understanding and development of inter-cell interference ...651719/fulltext01.pdf · inter-cell...
TRANSCRIPT
i
Understanding and Development of Inter-cell Interference Mitigation mechanism in LTE-A Heterogeneous Network
Förståelse och utveckling av Inter-interferens Mitigation mekanism i LTE-A
heterogent nätverk
Bilal Shah
Suman Ghimire
Faculty of Health, Science and Technology
Master’s Program in Electrical Engineering
Degree Project of 30 credit points
Supervisor: Arild Moldsvor (Karlstad University), Pechetty V. Prasad and Bengt Hallinger (Tieto Sweden AB)
Examiner: Jorge Solis (Karlstad University)
Date: 5th september 2013
Serial number:
iii
Abstract
In long term evolution Advanced (LTE-A), concept of heterogeneous network (HetNet) has
been introduced. Since, spectrum has become a rare resource these days; another mean is to be
looked after to improve the existing wireless technology. One possible way is to improve the
network topology so that frequency spectrum can be reused. In heterogeneous network, lower
power nodes like Pico/Femto cell are deployed inside Macro cell to increase the system
throughput and network coverage. Traditionally, cell selection for user equipment (UE) in LTE
is based upon the received downlink power but these Pico/Femto cell have a low power than the
Macro cell, meaning that few users can get access from Femto/Pico cells. UE should be close
to the Pico/Femto cell to get connected with it. So its solution is; cell selection based upon the
uplink path loss can be applied allowing more UE get connected to Pico/Femto cell. On doing
that area of the Pico/Femto cell will increase which is called range extension region.
Another problem arises when cell extension is applied, is that Macro cell imposes interference
towards the Physical channel and signal of the Pico/Femto cell UE in range extension region as
both Macro and Pico/Femto cell operate with same set of frequencies. 3rd Generation
Partnership Project (3GPP) LTE-A Enhanced Inter-Cell Interference Coordination (eICIC)
scheme has proposed Almost Blank Sub-frame (ABS) as a solution towards the above
mentioned interference problem by reducing the activity or muting the Macro sub-frame. So
that the corresponding Pico/Femto sub-frame can transmit the user information without
interference from Macro cell ABS. For the reason of backward compatibility ABS still transmit
certain physical channel and signals like CRSs, PCH, PBCH and PSS/SSS. So, interference still
remains in these signals and channels.
The main focus of thesis is reducing the impact of collision of cell-specific reference signal
(CRS) from Macro and Femto cell as CRS is used for channel estimation. We have developed
LTE link level system model for Macro and Femto cell in the Matlab simulator. Effect of
difference in power of Macro and Femto CRS on UE under different noise power is
investigated. It shows, higher the power of Macro, higher is the interference level. As a result
Femto channel estimation quality degrades which in-turn degrades system performance.
Combined receiver Interference cancelation (IC) methodology is implemented to reduce the
impact of interference between macro and Femto CRS collision, which is based upon the
reference signal received power (RSRP). System performance is evaluated with bit error rate
(BER) and block error rate (BLER) versus Signal to Noise Ratio (SNR) and compared with the
single cell system (without interference) and without IC system. Result confirms that IC method
system performance is far better than system without IC and as close to system performance of
single cell without interference. Furthermore, use of convolutional encoder, offer approximately
7dB coding gain in terms of SNR.
v
Acknowledgments
Firstly, we would like to thank Tieto Sweden AB, Karlstad for giving us opportunity to work
upon this thesis and providing such a great environment at their office. More importantly,
thanks to our company supervisors Pechetty V Prasad, Bengt Hallinger and internal supervisor
Arild Moldsvor, We are also thankful to Ireddy Chandra, Shilpika kappa and the whole base
band team who helped and guide us in our difficult time.
Secondly, we want to thanks to our families and friends for their continuous support and
patience.
vii
Contents
1 Introduction ............................................................................................................. 1
Introduction ................................................................................................................. 1 1.1
Goal ............................................................................................................................. 1 1.2
Problem definition ....................................................................................................... 2 1.3
Methodology ................................................................................................................ 2 1.4
Previous Work ............................................................................................................. 2 1.5
Delimitation and choices ............................................................................................. 3 1.6
Report Outline ............................................................................................................. 4 1.7
2 Overview of LTE ..................................................................................................... 5
Background .................................................................................................................. 5 2.1
2.1.1 Overview of LTE downlink physical layer ..................................................... 5
2.1.2 Requirements of LTE ..................................................................................... 5
2.1.3 Multiple Access Method in LTE .................................................................... 6
2.1.4 Cyclic-Prefix insertion .................................................................................... 6
2.1.5 Spectrum flexibility in LTE ............................................................................ 8
2.1.6 LTE Generic Frame Structure......................................................................... 9
LTE downlink channel and signals ........................................................................... 11 2.2
2.2.1 Downlink Reference Signals ........................................................................ 11
2.2.2 Downlink Physical Channels ........................................................................ 12
LTE-A Release 10 ..................................................................................................... 13 2.3
3 Heterogeneous network ........................................................................................ 15
Introduction to heterogeneous network ..................................................................... 15 3.1
3.1.1 Advantages of deployment of HeNBs .......................................................... 16
3.1.2 Properties of Macro eNBs and LPN ............................................................. 16
UE measurements in LTE.......................................................................................... 16 3.2
Introduction to range extension ................................................................................. 17 3.3
Problem caused by deployment of Femto eNBs ........................................................ 17 3.4
3.4.1 Evolution of eICIC ....................................................................................... 18
3.4.2 Carrier aggregation in LTE ........................................................................... 18
3.4.3 Almost Blank Sub-frame .............................................................................. 19
Colliding and Non colliding CRS .............................................................................. 20 3.5
4 Channel Model ...................................................................................................... 22
Introduction to Air Interface ...................................................................................... 22 4.1
4.1.1 Rayleigh distributed channel ........................................................................ 23
4.1.2 Rician distributed channel ............................................................................ 24
4.1.3 Maximum Doppler shift ............................................................................... 24
viii
Jake’s Channel model ................................................................................................ 24 4.2
5 Channel Estimation............................................................................................... 26
Channel estimation .................................................................................................... 26 5.1
Signal Model ............................................................................................................. 26 5.2
SINR .......................................................................................................................... 28 5.3
Least Square Estimator .............................................................................................. 28 5.4
Least Minimum Mean Square Error (LMMSE) Estimator ........................................ 29 5.5
Interpolation .............................................................................................................. 33 5.6
5.6.1 Frequency Interpolation ................................................................................ 33
5.6.2 Time Interpolation ........................................................................................ 33
Equalization ............................................................................................................... 34 5.7
6 System Model......................................................................................................... 36
Introduction to System Model ................................................................................... 36 6.1
Encoder ...................................................................................................................... 36 6.2
6.2.1 Cyclic redundancy check (CRC) .................................................................. 37
6.2.2 Convolutional encoder .................................................................................. 37
Digital modulation ..................................................................................................... 39 6.3
6.3.1 QPSK ............................................................................................................ 39
6.3.2 QAM ............................................................................................................. 39
MIMO- STBC ........................................................................................................... 40 6.4
Mapping of signal to the grid .................................................................................... 40 6.5
DFT/IDFT .................................................................................................................. 40 6.6
Cyclic Prefix (CP) ..................................................................................................... 42 6.7
Power Amplifier ........................................................................................................ 42 6.8
Air Interface ............................................................................................................... 44 6.9
Receiver ..................................................................................................................... 44 6.10
Core methodology ..................................................................................................... 45 6.11
6.11.1 Direct IC ....................................................................................................... 47
6.11.2 Joint channel Detection ................................................................................. 48
6.11.3 No IC ............................................................................................................ 50
7 Simulation and Results ......................................................................................... 51
Simulation set up ....................................................................................................... 51 7.1
7.1.1 NO IC Scheme Based results ........................................................................ 52
7.1.2 Conclusion based on No IC scheme ............................................................. 56
7.1.3 IC Scheme based Results .............................................................................. 56
7.1.4 Conclusion based on IC scheme ................................................................... 59
ix
8 Conclusion and Future work ............................................................................... 60
Conclusion ................................................................................................................. 60 8.1
Future Work ............................................................................................................... 61 8.2
9 References .............................................................................................................. 62
10 Appendix A ............................................................................................................ 65
A.1 LTE resource grid for LTE minimum bandwidth (1.4 MHz) .................................... 65
A.2 LTE resource grid for maximum bandwidth (20 MHz) ............................................. 66
xi
Table of figures
Figure 2.1: OFDM Orthogonal Sub-carriers ................................................................................. 6
Figure 2.2: Simple OFDM System ............................................................................................... 6
Figure 2.3: Graphical representation of Cyclic Prefix (CP) .......................................................... 7
Figure 2.4: OFDM and OFDMA graphical concept ..................................................................... 8
Figure 2.5: FDD and TDD ............................................................................................................ 8
Figure 2.6: LTE time domain frame structure. ............................................................................. 9
Figure 2.7: LTE time frequency resource grid ............................................................................ 10
Figure 2.8: LTE CRSs Position in a single Antenna ................................................................... 11
Figure 2.9: LTE CRSs mapping for 2x2MIMO .......................................................................... 12
Figure 3.1: Heterogeneous Networks .......................................................................................... 15
Figure 3.2: Range extension in Heterogeneous networks ........................................................... 17
Figure 3.3: Illustration of CSG and range extension user ........................................................... 18
Figure 3.4: Illustration of Carrier aggregation in LTE ................................................................ 19
Figure 3.5: Illustration of ABS in LTE ....................................................................................... 20
Figure 3.6: Illustration of Colliding and Non Colliding CRS ..................................................... 21
Figure 4.1: Multipath effect [25] ................................................................................................. 22
Figure 5.1: LTE time domain and frequency domain equalizer option ...................................... 34
Figure 6.1: System model ........................................................................................................... 36
Figure 6.2: Code rate 1/3, Convolutional encoder ...................................................................... 38
Figure 6.3: Convolutional encoding gain .................................................................................... 38
Figure 6.4: QPSK state diagram ................................................................................................. 39
Figure 6.5: QAM scheme diagram. A) QAM-16 B) QAM-64 ............................................... 39
Figure 6.6: DFT Block diagram .................................................................................................. 41
Figure 6.7: IDFT Block diagram ................................................................................................. 41
Figure 6.8: AM/AM & AM/PM mapping ................................................................................... 42
Figure 6.9: RAPP Amplifier smoothness factor is varying ......................................................... 43
Figure 6.10: RAPP Amplifier, Saturation level varying ............................................................. 43
Figure 6.11: 2x2 MIMO Configuration over air interface .......................................................... 45
Figure 6.12: Flow chart of IC Algorithm .................................................................................... 46
Figure 7.1: CSR, ABS and PDSCH configuration in LTE ......................................................... 52
Figure 7.2: UE receiving signal from Femto and Macro eNB .................................................... 52
Figure 7.3: BER when Femto RSRP is varying for QPSK ......................................................... 53
Figure 7.4: BER when Femto RSRP is varying for QAM-16 ..................................................... 53
Figure 7.5: BER when Macro RSRP is varying for QPSK ......................................................... 54
Figure 7.6: BER when Macro RSRP is varying for QAM-16 .................................................... 54
xii
Figure 7.7: BER when SNR is varying for QPSK ...................................................................... 55
Figure 7.8: BER when SNR is varying for QAM-16 .................................................................. 56
Figure 7.9: BER when SNR is varying for QPSK using IC scheme ........................................... 57
Figure 7.10: BER when SNR is varying for QAM using IC scheme.......................................... 58
Figure 7.11: Block error rate for QPSK, 1500 sub-frames ......................................................... 58
Figure 7.12: Block error rate for QAM-16, 1500 Sub-frames .................................................... 59
Figure A.1: LTE Resource grid for 1.4MHz bandwidth ............................................................. 65
Figure A.2: LTE Grid for 20MHz bandwidth ............................................................................. 70
xiv
List of Tables
Table 2.1: LTE downlink Bandwidths .......................................................................................... 9
Table 2.2: IMT advance requirements ........................................................................................ 13
Table 2.3: A comparisonof LTE and LTE-A Features and Characteristics [15] ........................ 14
Table 5.1: value for different modulation scheme and Code rate [35] .................................... 32
Table 7.1: Parameter used in Simulation .................................................................................... 51
xvi
List of Abbreviations
3G 3rd
Generations
3GPP 3rd Generation Partnership Project
4G 4th Generations
A ABS Almost Blank Sub-Frame
AWGN Additive White Gaussian Noise
B BER Bit Error Rate
BLER Block Error Rate
C CA Carrier Aggregation
CC Component Carrier
CCPCH Common Control Physical Channel
CDMA Code Division Multiple Access
CFR Channel Frequency Response
CIR Channel Impulse Response
COMIC Combined Interference Cancellation
CP Cyclic Prefix
CRC Cyclic Redundancy Check
CRS Cell Specific Reference Signal
CSG Closed Subscriber Group
D DDCE Decision Directed Channel Estimation.
