telecom-
munications Conference, pp. 1–5, IEEE, 2010.
[20] L. Zhang, L. Yang, and T. Yang, “Cognitive interference management for lte-a femto-
cells with distributed carrier selection,” in Vehicular Technology Conference Fall (VTC
2010-Fall), 2010 IEEE 72nd, pp. 1–5, IEEE, 2010.
[21] T. Kim and T. Lee, “Throughput enhancement of macro and femto networks by fre-
quency reuse and pilot sensing,” in Performance, Computing and Communications
Conference, 2008. IPCCC 2008. IEEE International, pp. 390–394, IEEE, 2008.
[22] P. Lee, T. Lee, J. Jeong, and J. Shin, “Interference management in lte femtocell systems
using fractional frequency reuse,” in Advanced Communication Technology (ICACT),
2010 The 12th International Conference on, vol. 2, pp. 1047–1051, IEEE, 2010.
[23] R. Juang, P. Ting, H. Lin, and D. Lin, “Interference management of femtocell in macro-
cellular networks,” in Wireless Telecommunications Symposium (WTS), 2010, pp. 1–4,
Ieee, 2010.
[24] J. Ikuno, M. Wrulich, and M. Rupp, “System level simulation of lte networks,” in
Vehicular Technology Conference (VTC 2010-Spring), 2010 IEEE 71st, pp. 1–5, Ieee,
2010.
[25] 3GPP, “Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Architec-
ture description,” TS 36.401, 3rd Generation Partnership Project (3GPP), Sept. 2008.
[26] FemtoForum, “Interference management in OFDMA femtocells,” TR , FemtoForum,
Mar. 2010.
[27] 3GPP, “Evolved Universal Terrestrial Radio Access (E-UTRA); Radio Frequency (RF)
system scenarios,” TR 36.942, 3rd Generation Partnership Project (3GPP), Sept. 2008.
59

[28] J. Javaudin, J. Lainé, D. Lacroix, and O. Seller, “On inter-cell interference in ofdma
wireless systems,” in Proceedings of the EUSIPCO Conference. Las Vegas, 2005.
[29] K. Begain, G. Rozsa, A. Pfening, and M. Telek, “Performance analysis of gsm net-
works with intelligent underlay-overlay,” in Computers and Communications, 2002.
Proceedings. ISCC 2002. Seventh International Symposium on, pp. 135–141, IEEE,
2002.
[30] F. Khan, LTE for 4G mobile broadband: air interface technologies and performance.
Cambridge University Press, 2009.
[31] S. Kim, K. Cho, D. Yoon, Y. Ko, and J. Kwon, “Performance analysis of downlink inter
cell interference coordination in the lte-advanced system,” in Digital Telecommunica-
tions, 2009. ICDT’09. Fourth International Conference on, pp. 30–33, IEEE, 2009.
[32] Y. Wang, S. Kumar, L. Garcia, K. Pedersen, I. Kovács, S. Frattasi, N. Marchetti, and
P. Mogensen, “Fixed frequency reuse for lte-advanced systems in local area scenarios,”
in Vehicular Technology Conference, 2009. VTC Spring 2009. IEEE 69th, pp. 1–5,
IEEE, 2009.
[33] 3GPP and Huawei, “Soft Frequency Reuse Scheme for UTRAN LTE,” R1 -050507,
TSG RAN WG1 Meeting 41, Athens, Greece, May 2005.
[34] Z. Xie and B. Walke, “Enhanced fractional frequency reuse to increase capacity of
ofdma systems,” in New Technologies, Mobility and Security (NTMS), 2009 3rd Inter-
national Conference on, pp. 1–5, Ieee, 2009.
[35] 3GPP and Huawei, “Further analysis of soft frequency reuse scheme,” R1 -050841,
TSG RAN WG1 Meeting 42, Aug. 2005.
[36] M. Assaad, “Optimal fractional frequency reuse (ffr) in multicellular ofdma system,”
in Vehicular Technology Conference, 2008. VTC 2008-Fall. IEEE 68th, pp. 1–5, IEEE,
2008.
[37] 3GPP, “Evolved Universal Terrestrial Radio Access (E-UTRA); Long Term Evolution
(LTE) physical layer; General description,” TS 36.201, 3rd Generation Partnership
Project (3GPP), Dec. 2007.
[38] Z. Xie and B. Walke, “Frequency reuse techniques for attaining both coverage and
high spectral efficiency in ofdma cellular systems,” in Wireless Communications and
Networking Conference (WCNC), 2010 IEEE, pp. 1–6, IEEE, 2010.
[39] M. Rahman and H. Yanikomeroglu, “Enhancing cell-edge performance: a downlink
dynamic interference avoidance scheme with inter-cell coordination,” Wireless Com-
munications, IEEE Transactions on, vol. 9, no. 4, pp. 1414–1425, 2010.
[40] B. Krasniqi, M. Wolkerstorfer, C. Mehlführer, and C. Mecklenbräuker, “Weighted sum-
rate maximization for two users in partial frequency reuse cellular networks,” 11th
MCM COST2100, Aalborg, Denmark, 2010.
60

[41] I. C802.16m-08/782, “Fractional Frequency Reuse in Uplink,” Tech. Rep. , J.Y. Han,
H.G. Cho (LG. Electronics), July 2008.
[42] Y. Kwon, O. Lee, J. Lee, and M. Chung, “Power control for soft fractional frequency
reuse in ofdma system,” Computational Science and Its Applications–ICCSA 2010,
pp. 63–71, 2010.
[43] 3GPP, “Physical layer procedures (FDD),” TS 25.214, 3rd Generation Partnership
Project (3GPP), Sept. 2008.
[44] R. Jain, The art of computer systems performance analysis. John Wiley & Sons, 2008.
[45] 3GPP, “Interference Mitigation by Partial Frequency Reuse,” R1 -060135, TSG-RAN
WG1 Ad Hoc Meeting on LTE, Helsinki, Finland, Jan. 2006.
61