DFT Discrete Fourier Transform
DPSK Differential Phase-Shift Keying
DVB-S Digital Video Broadcasting-Satellite
E EDGE Enhanced Data Rates for GSM Evolution
eICIC Enhanced Inter-Cell Interference Coordination
eNB Evolved Node Base Station
E-UTRA Evolved Universal Terrestrial Radio Access
ER Extension Region
xvii
F FDD Frequency-Division Duplex
G GSM Global System for Mobile Communication
H HeNB Home Evolved Node Base Station
HetNet Heterogeneous networks
HPN High Power Node
HSDPA High-Speed Downlink Packet Access
I IC+DDCE IC-Assisted Decision Directed Channel Estimation
ICIC Inter-Cell Interference Coordination
ID Identification
IDFT Inverse Discrete Fourier Transform
IFFT Inverse Fast Fourier Transform
IMT International Mobile Telecommunication
ISI Inter-Symbol Interference
ITU International Telecommunication Union
L LMMSE Least Minimum Mean Square
LOS Line Of Sight
LPN Low power networks
LS Least Square
LTE Long Term Evolution
LTE-A Long Term Evolution advanced
M MIMO Multiple Input Multiple Output
MMOG Multimedia Online Gaming
O OFDM Orthogonal Frequency Division Multiplex
OFDMA Orthogonal Frequency Division Multiple Access
P PBCH Physical Broadcast Channel
xviii
PCC Primary Component Carriers
PCH Paging Channel
PDCCH Physical Downlink Control Channel
PDF Power Delay Profile
PDSCH Physical Downlink Shared Channel
PHY Physical Layer
PRB Physical Resource Block
PSK Phase-Shift Keying
PSS Primary Synchronization Signals
Q QAM Quadrature Amplitude Modulation
QPSK Offset Quadrature Phase Shift Keying
R RB Resource Block
RSRP Reference Signal Received Power
RSRQ Reference Signal Received Quality
RSSI Received Signal Strength Indicator
RX Receiver
S SCC Secondary Component Carriers
SNR Signal to Noise Ratio
SSPA Solid State Power Amplifier
SSS Secondary Synchronization Signals
SIB System Information Block
T TDD Time-Division Duplex
TWTA Traveling Wave Tube Amplifiers
TX Transmitter
U UE User Equipment
UMTS Universal Mobile Telecommunications System
W Wi-Fi Wireless Fidelity
WiMAX Worldwide Interoperability for Microwave Access
1
Chapter 1
Introduction
Introduction 1.1
LTE is developed by the 3rd Generation Partnership Project (3GPP) in order to make sure
effectiveness of its standards in long term. Recently it is also known as 4th generation
technology. LTE is the evolution of 3rd
generation mobile technology also called as Universal
Mobile Telecommunications System (UMTS). The main challenges for LTE is to come up
with new radio access technology so that high data rates, low latency can be offered. LTE is
also known as E-UTRAN (Evolved UMTS Terrestrial Radio Access Networks).
Orthogonal frequency division multiplexing (OFDM) is the basis technology for the long term
evolution (LTE) where the channel estimation is done with the help of pilots or in LTE term
reference signals. LTE-A supports, heterogeneous network deployment which eliminate the
coverage holes of Macro base station and offers higher data rate and capacity. However such
deployments give rise to inter cell interference. In LTE-A, when inter-cell interference occurs
in HetNet Base station with higher power transmits ABS (almost blank sub-frame) to reduce
the interference however still there is a possibility of interferences due to the collision of
reference signal. General receivers in this situation cannot accurately estimate the channel so
advanced receiver with interference cancellation scheme is required for UE.
This master thesis investigates on the inter-cell interference caused by the collision of
reference signal from Macro and Femto eNBs and we have developed an interference
mitigation mechanism based upon the RSRP. It follows the specification of 3rd Generation
Partnership Project (3GPP) LTE-A. Detail description of LTE and OFDM is presented in
chapter 2. This chapter includes the Goal, Problem definition, Methodology, Previous work
and Thesis outline.
Goal 1.2
In General, goal of this thesis can be summarized in the following points.
To understand overall LTE system
To understand the interference that exists in heterogeneous networks
*To investigate on the 3GPP LTE interference cancellation and coordination schemes
like ICIC and eICIC
**Investigation on mechanism for successful decoding of the control and traffic
information under interference condition
*Modeling & simulation of physical layer for the Evolved Node Base Station (eNB),
Home Evolved Node Base Station (HeNB), UE and Channel conditions in Matlab
**Verifying the model with different power level of eNB and HeNB
Evaluating and find out BER/BLER at different SNR to check the system
performance
(Note: Throughout the thesis we have worked together, however there are some task
that is contributed mostly by individual. In the above ‘*’ represents the major
contribution made by Bilal Shah and ‘**’ represents the major contribution made by
Suman Ghimire)
2
Problem definition 1.3
The purpose of the project is to understand the interference between the Macro and Femto
cells with respect to the CRSs of Macro and Femto cell in heterogeneous Network which
operate with same set of frequencies, and come-up with mechanisms to minimize the inter cell
interference.
LTE Release 10 has the feature called eICIC which is developed to mitigate the interference
due to collision of CRSs from Macro and Femto cell. The eICIC is able to remove the
interference towards the data signal of the corresponding sub-frames of the co-channel victim
Femto cell with the help of Almost Blank Sub-frame (ABS). However, even with the usages
of ABS, interference from CRSs and other physical signals still appears. For the reason of
backward compatibility ABS transmits certain physical channel and signals like CRSs, PCH,
PBCH and PSS/SSS. In this work, we have considered the interference due to the collision of
CRSs from Macro and Femto cell. As CRSs are primarily used for the channel estimation and
collision of CRSs can significantly reduce the system performance. So, we have investigated
and developed a mechanism to minimize this interference caused by the collision of CRSs of
Macro and Femto cell.
The concept of colliding and non-colliding case of CRSs interference is described more
specifically and in depth detail in the section 3.5.
Methodology 1.4
LTE employs Orthogonal Frequency Division Multiple Access (OFDMA) and Single Carrier
Frequency Division Multiple Access (SC-FDMA) technology as the multiplexing scheme for
Downlink and uplink respectively.
We have considered the Heterogeneous network where low power base stations are deployed
inside High power base stations. We have modeled Macro Base station as a high power node
(HPN) and Femto base station as low power node (LPN). Moreover, UE is in the cell edge
region which also called cell extension region in LTE term, trying to access Femto cell. At the
same time UE is subject to high interference from Macro cell. Cell ID uniquely represent the
cell and in our thesis it is used for CRS generation. RSRP is used in channel estimation and
also for the measurement of system performance.
The IC algorithm is based on the received signal received power (RSRP) of both Macro and
Femto cell as cell ID and RSRP of both cells are known to UE. When UE receives a combine
signal from Macro and Femto Base Station, Based upon the RSRP difference of Macro and
Femto cell UE Estimate the Macro signal or Interfering signal with direct IC method or Joint
detection method and subtract it from the total combined signal to get the desired reference
signal from the Femto cell. Which is further used in channel estimation of Femto reference
signal and then interpolation is done to get the channel estimation of data signal. Detail
description of System Architecture, there choices and Limitation and a mathematical and
theoretical approach to minimize the interference is presented in Chapter System model.
Previous Work 1.5
Since this interference problem is lately discovered and still the hot topic among LTE
professionals, only few works have been done in Similar Interference scenario which can be
found in the [1], [2], [3] and [4]. The Interference cancellation technique used in this work is
modification and extension of [2] and [3].
3
In [1], Interference is due to the collision between reference signal and synchronization
signals, In this paper methodology is not described as Receivers algorithm are designed by
Manufacturers and they don’t usually disclose the methodology.
In [2], there is one macro cell, one Femto cell and one cell edge user in a Heterogeneous
Network. The interference is due to the collision of reference signal from both Macro and
Femto cell. Combined IC, Decision direct Channel Estimation and IC assisted Decision direct
Channel Estimation has been proposed. Results show that proposed methodology significantly
reduces the interference. In [3] the methodology is same but they have modeled more than one
Interfering cell i.e. Macro cell.
In [4], 3GPP ICIC technique is used where first, it is decided how much power Macro BS can
transmit so that Pico ER UEs get the desired DL SINR. This requires some form of
coordination between macro BSs and a Pico BSs. On the basis of decided power Macro BS
allocate RBs and transmit powers to its DL UEs, while respecting the power constraints
previously derived to support Pico ER UEs.
In the above mentioned references all other except [4] have used time domain eICIC method
for interference minimization. Beside above mentioned technique, sub-frame shift is a
solution to the inter-cell interference problem but sub-frame shift cannot be applied to LTE
TDD system and another method could be configuring ABS to be MBSFN, However MBSFN
cannot be always configurable.
Delimitation and choices 1.6
This project work is mainly focused upon the LTE Downlink so we have implemented
OFDMA system and we have followed the com-IC method of [2], we have used 20 MHz
bandwidth, Jakes Channel Model, Convolutional encoder, STBC, QPSK, QAM and RAPP
power amplifier. We have measured the system performance with BER and BLER, However
In [2], 5 MHz bandwidth, QPSK, WINNER II C2 (=EVA) and C3 (= ETU) channel model
and Turbo coding is used. The system performance is measured with System throughput and
BER. In [3], they have followed the same approach with theoretical information but
considered more than one Macro Interfering Cell.
The reason for the different choices we used is generally for easy implementation to avoid
complexity and more specifically to improve system performance. We implemented CRC and
convolutional encoder which are for error detection and correction respectively and also to
check the packet error rate or block error rate of the system which is more effective way to
analyses the system performance. The implementation of STBC delivers transmits antenna
diversity. Alamouti Scheme is one of the simplest schemes of STBC and more specific for
2x2 MIMO configurations and has full rate code i.e. rate 1code. In modulation scheme QPSK
is simplest for low and reliable data rate communication whereas QAM is demanded for the
high data rate in digital communication. Macro and Femto cell which are commonly
differentiated from each other in terms of power and user, so RAPP power amplifier fulfill our
requirements of power variation of transmit signal and it is simple and easy to implement.
Modeling of channel is complex and time consuming. So, we have used Jake’s channel model
provided by Company in order to save time.
4
Report Outline 1.7
The outline of this master thesis is as follows:
Chapter 2: Describes the detail of LTE downlink physical layer along with OFDM and,
reference signals.
Chapter 3: Describes the theory of heterogeneous network along with UE measurements in
LTE, range extension and eICIC
Chapter 4: Describes the air Interface which includes Jakes Channel model.
Chapter 5: Describes the theory behind Channel estimation technique like Least Square (LS)
error and Least Minimum Mean Square Error (LMMSE) and Interpolation that is
used in the methodology.
Chapter 6: Demonstrate the overall system model and present important aspect of the model.
Chapter 7: Analyses the simulation, results and finding of our thesis.
Chapter 8: Describe the conclusion that is drawn from the work and possible extension of the
work in the future.
5
Chapter 2
Overview of LTE
Background 2.1
The recent increase of mobile data usage and emergence of new applications such as MMOG,
mobile TV, Web 2.0, streaming contents have motivated the Third-generation Partnership
Project (3GPP) to work on the Long term evolution (LTE). The main objectives of LTE are
to minimize the system and User Equipment (UE) complexities, allow flexible spectrum
deployment in existing or new frequency spectrum and to enable co-existence with other
3GPP Radio Access [5]
The third Generation Partnership Project (3GPP) has introduced LTE as a next generation IP-
based OFDMA technology. The reason is to facilitate increasing demand of mobile data usage
and new multimedia applications. LTE is advancement of GSM/EDGE and UMTS/HSDPA
network technologies with additional capability to support bandwidth wider than [6].
2.1.1 Overview of LTE downlink physical layer
LTE physical layer is designed by considering the requirements for high data rate, spectral
efficiency, and multiple channel bandwidths. These requirements can be fulfilled by using
orthogonal frequency division multiplex (OFDM).
OFDM is a technology that was published in 1960’s. In the 1980’s, OFDM has been studied
for high speed modems. It was considered for 3G systems in the mid-1990s before being
determined too immature. Developments in electronics and signal processing since that time
has made OFDM a mature technology widely used in other access systems like 802.11 (Wi-
Fi) and 802.16 (WiMAX) and broadcast [7].
In addition to OFDM, MIMO (multiple input multiple output) is another basis technology that
helps meeting the requirement of an LTE. It can either increase channel capacity (spatial
multiplexing) or enhance signal robustness (space frequency/time coding).
2.1.2 Requirements of LTE
Higher User throughput of 3-4 times that of HSDPA (Rel.6) in downlink and 3-4
times that of enhanced Uplink (Rel.6).
Mobility optimized for low mobile speed from 0 to , also supports higher
mobile speeds.
Optimal cell size of 5 km, 30 km sizes with reasonable performance and up to 100 km
cell sizes supported with acceptable performance.
Supports a large number of users per cell of at least 200 users/cell
Spectrum flexibility with 1.4, 3, 5, 10 ,15 and Bandwidth
Co-existence with legacy standards
6
2.1.3 Multiple Access Method in LTE
OFDM is a multicarrier transport technology for high data rate communication system. The
main concept in OFDM is spreading high speed data over the low rate carriers. These low rate
carriers are called sub-carriers. Sub-carriers are generated using IFFT (Inverse Fast Fourier
Transform) digital signal processing. IFFT is an efficient scheme for generating orthogonal
subcarriers. These orthogonal subcarriers are mutually orthogonal in frequency domain which
avoids the inter-symbol interference (ISI) as shown in figure 2.1 below. Each sub-carrier has a
bandwidth less than channel coherence bandwidth which is why these subcarriers experience
flat fading.
f
Figure 2.1: OFDM Orthogonal Sub-carriers
Firstly, Bit stream from the encoder are modulated using QPSK or QAM-16 or QAM-64 into
symbols. These symbols are then fed to serial to parallel converter. The parallel data symbols
from serial to parallel converter are then spread over N IFFT orthogonal subcarriers. The
meaning of orthogonal subcarrier is that there is no interference between the subcarriers.
Again the outputs from the IFFT which are modulated subcarriers are converted back to the
serial form with the help of Parallel to serial converter. This again passed through the digital
to analog converter before transmitting to the air as shown in figure 2.2. In the receiver the
reverse operation is done to the OFDM symbol to retrieve the data stream.
IFFTSerial to parallel
Parallel to Serial
Digital To Analoge
Convertor
Modulator
Bit stream from encoder
N complex Symbols
Figure 2.2: Simple OFDM System
2.1.4 Cyclic-Prefix insertion
One of the drawbacks of OFDM signal (Block of transmitted carriers) is that when it is
propagating through the channel i.e. free air, it convolves with channel impulse response in
7
time domain, and after convolving with channel the length of convoluted signal increase by
factor of (L-1), where L is the length of channel impulse response. Due to this increase of
length (L-1), symbols will overlap with next consecutive symbol and is called ISI. Reason for
adding CP is illustrated graphically in the figure 2.3.
X[2]X[0] X[1] X[M-1]. . . X[M]
l+1 symbol
X[2] X[M-1]. . . X[M]X[0] X[1]
l symbol
X[M-1]. . . X[M]Symbol Over lap
(L-1)
l symbol[M+L-1]
X[M-1]. . . X[M]Symbol Over lap
(L-1)Symbol Over lap
(L-1)
l+1 symbol[M+L-1]
A
B
X[M-1]. . . X[M]. . . X[M]X[M-L+1] X[0]
l symbol
X[M-1]. . . X[M]. . . X[M]X[M-L+1] X[0] . . . X[M]X[M-L+1]
C
CP (L-1) CP (L-1) CP (L-1)
Discard CP Discard CP DiscardX[n] C[n]X[n] C[n]
D
Figure 2.3: Graphical representation of Cyclic Prefix (CP)
In above figure 2.3(A), any two transmitted symbol I and I+1 of length M are considered,
whereas in figure 2.3(B), the channel impulse response (CIR) of length L is considered when
it convolved with two symbols its length become M+L-1 instead of M. So the length L-1 of
symbol I overlap the next I+1 symbol and over write Data there. To overcome this difficulties
CP is appended to the front OFDM symbol and which is the copy of last (L-1) subcarriers of
that symbol. After insertion of CP to all OFDM symbol and then convolving with channel
impulse response the received symbol get affected only in the CP part as shown in figure
2.3(C). This way actual symbol can be kept safe and reduce the ISI. After receiving the
symbol we discard the CP part in the receiver as shown in figure 2.3(D).
The disadvantage of CP insertion is that it does not carry any new data information, which
decreases the transmitted energy per information bit, so there is power loss and reduced the
SNR [8], second there is reduction in symbol rate which is the loss in terms of bandwidth and
can be reduced by decreasing the inter carrier spacing [9], [10].
In LTE physical layer, Orthogonal Frequency Division Multiple Access (OFDMA) scheme is
used for downlink transmission. OFDMA make use of OFDM technology to multiplex
different users by assigning specific patterns of subcarriers in time and frequency domain. The
scheduling and assignment of resources makes OFDMA different from OFDM. In the OFDM
figure 2.4 shows that the entire bandwidth is assigned to single user for a certain period of
time but in OFDMA multiple users are sharing the bandwidth at each point in time [11].
8
Sub
Carr
iers
Time Time
OFDM OFDMA
Sub
Carr
iers
Figure 2.4: OFDM and OFDMA graphical concept
2.1.5 Spectrum flexibility in LTE
One of the main characteristics of LTE radio-access technology is Spectrum flexibility which
allows deployment of LTE radio-access in different frequency bands with different sizes since
spectrum has become a scarce resource. This flexibility includes 2 main areas as follows:
Flexibility in duplex arrangements
Bandwidth flexibility
In Flexibility in duplex arrangements, the communication can take place both in paired and
unpaired bands. In former case the uplink and downlink transmissions use separate frequency
bands, while in later case downlink and uplink share the same frequency band. LTE support
both Frequency-Division Duplex (FDD) and Time-Division Duplex (TDD).
In FDD uplink and downlink transmissions take place in different and sufficiently separated
frequency bands, whereas in TDD uplink and downlink operate in different non-overlapping
time slots as shown in figure 2.5. In our thesis we have considered only FDD.
Time (t)Time (t)
Frequ
ency (f)
Frequ
ency (f)
F-DL
F-ULF-DL+ F-UL
FDD TDD
Figure 2.5: FDD and TDD
9
Bandwidth flexibility is an important aspect in the LTE operation. It is due to the fact that
LTE can be operated in different transmission bandwidths for uplink and downlink. The
reason for having different bandwidth is that LTE deployment depends upon the frequency
bands and on the operator. This bandwidth flexibility arise chances of gradual frequency
bands migration from other radio-access technologies.
2.1.6 LTE Generic Frame Structure
OFDMA is the best radio access for LTE downlink comparing to others. However, resource
allocation is complicated in OFDMA. It is far better than packet oriented approaches in the
context of efficiency and latency. In LTE, users are assigned a certain number of subcarriers
for predetermined period of time; these subcarriers are called physical resource blocks
(PRBs). These blocks have both time and frequency dimension [12].
Figure 2.6: LTE time domain frame structure.
As can be seen in the figure 2.6, LTE frame is of 10 millisecond ( ) duration which is
divided into 10 sub-frames of 1msec duration. Each sub-frame is further divided into two slots
each has a length of 0.5 . Each slot comprises either 6 or 7 OFDM symbols depending
upon normal or extended CP is used.
Table 2.1: LTE downlink Bandwidths
Bandwidth (MHz)
1.4
3
5.0
10.0
15.0
20.0
Subcarrier bandwidth (kHz)
15
Physical resource block (PRB)
bandwidth (kHz)
180
Number of available PRBs
6
12
25
50
75
100
10
Table 2.1 shows LTE specification defines different channel bandwidths from 1.4 to 20 .
Different channel bandwidth has different number of available PRBs. A PRB is comprises of
12 consecutive subcarriers for one slot as shown in figure 2.7 below. A PRB is the
smallest unit that base station can allocate to the user. Subcarrier bandwidth and PRBs
bandwidths are 15 and 180 respectively for all system bandwidths.
The figure 2.7 is shown for the case of normal CP. The transmitted downlink signal consists
of subcarriers for the duration of OFDM symbols.
1 Frame (10 msec)
1 Sub-Frame (1.0 msec)1 Slot (0.5 msec)
12
su
bca
rrie
rs
N s
ub
carr
iers
RESOURCE BLOCK7 symbols x 12 subcarriers (normal cp), or6 symbols x 12 subcarriers (extended cp)
Resource element
Time domain symbol
Fre
qu
en
cy d
om
ain
su
b-c
arr
iers
Figure 2.7: LTE time frequency resource grid
A Resource element is the smallest physical resource and it consists of one subcarrier during
one OFDM symbol. A group of resource elements are referred as physical resource blocks i.e.
PRBs. A PRB has a seven OFDM symbols having duration of one time slot 0.5ms and 12
subcarriers having a bandwidth of ( , so each resource
block in the case of normal cyclic prefix has of 84 resource elements (
) whereas in the case of extended cyclic prefix
resource block has 72 resource elements (
.
11
LTE downlink channel and signals 2.2
2.2.1 Downlink Reference Signals
Packet oriented networks use PHY preamble to facilitate carrier offset estimate, channel
estimation, timing synchronization etc. However, in LTE special reference signals are mapped
into the PRBs to allow coherent demodulation at the user equipment. Downlink reference
Signals are placed in the first and third last OFDM symbol in each slot. It means for the case
of normal CP reference signals are depicted in first and fifth OFDM symbol and of the
extended CP reference signals are depicted in first and fourth OFDM symbol. Frequency
spacing between two reference signals is six subcarriers. Therefore there are four reference
signals in one PRB. These reference signals are used to estimate the channel response on
subcarriers bearing them. Interpolation is employed to estimate the channel response on other
subcarriers [12]. Figure 2.8 given below illustrated are for single antenna LTE system with
normal CP.
0
1
2
3
4
6
7
8
9
10
1 2 3 4 5 6 0 1 2 3 4 5 6
R
R
R
R
R
R
R
R
Slot 0Subframe
12 S
ubca
rrie
rs
Slot 1
Figure 2.8: LTE CRSs Position in a single Antenna
In the case of 2x2 MIMO operations, the receiver must determine channel impulse response
from each transmitting antenna. For the accurate channel estimation, when one antenna is
sending the reference signal other antenna does not transmit anything at the same time and
frequency position of reference signals. Mapping of reference signals for 2x2 MIMO
operations is shown in figure 2.9.
12
Reference signal From Antenna 0
Reference signal From Antenna 1
Unused Resource element
Su
bca
rrie
rs(f
req
ue
ncy
)
Antenna 0
OFDM Symbols (time)
Antenna 1
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
R1
R0
R0
R0
R0
R0
R0
R0
R0
R1
R1
R1 R1
R1
R1
R1
R0
R1
X
Figure 2.9: LTE CRSs mapping for 2x2MIMO
In our thesis we have generated the reference signals as per the technique described in 3GPP
TS 36.211 version 10.0.0 Release 10 [13] and mapping to resource grid is done as described
above. We have used cell ID 300 for Macro cell and 3 for the Femto cell so that CRS of
Macro and Femto collide. CRS collide with each other when criteria given by equation (2.1) is
fulfilled otherwise they don’t collide.
( ) (
) (2.1)
Where, and
is Macro and Femto cell ID.
2.2.2 Downlink Physical Channels
There are mainly three different types of physical channels are defined for LTE downlink.
Physical channels are used for conveying information from higher layers to the LTE stack.
However, Physical signals like reference signal and synchronization signal are used just in the
physical layer. LTE downlink physical channels are
Physical downlink shared channel (PDSCH)
Physical downlink control channel (PDCCH)
Common control physical channel (CCPCH)
PDSCH is used for transportation of data and multimedia. Modulation schemes for PDSCH
include QPSK, QAM-16 and QAM-64 as they are designed for high data rates.
PDCCH is generally used for transportation of UE specific control information. Reliability
and robustness are the important aspect for PDCCH so only QPSK is available as a
modulation scheme.
CCPCH is used for carrying cell wide control information. As PDCCH, reliability and
robustness are the key aspect for CCPCH also, therefore only QPSK is available as a
modulation scheme. They are transmitted on the 72 active subcarriers centered on the DC
subcarrier.
13
All the down link physical channels and signals and their specific mapping in LTE resource
grid for bandwidth 1.4MHz and 20MHz is shown in Appendix B.1 and B.2 respectively.
LTE-A Release 10 2.3
The LTE release 10 was introduced in order to fulfill the IMT (International Mobile
Telecommunication) -Advanced requirements; some of these are listed in table 2.2. The ‘‘IMT
is a global broadband multimedia international mobile telecommunication system that the ITU
(International Mobile Telecommunication) has been coordinating along with governments,
industry and private sector. IMT-Advanced is the term that ITU uses to describe radio-access
technologies beyondIMT-2000’’ [14].
Table 2.2: IMT advance requirements
Channel bandwidth Supporting up to
Peak spectral
efficiencies
15 bit/s/Hz in down link and corresponds to peak rate of
6.75 bit/s/Hz in uplink and corresponds to peak rate of 270Mbits/s
Control plane latency less than
User plane latency less than
The LTE release 10, radio-access technology is fully compliant with the IMT-advanced
requirements and the main reason for LTE release 10 to be called LTE-Advanced.
The Carrier aggregation, extended multi–antenna transmission, almost blank sub-frame
(ABS), relaying and Heterogeneous deployments are some of the important enhancements and
features introduced in LTE Advance. Table 2.3 shows the feature and characteristics of LTE
and LTE-A for comparison purposes.
15
Chapter 3
Heterogeneous network
Introduction to heterogeneous network 3.1
Recently the number of mobile broadband users has been grown exponentially and it will
definitely grow more in the future. This is due to the fact that evolution on terminal
capabilities which can handle new and different kind of services. It is predicted that traffic per
year is going to be double in next five years, therefore by 2014 average traffic uses 1GB of
data per month compared to 100 or 200 MB now [16]
Scarcity of frequency spectrum is always a big issue. Spectral efficiency of wireless networks
is reaching its theoretical limits, it is necessary to increase the node density which can
improve the network capacity. If macro cell deployments are sparse then adding another cell
does not create significant inter-cell interference and cell splitting gain can be achieved.
However, if macro cells are already dense, cell splitting gains are significantly reduced
because of inter-cell interference. Instead of deploying macro cell another low power nodes
can be deployed. These low power nodes are called Pico cell and Femto cell. Those network
which consist of macro and low power nodes are called heterogeneous network [17]. In
heterogeneous network Pico/Femto cell with same set of frequencies are deployed within the
Macro cell in an unplanned way depending upon the areas that has more traffic. In one hand
these deployment increases the overall network capacity, data rate and the performance of the
cell-edge user and on the other hand such deployments cause severe Macro-Pico/Femto inter-
cell interference. Figure 3.1 shows the typical scenario of heterogeneous networks, where in
each cell there is one Macro base station, one Femto base station and one user equipment i.e.
UE.
Femto Base Station
Macro Base Station
UE
Figure 3.1: Heterogeneous Networks
In this thesis Femto cell and Pico cell are considered as same. Therefore we have taken Femto
cell that represents low power node (LPN). We have also considered the scenario where
Femto cell is deployed inside macro cell. In 3GPP LTE term base station is referred as
16
evolved node base station also abbreviated as eNBs for macro base station and HeNBs for
Femto base station.
3.1.1 Advantages of deployment of HeNBs
They can be deployed to eliminate coverage holes
Offer high data rate and capacity where they are deployed
less costly and easy to deploy comparing to macro cell
do not require air conditioning unit in the power amplifier
Cell splitting gain can be achieved
Traffic offloading to the low power nodes can be done
3.1.2 Properties of Macro eNBs and LPN
Macro eNBs
Power range around +45dBm
Covers outdoor with around cell site ~5km
Operator deployed
Femto eNBs
Power range around +15dBm
Covers indoor residence 50m
Operator deployed or User deployed
Equipped with Omnidirectional antenna
Backhaul uses the existing internet connection such as DSL
Pico Cell [17]
Power range around 23.97dBm for outdoor and 20dBm
Pico cell are regular macro cell except for low transmitting power
Equipped with Omnidirectional antenna
Backhaul uses the X2 interface for data communication and interference management
Relay [17]
Low power nodes (LPN) used in HetNets to enhance coverage.
No wired Backhaul ,wireless backhaul by relaying the signal from mobile station to
Macro eNB
In-band relay node uses the same UL and DL frequency where in out-of-band uses
different frequencies in UL and DL.
UE measurements in LTE 3.2
Reference signal received power (RSRP) is defined as the as the linear average over the power
contributions ( of the resource elements that carry cell-specific reference signals within
the considered measurement frequency bandwidth .For RSRP determination the cell-specific
reference signals (from first antenna) according to TS 36.211 [18] be used. If the UE can
reliably detect that (from second antenna) is available it may use in addition to to
determine RSRP [19].
17
Reference signal received quality (RSRQ) is defined as the ratio N×RSRP / (E-UTRA carrier
RSSI), where N is the number of RB’s of the E-UTRA carrier RSSI measurement bandwidth.
The measurements in the numerator and denominator shall be made over the same set of
resource blocks. E-UTRA Carrier Received Signal Strength Indicator (RSSI), comprises the
linear average of the total received power ( observed only in OFDM symbols containing
reference symbols for antenna port ‘0’, in the measurement bandwidth, over N number of
resource blocks by the UE from all sources, including co-channel serving and non-serving
cells, adjacent channel interference, thermal noise etc. [20].
Introduction to range extension 3.3
Cell selection in LTE is done based upon UE measurements of RSRP. In traditional Macro
cell networks the eNB having the high RSRP is selected [21]. However, in a heterogeneous
network there are base stations transmitting with different power level. So it is unfair to the
base station transmitting with lower power because terminal always chooses higher power
base station. This problem can be minimized if cell selection is done based upon uplink path
loss, which in practice is done by applying cell specific offset of the RSRP and RSRQ. This
method of cell selection increases the coverage area of the low power base station and this
increased area is named as range extension as shown in figure 3.2.
Range extension region
Figure 3.2: Range extension in Heterogeneous networks
Problem caused by deployment of Femto eNBs 3.4
As it is mentioned above that Femto eNBs which are also called HeNBs can be operator
deployed or user deployed. If it is operator deployed Macro user also have access to it which
is called open access but if it is user deployed Macro user cannot get connected to it only
limited number of users like family members of that particular home will have access to it
which is called closed subscriber group (CSG).
In this situation when Macro UE (user equipment) is in edge of Femto cell there will be strong
interference from Femto cell and may even not be able to access the Macro cell at all. In
another situation strong Macro cell interferes the Femto UE. This happens when sometime it
is preferable to connect UE to the Femto cell even if the received power from the Femto cell is
weaker than Macro cell. This is useful when strong cell has weak backhaul quality or when it
is necessary to offload the traffic to Femto cell and to achieve true cell-splitting gains in the
network [22].
Let’s take an example when UE is in cell extension region and it is connected to Femto base
station. But received power by the UE from the Macro is much higher than the power received
18
from Femto. This Power difference cause severe interference to the Femto user by Macro base
station. This scenario can be demonstrated by the figure 3.3.
Range extension region
UE1
UE2
Figure 3.3: Illustration of CSG and range extension user
As we can see UE2 is in well inside the Femto cell which is also called center Femto user.
Here, the received power of the Femto is stronger than the received power of Macro so there
is no issue of interference. But if we see the UE2 it is in the cell extension region and it is
connected to the Femto base station. Received Power by this user from the Macro is much
higher than power received from Femto. So, UE2 undergoes severe interference imposes by
Macro base station.
So this is one of the main disadvantages of heterogeneous network. Our thesis deals with the
minimization of this interference. We have assumed the heterogeneous network of Macro and
Femto base station. And there is an UE in the cell extension region which is interfered by the
Macro base station.
3.4.1 Evolution of eICIC
In order to reduce Inter-cell interference in the LTE, the Release 8 of the 3GGP
standardization body proposed the ICIC (Inter-cell Interference Coordination) feature. It is
one of the new RRM (Radio Resource management) functions. The task is to determine the
resources (frequency time and power) available in each cell and schedule those resources to
the users. The ICIC, as described in [23], is located in the eNB and it has the task of managing
the PRBs such that inter-cell interference is kept under control. The coordination between
neighboring eNBs is based on the exchange of interference information. The main problem
with ICIC is that it is able to mitigate the interference between data channels but control
channels are still under interference.
In LTE-A enhanced Inter-Cell Interference Coordination (eICIC) is introduced to address this
problem. They proposed two methods one is frequency domain multiplexing inter-cell
interference coordination scheme and other is Time domain multiplexing inter-cell
interference coordination scheme (Almost Blank Sub-frame).
3.4.2 Carrier aggregation in LTE
Carrier aggregation is used in Frequency domain multiplexing inter-cell interference
coordination scheme which is one of the most important features of LTE-A. This scheme
allows LTE-A user to connect to several carriers at the same time. It also allows resource
19
allocation across the carriers as well as fast switching between carriers without time
consuming handovers, which means a node, can schedule its control and data information on
separate carriers. Main principle in a HetNet scenario is to partition the available spectrum
into two separate component carriers and assign the primary component carriers (PCC) to
different network layers as shown in figure 3.4 below.
control
data
f1
f2
f1
f2
Macro
Femto
Figure 3.4: Illustration of Carrier aggregation in LTE
Primary component carriers is the cell that provides the control information to the UEs, by
assigning this different frequencies interference on control channels between network layer
can be avoided. By means of so called cross carrier scheduling each network layer can still
schedule UEs on the other CCs called secondary component carriers (SCC) [24]
In the figure 3.4, f1 and f2 are PCC and SCC respectively where 5 sub-frames are shown in
each carrier. These carrier’s components are used by both Macro layer and Femto layer. The
sub-frames contain PDCCH, PCFICH and PHICH in the beginning the blue part which is
control signal and green part is data signal.
As shown in figure 3.4 Macro layer can schedule its control signal on f1 but can still schedule
its users on both f1 and f2 so by scheduling control and data information for both Macro and
Femto layers on different component carriers, interference on control and data can be avoided.
As can be seen in the sub-frame 3 of figure 3.4, Macro layer has scheduled the user in the
same frequency as Femto layer scheduling its user, this can be done when Femto user is center
Femto user as the interference from Macro layer on center Femto user can be tolerated. If
Femto user is in range extension region, Femto users are scheduled in different carrier then in
Macro users. The disadvantage of carrier aggregation with cross carrier scheduling is that it is
not backward compatible to release 8 and 9 terminals. It supports by release 10 terminals and
onwards [14].
3.4.3 Almost Blank Sub-frame
The almost blank sub-frame is a sub-frame with reduced downlink transmission power and
activity. In this, transmission from Macro eNB causing high interference onto Femto eNB
users is periodically muted during entire sub-frames. Thus muted sub-frames are referred to as
almost blank sub-frame. Using this approach, Femto cell users that are suffering from a high
level of interference from the aggressor Macro eNB have a chance to be served. However, for
the backward compatibility or to avoid radio link failure reasons certain signals needs to be
transmitted in ABSs.
20
Such signals are:
Cell specific reference signals (CRS) which are also called pilots
SIB-1 and paging with their associated PDCCH
Physical broadcast channel (PBCH)
Primary and secondary synchronization channels (PSS/SSS)
Collisions of sub-frame muting with PSS, SSS, SIB-1 and paging should be minimized. PSS,
SSS, SIB-1 and paging are transmitted in sub-frames #0, #1, #5 and #9.Without carefully
handling of these channels in the ABSs significant performance degradation can occur in
some situations.
ABSs muting patterns are configured semi-statically and signaled between eNBs over the X2
interface. These patterns are signaled in the form of bitmaps of length 40, i.e. spanning over 4
frames for FDD and 2 to 7 for TDD. Hence they can be configured dynamically by the
network.
controldata
Macro
Femto
ABS ABS
Rec
eiv
ed S
INR
fem
to u
ser
Time
Figure 3.5: Illustration of ABS in LTE
As shown in figure 3.5, Usage of ABS causes different level of interference between the sub-
frames of Macro and Femto. If Femto user is in range extension region and it is interfered by
strong Macro then Femto user can be served during the time when Macro transmit ABS else
Femto user can be served during ABS or non-ABS.
Colliding and Non colliding CRS 3.5
As it is explained earlier for the sake of backward compatibility and to avoid radio link failure
signals like CRSs, PCH, PBCH and PSS/SSS are still needs to be transmitted in the ABSs.
For simple study and investigation on the interference issue, we have considered a Macro cell
transmitting ABS with only CRS and a Femto is transmitting a sub-frame with PDSCH
channel and CRS. Since, the CRSs are spread all over the sub-frame in both in time and
frequency domain there is a chances that CRS of Femto sub-frame and CRS of ABS collide,
which is one of the worst scenarios encountered in the heterogeneous environment.
21
This situation is called colliding CRS case as shown in figure 3.6(B). As CRSs are primarily
used for channel estimation, if they collide, Channel estimation quality won’t be good and
system performance will degrade. The effect of interference from PCH, PBCH and PSS/SSS
in ABS can still be controlled by not assigning the sub-frames where those channel appears as
ABS, and also by using sub-frame shift between the Macro and Femto [1], which is sometime
referred as non-colliding case shown in figure 3.6(A).
(A) Non Colliding Case (B) Colliding Case
Figure 3.6: Illustration of Colliding and Non Colliding CRS
22
Chapter 4
Channel Model
Introduction to Air Interface 4.1
It is very important to have a good channel model for simulations as close to the reality as
possible. Correct channel models are also essential for testing, parameter optimization,
performance evaluation of communication systems and deployment of communication system
for reliable transfer of information between transmitter and receiver. On the basis of time,
channel may be time-invariant i.e. channel is constant over time or time-variant channel i.e.
channel varies over time. The time-invariant channels are relatively easy to estimate since it
only requires one estimate at the beginning of the reception of the signal whereas time-variant
channel is much harder to estimate since it requires continuous estimation.
The major complication in modeling of a wireless channel is due to complex propagation
processes. Multiple copies of transmitted signal received from transmitter through different
propagation mechanisms shown in figure 4.1.
Figure 4.1: Multipath effect [25]
The multipath propagation occurs due to the following reasons [24], [26].
Reflection, when electromagnetic waves get contact with flat and even surfaces then
they are reflected
Diffraction, caused by bending of electromagnetic waves around corners of buildings
and other objects.
Scattering, when electromagnetic waves get contact with the objects having uneven or
irregular surface then they are get scattered.
Energy of the wave is partially absorbed when it passes through some special objects.
If channel is varying due to relative motion of large objects to that of radio device causes
often slow fading, on the other hand multipath components of the signal that mix up either
23
constructively or destructively causes fast fading. As a result of the terminal movements on
the order of half a wavelength a constructive superposition becomes destructive and vice
versa.
As shown from Figure 4.1 a receiver, receives multiple copies via different paths from
transmitter i.e. a multipath channel and is called as frequency selective channel hence, the
channel varies in frequency. On the other hand, when receiver receives only one copy of the
transmitted signal which results in a flat frequency fading, which in practice not the case
often. Instead the radio signal is reflected by objects before it reaches the receiver [8] and can
be modeled in time domain as
(4.1)
Where, and are stationary, statistically independent and real-valued Gaussian
processes and are typical not white processes, but instead colored Gaussian.
4.1.1 Rayleigh distributed channel
The central limits theorem will model the channel as a Rayleigh distributed channel, when
there are enough different paths, hence, and are Gaussian with zero-mean in
equation (4.1). The Equation (4.1) in polar form is given below
(4.2)
Where
√
(4.3)
And
(4.4)
This distribution represents the sum of amplitudes of the same order of a large number of
uncorrelated components with phases uniformly distributed in the interval (0, 2π). The
power delay profile (PDF) of Rayleigh distributed channel is described by equation below
(4.5)
24
Where, [ ] [
] and for < 0. The parameter is the root mean
square value of the received signal.
4.1.2 Rician distributed channel
In the presence of line of sight (LOS) component, the channel can be modeled as a Rician
distribution and is given by equation (4.6) as in [27].
(
)
(4.6)
Where modified Bassel function of first kind and Zero order and is the amplitude of the
line of sight component (LOS). In the absence of LOS component, Rician distribution can be
simply expressed as Rayleigh distribution. Since in this thesis MIMO is implemented and it is
assumed that there is no line of sight between the transmitter antennas and the receiver
antenna, so Rician distributed channel model is not our interest here.
4.1.3 Maximum Doppler shift
In statistical analysis of an extended Clarke’s model, the time varying nature of the wireless
channel is described by the wiener processes and where fixed Doppler shift through the multi-
path signal is introduced to describe the situation of a moving receiver and transmitter. A
closed-form expression for the autocorrelation function of the fading channel is derived which
contain a zero order Bessel function and decaying factor. The Bessel is the result of the
Doppler shift introduced due to the moving receiver and transmitter [28]. The maximum
Doppler shift is given by
(4.7)
Where is the Doppler shift, is the carrier frequency, is the relative velocity and is the
propagation speed of signal (speed of light) The above equation describes fast and slow fading
in batter way, since the receiver and transmitter moving relative to each other is the main
cause of fading.
Jake’s Channel model 4.2
In typical radio mobile communication scenario the transmitter is generally fixed whereas the
receiver is moving and reflection through objects produces a multi-path signal effect. The
direct LOS signal is not available in many cases and hence only signal reflected through
objects reach the moving receiver. The total received signal at moving receiver is well defined
by Jake’s model [29]. Equation (4.8) shows the impulse response of the channel;
∑ (4.8)
25
Where, is the delay of path and is the corresponding complex amplitude. Jakes
channel model does not take into consideration of the changing of delay paths M and the
delays over time. All is a complex Gaussian zero mean and can be defined as a
function of an angle and the distance between the receiver and the transmitter as
follows
∑ (4.9)
Zero order Bessel function, which depends upon the time difference and Doppler
shift, is the autocorrelation of the different transmission paths . The factor determines
the variation of the channel between successive symbols.
The way how the power is spread over the time is difficult to estimate for any channel model.
However, most simple type of power delay distribution is uniformly and exponentially. In
uniform distribution, power of the channel response is uniformly distributed over a certain
period of time, whereas exponential power delay distribution which is defined in [30] has
power distributed according to equation (4.10).
(4.10)
Where is an arbitrary constant and is the root mean-square delay spread of the channel.
The delays are all uniformly distributed over the delay time of the channel. Taking this in
to account the channel can now be defined as
∑ √
(4.11)
In the OFDM case different sub-carriers has different frequencies. Therefore it is important to
have time varying transfer function of the channel.
∫
∑ √ (4.12)
Above expression is the time varying transfer function of the channel which assumes that an
exponential power delay profile is used with set so that the average power is normalized to
one.
In this thesis, Jake’s channel model with an exponential power delay Profile is used to
generate time and frequency-variant channels as described in equation (4.12).
26
Chapter 5
Channel Estimation
Channel estimation 5.1
Channel estimation is an important part of receiver section. Without channel estimation the
received data in any receiver cannot be decoded correctly while propagated through time
varying environment (channel) or in other hand receiver have to use differential phase-shift
keying (DPSK), in the latter case signal-to-noise ratio (SNR) of 3-dB is less compared with
coherent phase-shift keying (PSK) [31]. For time varying channel, receiver must have a
continuous channel estimation and tracking. If it is not continuous then receiver performance
will degrade drastically.
In this thesis report two type of channel estimation i.e. LS and LMMSE are studied and
implemented. As in LTE, the CRSs are scattered in time frequency resource grid as describes
in LTE technical specification, 3GPP TS 36.211 [18] as shown in figure 2.8 and figure 2.9
for SISO and 2x2 MIMO respectively. Due to the scattered nature of CRSs in time frequency
resource grid, first channel is estimated at CRSs position using the received CRS and
Transmitted CRS by least squared estimate. While using frequency interpolation over the
estimate of CRS position, for the channel estimate of OFDM symbol having the CRS (OFDM
symbols #1, 4, 8 & 12 for 2x2 MIMO configurations). Finally channel is estimated over
OFDM symbol having no CRSs (OFDM symbols # 2, 3, 5, 6, 7, 9, 10, 11, 13, 14 for 2x2
MIMO configuration) using time interpolation on OFDM symbols having CRS.
Signal Model 5.2
In simulation work we use multi sub-frame to analyses the interference mitigation via
advanced interference cancelation schemes in LTE Down link system model. One LTE sub-
frame consists of 14 OFDM symbols as sown in figure 2.8. But for channel estimation of LTE
downlink we considered one OFDM symbol from single eNB in frequency domain and is
given by equation (5.1).
(5.1)
Where, is a frequency response vector of unknown channel to be estimated and having
the dimension of . is the diagonal matrix of transmitted data sample including
CRSs and Zeros and have the dimension of and is the AWGN noise.
The equation (5.1) in terms of channel impulse response (CIR) is given by equation (5.2) [32].
(5.2)
27
Where, and is the DFT matrix and have the dimension of . In
mathematical form it can be expressed as
[
]
(5.3)
As discussed above the channel estimation primarily used CRSs for the channel estimation
and these CRSs are time frequency distributed. So in order to avoid the complexity in the
estimators we considered some simplification in signal model.
In frequency selective channel by so many path transmitted signal reached to receiver and
referred to as channel taps but every channel taps have not sufficient energy for consideration
of channel estimation, here we considered that channel taps having enough energy. If L
represents the length of the channel then we take first L columns equation (5.3) of the matrix
F corresponding to the maximum delay at taps L-1.
As in equation (5.2) Matrix X, represents data, CRS and Zeros but now onward for channel
estimation we considered which represents only the rows CRS position and have the
dimension of and similarly Matrix F represent only the rows which correspond
to these CRS position. Now equation (5.2) can be reform as:
(5.4)
Where the output vector at CRS position and AWGN vector, having dimension of
. The vector and represent Fourier matrix associated with Transmitted
CRSs and unknown CIR coefficients to be estimated at CRSs position respectively and have
the dimension of . The matrix can be written as fallow:
√
[
]
(5.5)
Where referred to as the starting index of CRS position and represents
with the
point of DFT.
28
Equation (5.2) can also be written in similar fashion to that of equation (5.4) for the non CRSs
position means for the data position and is given by
(5.6)
And,
√
[
]
(5.7)
SINR 5.3
The SINR is the ratio of signal to interference plus noise. In heterogeneous network the
serving BTs is most often interfered by the non-serving BTs. There interference may be intra-
interference or inter-interference or both. So the SINR in heterogeneous network for one
Macro cell and one Femto cell and one can be defined as
(5.8)
Where, is the power of the total received signal and is the received power of interferer
and is Nose power. The SINR is mainly used in LMMSE estimation and which play an
important role in the IC algorithm.
Least Square Estimator 5.4
The least square estimator uses no information from channel i.e. channel statistics and is the
simplest of all the estimators in uses for channel estimation. The LS estimator uses received
CRS and known transmitted CRS for estimating the channel estimation at the CRS position in
the following way:
[
] (5.9)
29
Where is the estimated channel frequency response of the sub-carrier at the CRSs
position in OFDM symbol. The frequency and time interpolation is done according to section
5.1 to fully estimate the channel frequency response at the data position in OFDM symbols.
Due to low complexity the LS estimate is used widely but it has inherent disadvantage that it
has high mean square error. The LS channel estimation given in equation (5.9) is based on
minimization of the squared error without noise as described in [33], [34].
| | | |
| | (5.10)
After some mathematical calculation and by taking derivatives of squared error we get LS
estimation given bellow:
( )
( ) (5.11)
(5.12)
Where k = 1, 2, 3. . . N
Least Minimum Mean Square Error (LMMSE) Estimator 5.5
Least minimum mean squared error estimator is another type of 1D estimator, which gives the
estimate of channel impulse response (CIR) or channel frequency response (CFR) at the CRS
position in OFDM symbol. The LMMSE has advantage over the LS estimator that it gives
least mean square error means better performance as compared to LS, but the LMMSE
estimator has higher degree of computational complexity and also requires second order
channel statistics. The LMMSE performance criteria are based on the minimization of the
mean square error between the actual and estimated CIRs. These CFRs can be found while
using the LMMSE as describe in [32]. Now error is given by;
(5.12)
And
{ | | } { | | } { } (5.13)
Where in equation (5.13) { } is the expectation and channel AWGN noise assumed to be
uncorrelated, after some mathematical calculation we get
30
(5.14)
Where is the auto covariance matrix of and
is the cross covariance
matrix of the vector H and .
[
]
[ ]
[ (
)]
[ (
)]
[ ]
[
] (5.15)
In equation (5.15) (AWGN) noise and Channel frequency response H are uncorrelated so
their expectation value is Zero.
(5.16)
In equation (5.16) is the auto covariance matrix of
[
] (5.17)
[ ]
[ (
)]
[ ]
[
]
[ ] [
]
31
(5.18)
In equation (5.18) is the noise variance and is given by equation
{|
| } (5.19)
By using equations (5.11), (5.16) and (5.18) we can write the equation (5.14) for LMMSE as
below:
(
)
(5.20)
The above equation of justify the fact of computational complexity due to the
presence of term ( )
which is changes every time for the news input (next OFDM
symbol ) of in the case of CRS. This complexity in equation (5.20) can be reduced by
replacing complex matrix inversion by the expectation {( )
} . Then the above
equation (5.20) can be reformed as
(5.21)
Where is the modulation index and depends upon the constellation point which is
consequently directly depends on type of modulation scheme and code rate used. The value
for QPSK and QAM-16 are 1 and 17/9 respectively for the un-coded system which is
calculated as equation (5.22).
{ | | } { |
| } (5.22)
For the coded system the value depend upon the different code rate and modulation scheme
as given in table 5.1 [35].
32
Table 5.1: value for different modulation scheme and Code rate [35]
And , the average SNR is given by;
{ | | }
(5.23)
The factor in equation (5.21) has still worse effect in channel estimation of at CRSs
position because this factor is the cross correlation of CRSs and the whole channel in
frequency domain. This worse effect of can minimised by separating the whole
channel matrix into CRSs channel vector and Non CRSs (Data) channel vector as given
bellow:
33
[
] (5.24)
Where are the CFRs at data position but not of CRSs position. By using equation (5.24)
in equation (5.21) estimation and interpolation can be split into two separate expressions as
given below:
(5.25)
(5.26)
Interpolation 5.6
In order to estimate the channel over data is necessary for achieving accuracy in demodulation
and good performance at receiver. In LTE downlink CRSs are time frequency distributed as
shown in figure 2.8 and figure 2.9. First frequency interpolation is needed in order to
optimally estimate the whole OFDM symbol containing CRSs and after that time interpolation
is done for other data OFDM downlink symbols on the basis of frequency interpolated OFDM
symbols.
5.6.1 Frequency Interpolation
There are different techniques for frequency interpolation in OFDM system as described in
[36]. In this thesis we used frequency interpolation for the OFDM symbols containing CRSs
and interpolates as whole symbol in frequency domain for the estimation CFRs at data
positions while using the estimated CFRSs of the equation (5.25) and is given by the equation
(
)
(5.27)
Instead of equation (5.27), equation (5.26) can also be used for the frequency interpolation.
The main difference between equation (5.26) and (5.27) is that the earlier equation estimate
the CFRs at the data position in OFDM symbols only whereas the later one estimated the
whole symbol.
5.6.2 Time Interpolation
As mention in section (5.1) the CFRS of the resource elements in OFDM symbols having no
CRSs can be determined by the time interpolation. On the bases of CFRS of the resource
elements in estimated OFDM symbols that are interpolated by frequency interpolation and
CFRs, the CFRs of the resource elements in non CRSs symbols are interpolated in time
domain by simple linear time interpolation to avoid the complexity. For time interpolation it is
assumed that OFDM symbols with in a sub-frame have strong correlation of the channel
frequency response. On the bases of previous and next OFDM symbol i.e. frequency
34
interpolated, the resource element of in the middle OFDM symbol can be interpolated as
follow [8].
(5.28)
Where and are the channel estimate of the previous and next
resource elements on the same carrier of frequency interpolated OFDM symbols (CRSs
OFDM symbols) respectively. Whereas and are the location or position of the
previous and next resource element on the same carrier from the respectively. As this
interpolation is based on the channel variation, which is directly related the channel coherence
time. In this thesis it assumed that the channel is constant within a resource block as mention
equation (6.26). Second in our simulation the channel model based on Jack’s channel model
criteria is not varying too much and the coherence time chosen is 0.01sec. Based on this
coherence time 138 OFDM symbol can be transmitted without degradation (One OFDM
symbol time is since one sub-frame time is 1 ).
Equalization 5.7
As the transmitted data passed through the time varying channel it convolve with the CIRs of
the channel due to which the received sub-carrier in OFDM data symbol suffer from distortion
in amplitudes and phase shift. So in order to compensate this distortion in amplitudes and
phase shift due to the multiplicative inverse of the complex channel equalization must be
performed. Two types of equalizations are described in [25] that are:
Time domain equalization
Frequency domain equalization
CP RemovalSub-CarrierDe-Mapping
N-Point DFT
Time Domain EqualizationFrequency Domain
Equalization
. . .
. . .
Figure 5.1: LTE time domain and frequency domain equalizer option
Time domain equalization is done on data symbols in time domain in order to equalize it with
the transmitted data symbol as show in figure 5.1. The time domain equalization is used in
OFDM receiver where the performances of the other equalizer are not important as in time
varying propagation condition.
Frequency domain equalization try to reduce the error between receive sub-carrier of OFDM
data symbol and transmitted OFDM data symbol in frequency domain. In frequency domain
equalization, received time domain OFDM data symbols are first transformed into frequency
35
domain while using N- point DFT and then equalization is done by frequency domain
filtering.
In simulation we performed the frequency domain equalization and the frequency domain
system model of equation (5.6) can be written as for the data sub-carrier as
(5.29)
Where is the resource element of the . The equalized output can be
writing as below
‖ ‖
(5.30)
36
Chapter 6
System Model
Introduction to System Model 6.1
Since we have considered Heterogeneous Network with Macro and Femto cell and Femto cell
UE in the cell extension region, we have modeled two transmitter base station one for Macro
as interferer cell and Femto as serving cell and one receivers.
We have mentioned in the methodology that Macro transmits ABS with CRS. For the
simplicity, we have shown general transmitter model for both Macro and Femto. It starts with
the generation of data bits. Now onwards we have described the model with block by block
basis. Close pictorial image of the system model is shown in figure 6.1.
Channel estimation
Tx CRS
Receiver
Bit genertaion CRC
Convolencoding
Ant 1
Ant 2
Ant 1
CRS genertaion
Modulation(QPSK,QAM)
Femto Transmitter(Serving HeNB)
STBC
Ant 2
Maping
Ant 1
Ant 2
FFT
Ant 1
Ant 2
CP Insertion
Ant 1
Ant 2
PA
Channel (Air interface)
Bit genertaion
CRCConvolencodin
g
Ant 1
Ant 2
Ant 1
CRS genertaion
Modulation(QPSK,QAM)
STBC
Ant 2
Maping
Ant 1
Ant 2
FFT
Ant 1
Ant 2
CP Insertion
Ant 1
Ant 2
PA
Macro Transmitter(Interfering eNB)
Ant 1
Ant 2
STBC
Extraction
Ant 1
Ant 2
IFFT
Ant 1
Ant 2
CP Removal
Ant 1
Ant 2
AWGN
RX Data CRC Check
Viterbi
DeModulation
(QPSK,QAM)Rx Data
Rx CRS
Equilization
Femto Ant 1
Femto Ant 2
Macro Ant 1
Macro Ant 2
RX Ant 1
RX Ant 2
Encoder
Encoder
Decoder
Figure 6.1: System model
Encoder 6.2
At first, 24 bits CRC is appended at the end of the generated bit, the output is called ‘code
word’. The code word is then passed through the convolutional encoder of code rate 1/3.
37
6.2.1 Cyclic redundancy check (CRC)
The cyclic redundancy check (CRC) is a technique which is mostly used in digital data
transmission for detecting errors. The CRC can’t be used for corrections of detected error bits.
In CRC technique, a certain number of parity bits, which are generated with some cyclic
polynomial generator, often named as checksum, are appended to the transmitted information
(message). After receiving the message along with CRC checksum, the receiver check it
whether it is agree with the data or not. If agree message go forward otherwise, the receiver
sends a “negative acknowledgement” (NAK) back to the transmitter for requesting that the
message send back again [27], [37]. In this thesis, 24 bit cyclic generator polynomial as
specified in “3GPP TS 36.212 [18] multiplexing and channel coding, release 10” is used.
6.2.2 Convolutional encoder
Convolutional codes are generally represented by three parameters i.e. n, k and m.
The quantity k/n is called code rate, which measures the efficiency of the code.
Manufacturers’ often use parameters and to represent the convolutional codes. Where,
is the constraint length of the code, which defines number of bits in the memory of the
encoder that effect the generation of n output bits. And can be written as,
(6.1)
However, in commercial specification codes are represented by where, represents
code rate and is the constraint length.
Encoding is done to improve the channel capacity; it is done by adding some redundant to the
message bits. Another type of coding is block coding. The Convolutional coding can be
applied to serial data, one or a few bits at a time whereas, Block coding can be applied to large
data up to several hundreds of bytes or blocks. Convolutional code has the ability to detect
and correct error bits up to some extent and is called as error detection and correction code
[27], [38].
38
u u u-11 01
v
v
1
2v
3
(1,1,1)
(0,1,1)(1,0,1)
u
Figure 6.2: Code rate 1/3, Convolutional encoder
Figure 6.2 shown above is convolutional encoder, where each input bit is coded to 3 output
bits. So this is a convolutional encoder of code rate 1/3. The constraint length of the code is 2.
3 output bits are produced by 3 modulo-2 adders by adding specific bits from the memory
registers. These specific bits are generated by the so called generator polynomial ‘g’.
In our thesis, we have used generator polynomial as specified in “3GPP TS 36.212 [18]
multiplexing and channel coding, release 10”. The first, second and third output bit has a
generator polynomial of (1, 1, 1), (0, 1, 1) and (1, 0, 1) respectively. The output is sum of
these bits.
(6.2)
(6.3)
(6.4)
Figure 6.3 illustrate the 7dB coding gain achieved when convolutional encoder is
implemented in the single system without interference for QAM-16. Red curve with circle is
un-coded BER and blue curve with asterisk coded BER of the system.
Figure 6.3: Convolutional encoding gain
-5 0 5 10 15 2010
-5
10-4
10-3
10-2
10-1
100 QAM16, Jakes Channel Model
Serving cell SNR(dB)
Bit E
rro
r P
rob
ab
ility
Uncoded
coded
39
Digital modulation 6.3
The encoded sequence of bits is mapped for either QPSK or QAM scheme. In our thesis, we
have simulated the system with QPSK, QAM-16 and also in some case with QAM-64, short
description is given below:
6.3.1 QPSK
One of the general types of modulation scheme is QPSK. They are used in different channel
access technologies like OFDMA, CDMA in wireless telecommunication along with in
Iridium (a voice/data satellite system) and DVB-S (digital video broadcasting-satellite).
Quadrature means the signal shifts from 45 to 135 or -45 or -135 degrees. These points can be
easily implemented in and modulator. There are states in QPSK because there are
only two values and two values are required as shown in figure 6.4.
Figure 6.4: QPSK state diagram
6.3.2 QAM
Another type of digital modulation is Quadrature Amplitude Modulation (QAM). They are
also mostly used in OFDMA, CDMA in wireless telecommunication along with in DVB-S
(digital video broadcasting-satellite), and modems. In QAM-16, there are 16 possible states of
a signal. Therefore, there are four values and four values. This modulation
scheme is more spectral efficient than QPSK. Similarly, QAM-64 has 64 states.
Therefore, there are six values and six values. Figure 6.5 illustrates state diagram QAM.
B)64QAMA)16QAM
Figure 6.5: QAM scheme diagram. A) QAM-16 B) QAM-64
40
MIMO- STBC 6.4
Recently, study and research in MIMO technologies is growing, as it is one of the promising
techniques for high data rate transmission. Space Time Block Coding (STBC) is one of the
MIMO technologies which provide diversity gain at the receiver by transmitting space coded
signal through multiple antennas. As we have discussed earlier OFDM is a technology that is
famous for high data rate transmission and its robustness towards inter-symbol interference.
So, STBC-OFDM can be one of the best system configurations for generation mobile
system. STBC applies coding across the number of OFDM symbols equivalent to number of
transmit antennas [36] . There are mainly three kinds of STBC schemes; they are Alamouti’s,
Tarokh’s and quasi-orthogonal schemes for OFDM system.
In our work we have implemented 2x2 MIMO Alamouti’s scheme for channel coding. So we
have just described about the Alamouti’s scheme below.
Alamouti’s scheme is specifically designed for two antennas transmit diversity case and the
scheme work over two symbol periods and channel gain is assumed to be constant over this
period. Over first symbol period two different symbols and each with energy of are
transmitted from both antenna 1 and 2 simultaneously and over next coming symbol period,
symbols and
(each with energy of ) are transmitted from antenna 1 and antenna 2
respectively and simultaneously. Coding matrix for Alamouti’s scheme is given below by
equation (6.5).
S*1
S2
- S*2
S1
Transmitting Antenna
Time
Slot
X
(6.5)
Where, * represent complex conjugate.
Mapping of signal to the grid 6.5
The CRSs is generated depending upon its position and cell ID. The CRSs are the complex
valued numbers. These CRSs are generated by the product of two dimensional pseudo random
sequences and two dimensional orthogonal sequences. In LTE specification there are 510
different cell IDs and their corresponding CRSs sequences [18]. Generation and mapping of
CRS for two antenna resource grid is described in chapter one.
Once CRS are mapped to the grid, STBC encoded data is mapped, which in our case is
PDSCH to the rest of the grid. In the system we have used one whole sub-frame. As we have
mentioned before that sub-frame contain other physical channel and signals but for the
simplicity we have just used CRS as physical signal and PDSCH as physical channel. After
mapping we transmit the each LTE symbol to IDFT.
DFT/IDFT 6.6
DFT can be viewed as a linear input/output processor. Let’s, take an example that we have an
L input symbols. And let’s take the DFT size of length N. If N is less than L, then we choose
N samples out of the L and perform DFT calculation on these N points. In another case, If N
41
is larger than L then we add some zeroes to input symbols so that its length is equal to N. DFT
operation is shown in figure 6.6.
Discrete Fourier Transform(DFT)
x[0]
x[1]
x[N-1]
X[0]
X[1]
X[N-1]
Figure 6.6: DFT Block diagram
The mathematical expression for DFT is;
[ ] ∑
0≤n≤N-1 (6.6)
Where, is the transmitted symbol in subcarrier and X[n] is an OFDM symbol, N is
the total number of subcarriers, n is discrete time domain, is frequency of subcarrier
and is equal to
.
Inverse Discrete Fourier
Transform(IDFT)
X[0]
X[1]
X[N-1]
x[0]
x[1]
x[N-1]
Figure 6.7: IDFT Block diagram
IDFT operation is shown in above figure 6.7 and its mathematical expression is;
[ ]
∑
0≤n≤N-1 (6.7)
The only difference between DFT and IDFT is that IDFT has a scaling and change of the sign
of the exponent. DFT is implemented in receiver whereas IDFT is implemented in transmitter.
42
Cyclic Prefix (CP) 6.7
After IFFT, CP is added to the signal. Detail description of CP is done in section 2.1.4
Power Amplifier 6.8
The power amplifier is used to boost up the radio frequency to achieve desired power level for
the transmission on air interface. In our case, there is Heterogeneous Network i.e. Macro-eNB
and Femto-HeNB, both operates at different power level.
A baseband equivalent nonlinearity characterized by AM/AM (amplitude to amplitude) and
AM/PM (amplitude to phase) mapping is illustrated in figure 6.8 where the notation and
are actually functions operating on amplitude of the base band signal [39].
Figure 6.8: AM/AM & AM/PM mapping
Where,
(6.8)
( ) (6.9)
In our simulation, we have used Rapp Power Amplifier Model which is also named as the
Solid state power amplifiers (SSPA). This is more linear to small signal as compared to the
Traveling Wave Tube Amplifiers (TWTA) and clipping for large signals. The characteristics
of the Rapp model is given by equation (6.10) [39].
⌊ (
)
⌋ (6.10)
(6.11)
Where and are the mapping from the amplitude and phase of the base band
input to amplitude and phase of the base band output respectively, k is the small gain, is
𝑔𝐴 𝑔 𝐺 𝑥 𝑡
𝑥 𝑡 𝑦 𝑡
43
the level of limiting amplitude and p is the smoothness factor of the transition region to
limiting amplitude. The and are often known as AM/AM (amplitude to
amplitude) and AM/PM (amplitude to phase) mapping. The is assumed to be zero for
SSPA [39].
In below figures 6.9 and 6.10 Rapp model is investigated first by varying smoothness factor
‘ ’ keeping ‘ ’ and constant, and then by varying keeping and constant.
First In the figure 6.9, Rapp model is investigated by varying smoothness factor ‘ ’ from 1, 2
and 3keeping 500 and constant. Similarly, in the second figure 6.10, Rapp model
is investigated by varying saturation level keeping and p=1.
Figure 6.9: RAPP Amplifier smoothness factor is varying
Figure 6.10: RAPP Amplifier, Saturation level varying
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
0.8
1
1.2
1.4
1.6
1.8
2RAPP PA when smoothness factor is varying
Input Amplitude
Out
pu
t A
mp
litud
e
p=1
p=2
p=3
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5RAPP PA when saturation level is varying
Input Amplitude
Outp
ut
Am
plitu
de
sat=1
sat=1.25
sat=1.5
44
Air Interface 6.9
The transmitted data from each antenna of Macro eNB and Femto eNB after amplifying by
the RAPP amplifier is propagating through the air interface and get affected by the
characteristic of wireless channel. The characteristics of wireless channel depend upon the
distance between TX and RX antennas, the path or paths taken by the TX signal, the
environment around the taken path (buildings and other objects) and the relative motion of the
transmitter and receiver antenna and object between it. The transmitted signals from Macro
and Femto eNBs added up in the air interface before receipt by the receiver as both the signal
have same carrier frequency. In air interface the signal from eNB and HeNB is convolved
with the channel impulse response in time domain which is equivalent to multiplication
infrequency domain. In our simulation work we studied and implemented the Jack’s channel
model in order to realize the real time channel impulse response as discussed in section 4.2.
The combined signal received at UE after channel convolving is given by equation (6.12) in
frequency domain.
(6.12)
Where, Y is the received signal of the mth OFDM symbol at the kth subcarrier and the
superscript F and M represent the Femto and Macro eNB respectively. Where H and X are the
channel impulse response and the transmitted signal correspond to the mth OFDM symbol and
kth subcarrier respectively. Z is the additive noise of AWGN.
Receiver 6.10
The receiver is more commonly reverse of the transmitter. In the receiver, first of all cyclic
prefix is removed which is added at the end of the transmitter. The CP removed time signal is
then changed to frequency domain signal by taking the Fast Fourier Transform (FFT). In
frequency domain noisy modulated data and CRSs are extracted from the frequency time
domain grids of each antenna from the same position where data and CRSs were mapped to
antenna grid.
In our case i.e. colliding case where the Femto eNB and Macro eNB has the same CRSs
mapping position in time frequency resource grid or they have to fulfill the below equation
(6.13) and graphically is shown in figure 3.6(B) of section 3.5.
( )
(6.13)
Where and
represent the Femto Cell identity and Macro Cell Identity, respectively.
So, the received extracted noisy CRSs are the sum of Femto CRSs, Macro CRSs and AWGN
noise and these received CRSs are used for channel estimation of and
. Due to using
of 2x2MIMO there are 2 extracted received copies for each CRS from each antenna as shown
in figure 6.11 below. Where, ,
and ,
are the CRS of Macro and Femto at
sub-carrier k and k+1 respectively and and are the received CRS at sub-carrier k and
k+1 respectively. Where represent the channel impulse response for the specific
receive and transmit antenna.
45
K+1
K+1
Macro eNB
Femto eNB
Ant 2
Ant 1
Ant 1
Ant 2
UE
Ant 1
Ant 2
h12
h21
0XM
XF
0
Y =k
Y =K+1
h11 h11+
X0
F
K
XF
K
X0
MK
X MK
XM
K+1h12 XF
K+1h12
h21+ XF
KX MK
XM
K+1h22 XF
K+1h22
h21
Air Interface
Figure 6.11: 2x2 MIMO Configuration over air interface
Core methodology 6.11
Now we have combined received CRSs from both Macro and Femto eNBs and furthermore
after cell search the UE also have information about transmitted CRSs positions, reference
signal received power (RSRP) and Cell IDs of both the Macro and Femto eNBs. Using this
information serving cell (Femto) data can be correctly got after using advanced inter-cell
interference mitigation schemes.
Our main purpose in this thesis is to mitigate the interference in CRS, as CRS is mainly
related to the channel estimation and their collision drastically degrades the performance of
the receiver. So in order to minimize the effect of interference on data, ABSs are transmitted
from the interferer cell (Macro eNB) according to Rel. 10 eICIC as described in [1].
In this work RSRP based CRS interference cancellation scheme is studied. The RSRP of
Femto and Macro cell is represented by and , respectively. Based on the actual RSRP of
CRS, interference cancellation scheme can be divided into following three cases;
Direct IC :
Joint channel Detection : | |
No IC :
By combining the above three cases it can write as and named as
Combined IC (‘Com IC’).
The flow chart for implementation of above mentioned IC algorithm is shown in figure 6.12.
Since the IC algorithm is implemented in receiver we have shown the receiver IC
implementation in the flow chart. Flow chart starts with Macro cell is transmitting ABS
having CRS and Femto is transmitting sub-frame having data signal and CRS. RE Femto UE
receives signal from Macro and Femto at the same time as we have assumed that Macro and
Femto sub-frames are synchronized. Receiver first calculates the RSRP of both Macro and
Femto cell. Depending upon the RSRP difference level either Direct IC or Joint channel
detection or No IC method is selected. Detail description of Direct IC, Joint channel detection
and No IC is described in 6.11.1, 6.11.2 and 6.11.3 respectively. From these methods we get
estimated received Femto CRS signal, which is further utilized to get the Femto CRS channel
estimation. And this CRS channel estimation is used for the interpolation to get channel
estimates of all Femto data signals. Finally, Equalization is done over Femto data signals and
desired Femto signal is received.
46
Femto eNB Macro eNB
Air Interface
Start
𝑃𝑓 𝑃𝑚 𝑑𝐵
|𝑃𝑚 𝑃𝑓| 𝑑𝐵 𝑃𝑚 𝑃𝑓 𝑑𝐵
RSRP Calculation of
Femto & Macro
Direct IC Joint Detection
No IC
End
Equalization
Femto eNB
Signal
Detection
Figure 6.12: Flow chart of IC Algorithm
47
6.11.1 Direct IC
The main theme of the direct IC scheme is to estimate the Macro received CRSs using Macro
channel estimation and subtracting it from the total received signal i.e. Femto plus Macro and
Noise and then using the subtracted signal for Femto channel estimation. As UE is closer to
the Macro eNB, then will be fulfilled and in such situation Femto eNB
signal can be merged with AWGN noise as Noise i.e.
is considered as
Noise in equation (6.12) Macro channel can be estimated using Least Square estimate as
derived in equation (5.12).
(
)
(6.14)
Putting the value of in equation (6.14) from equation (6.12), while the sub-carrier index k
is absent for simplicity in calculation
{
}(
)
(6.15)
( )
( )
(6.16)
Least minimum mean square error (LMMSE) estimator can be applied to LS estimate to get
rid-off unwanted noise. Using equation (5.21) Macro CRSs channel can be estimated. After
applying LMMSE estimate of Macro CRSs channel over Transmitted Macro CRSs, the
received estimated Macro CRSs can be obtained as given by equation (6.18).
(
)
(6.17)
Where,
(
)
(
) (6.18)
By subtracting the received Macro estimate of CRSs ((
)
) from equation (6.12),
we left with Femto noisy signal.
(
)
(6.19)
(
)
(6.20)
48
Equation (6.20) shows that when the received Macro estimate is good then noisy effects from
the Interferer will ultimately less and in a consequence Femto desired signal estimate will
healthy. Using this Femto noisy signal estimate for Femto channel estimation in LS and
LMMSE estimators, and after equalization of the received data and demodulation serving cell
(Femto cell) data can get with less error probability.
6.11.2 Joint channel Detection
Joint channel detection means that estimate channel of both Macro and Femto eNBs as this
case related to the RSRP level difference of Macro and Femto | | , which
means that UE receives approximately same power level from both the Macro and Femto
eNBs. In this scenario no one signal can be taken as Noise, but both the signal are important
and channel of both Macro and Femto are estimated at the same time and hence called Joint
channel detection algorithm.
For 2x2 MIMO system configuration of figure 6.11, the received output for one carrier
frequency of both Macro and Femto in the presence of AWGN noise can be expressed as.
(6.21)
(6.22)
The received signal of equation (6.21) & (6.22) from antenna 1 and 2 in matrix notation
can be written as below
[
] [
] [
]
[
] [
] [
] (6.23)
In equation (6.23) the antenna numbers are represented by subscript 1 and 2 and NRX and
NTX notation is used for channel representation of receive and transmit antenna. In LTE
CRSs are not transmitted at the same time from both antennas of the same eNB. So when
antenna 1 of Femto transmit CRSs the other antenna of Femto will not transmit at time and
hence
||
(6.24)
By applying above condition, when only transmitter antennas 1 of both eNBs are in
operational mode then equation (6.23) can be modeled as
49
[
] [
] [
] [
] (6.25)
In equation (6.25) there are four unknown i.e. two Femto channel impulse responses
(
) and two Macro channel impulse response (
). From above
equation it is not possible to determine the values of four unknown parameters while using
single CRS symbol. So in order to take the advantage of time invariant characteristics of the
channel, as in LTE with in a resource block (RB) channel remain constant as far as the
Doppler spread remains small. So,
(6.26)
While following the above condition, for 2x2 MIMO system configurations with in a RB
symbol 1 and symbol 4 have the CRS as shown in a figure 2.8 . The equation (6.25) can be
reformed as below;
[
] [
] [
] [
] (6.27)
By taking the pseudo inverse of
Matrix, channel impulse response of
,
can be calculated as:
[
] ([
] [
]) [
]
(6.28)
By using similar calculation, remain channel impulses responses of Femto
and
Macro
can be calculated as fallow:
[
] ([
] [
]) [
]
(6.29)
Using equation (6.28) and (6.29) Macro and Femto cell channel at CRS position are
calculated. The pseudo inverse in equation (6.28) and (6.29) is actually least square estimate
at CRS positions. LMMSE can be applied to the Joint detection LS estimate for better
accuracy and performance by removing unwanted noise as appears in LS estimate equations.
Equations (6.17) to (6.20) are used for the interference cancelation while using LS estimate of
50
Macro channel estimate in order to minimize the effect of Macro interference on Femto
signal.
6.11.3 No IC
In No IC case the Femto eNB power is greater than Macro eNB i.e. . In this
case either the UE is much closer to the HeNB or Macro eNB is too far away from HeNB and
the RSRP level difference is to enough that UE can correctly estimate the Femto channel and
demodulate the Femto received data with accuracy.
51
Chapter 7
Simulation and Results
Simulation set up 7.1
In our simulation we have used 2x2 MIMO antenna systems with STBC. We have also used
convolutional encoder of code rate 1/3 and constraint length of 7 and 24 bit CRC. LTE
channel bandwidth is The Channel model we have used is Jakes channel model. All
the parameters used in simulation are shown in the Table 7.1 below.
Table 7.1: Parameter used in Simulation
Transmission Bandwidth 20 MHZ
Sub-frame duration 1ms
Sub carrier spacing 15 kHz
Sampling frequency 30.72 MHz
FFT size 2048
Number of occupied subcarrier 1200
Number of OFDM symbol per subframe (short CP) 14
Number of OFDM symbol per slot (short CP) 7
CP length 5 µs
Number of channel taps 10
τ (rms delay spread) 25ns
Maximum doppler shift 20 Hz
rms
Modulation QAM, QPSK
Range of SNR -5 to 20 dB
Number of simulation cycle per SNR 20
Range of SNR -5 to 20 dB
Code rate (convolutional encoder) 1/3
Number of CRC bit 24
In the evaluation, two eNBs are assumed one is Macro as dominant interference cell and other
is Femto as a serving cell. In addition, we have assumed that transmission time between
Macro and Femto is aligned i.e. Both are synchronized in time domain.
We have also assumed the case when Macro is transmitting the ABS. We have taken the CRS
as a physical signal that is transmitted in the ABS. Femto cell is transmitting PDSCH as data
signal and CRS as a physical signal, CRS of both Macro and Femto cell are mapped at the
52
same position and they are collided at the receiver. We can see from the diagram 7.1 below
that interference exist just only in the CRS of Macro and Femto cell.
Macro eNB(Almost Blank Subframe)
Femto eNB
One subframe
Interfering CRS
Serving CRS
PDSCH (data signal)
Time allignment between macro and femto
Time
CRS Symbols
Figure 7.1: CSR, ABS and PDSCH configuration in LTE
We have considered that the UE is at the range extension region trying to get access from
Femto HeNB but Macro eNB is imposing high interference to it, as can be illustrated from the
figure 7.2 below.
Range extension region
UE1
Figure 7.2: UE receiving signal from Femto and Macro eNB
7.1.1 NO IC Scheme Based results
In this section we have discussed the impact of power variation of Macro and Femto eNBs on
range extension Femto UE.
Firstly, Macro power and noise power is set at constant level but Femto power is varied. Then
we have analyzed the quality of desired received signal with a graph as shown in figure 7.3.
Furthermore, the case is studied for different value of SNR to see the noise effect.
53
Figure 7.3: BER when Femto RSRP is varying for QPSK
Macro power is set at 1dB and Femto power is varied from 1 to 16 dB for each curve. And the
case is studied for SNR value of 0, 5, 10 and 20 for red, blue, green and black curve
respectively. From figure 7.3, it is concluded that when the Femto power is increased from the
same level of Macro power there is no significant interference from Macro cell. Signals can be
easily detected for higher value of SNR. In another word, received signal quality gets better
for low power of noise. We observed the similar results for the case when signal is modulated
with QAM-16 as shown in figure 7.4. The only difference is that the converging rate of the
curve is slower as compared to the curve of figure 7.3.
Figure 7.4: BER when Femto RSRP is varying for QAM-16
Now it is interesting to see what results we get when Femto power is kept constant and Macro
power is varying under constant serving cell SNR.
-4 -2 0 2 4 6 8 10 12 1410
-5
10-4
10-3
10-2
10-1
100 Femto Power increasing,QPSK, Jakes Channel Model
Femto RSRP/Macro RSRP (dB)
Bit E
rro
r P
rob
ab
ility
SNR=0
SNR=5
SNR=10
SNR=20
-2 0 2 4 6 8 10 1210
-5
10-4
10-3
10-2
10-1
Femto Power increasing,QAM-16, Jakes Channel Model
Femto RSRP/Macro RSRP (dB)
Bit E
rro
r P
rob
ab
ility
SNR=0
SNR=5
SNR=10
SNR=20
54
Femto power is set at 1dB and Macro power is varied from 1 to 16 dB for each curve. And the
case is studied for SNR value of 0, 5, 10 and 20 for red, blue, green and black curve
respectively. Then we have studied the BER performance against ratio of Femto RSRP by
Macro RSRP under different value of serving cell SNR as show for QPSK in figure 7.5.
Figure 7.5: BER when Macro RSRP is varying for QPSK
It is concluded from the figure 7.5 that when the Macro power is increasing from the same
level of Femto power for each curve, BER is also increasing which means interference
imposes by Macro on Femto keeps increasing and system performance gets worst. This case is
also studied for QAM-16 as can be seen in the figure 7.6 below. We observed similar results
like as for QPSK. But the only difference is the initial BER for QPSK is less and for QAM-16
it is high.
Figure 7.6: BER when Macro RSRP is varying for QAM-16
-18-16-14-12-10-8-6-4-210
-4
10-3
10-2
10-1
100 Macro Power increasing,QPSK, Jakes Channel Model
Femto RSRP/Macro RSRP (dB)
Bit
Err
or
Pro
bab
ility
SNR=0
SNR=5
SNR=10
SNR=20
-18-16-14-12-10-8-6-4-2
10-0.6
10-0.5
10-0.4
Macro Power increasing,QAM-16, Jakes Channel Model
Femto RSRP/Macro RSRP (dB)
Bit
Err
or
Pro
bab
ility
SNR=0
SNR=5
SNR=10
SNR=20
55
Finally, in No IC scheme Femto power and Macro power is kept constant at different RSRP
level and serving cell SNR is varying from -5dB to 20dB. Then we have studied the BER
performance against ratio of Femto RSRP by serving cell SNR and results for QPSK and
QAM are shown in figure 7.7 and figure 7.8.
Figure 7.7: BER when SNR is varying for QPSK
From the above figure 7.7 we examined that interference exist when the power level of Macro
RSRP is greater or equals to 3 dB with respect to Femto RSRP. This is due to the fact that
stronger the Macro power, channel estimation quality degrade more. As mentioned before
channel estimation in LTE is based upon CRS signals. And RSRP is the linear average power
of the received CRS. The nonlinear characteristics from SNR 3 dB to 6 dB in the blue or
curve with asterisk can be removed if we send more number of sub-frames; in this graph we
have transmitted just 20 sub-frames. We have transmitted 1500 sub-frames for block error rate
calculation and there are no such non-linearities as you can see the block error rate in figure
7.11 and 7.12.
In the case of QAM-16 the results are worst as QAM symbol have more bits per symbol and
results are shown in figure 7.8 below.
56
Figure 7.8: BER when SNR is varying for QAM-16
7.1.2 Conclusion based on No IC scheme
The conclusions from the results of figures 7.3-7.8 which are based on No IC scheme; When
Femto power is increased from the same level of Macro power, the interference is decreasing
meaning that system is performing well. When Macro power is increased from the same level
of Femto power, the interference is also increasing. Consequently the system performance
degrades due to collision of CRSs as channel estimation depends on the CRS signals. If we
don’t use any IC scheme to coup this severs interference from the Macro eNB, in other words
we have to find some technique to increase channel estimation quality.
7.1.3 IC Scheme based Results
So we have used RSRP based interference cancellation method as mentioned in methodology
section 6.11. Here, Femto data bits are QPSK modulated. Macro RSRP is set 16dB and Femto
RSRP is set 4dB maintaining 12dB power difference. Serving cell SNR is varied from -5 to
20dB as shown in the figure 7.9 below.
-5 0 5 10 15 2010
-2
10-1
100 QAM-16, Jakes Channel Model
Serving Cell SNR(dB)
Bit
Err
or
Pro
bab
ility
Macro RSRP-Femto RSRP<3dB
Macro RSRP-Femto RSRP=3dB
Macro RSRP-Femto RSRP>3dB
57
Figure 7.9: BER when SNR is varying for QPSK using IC scheme
Three BER curves against the serving cell SNR are plotted. Double dashed or blue curve is for
the case when no Interference cancellation (IC) technique is applied. For no IC, Bit error rate
is not converging means system performance is worst. This is because Macro cell power is 12
dB higher than Femto cell power which causes high interference toward the Femto CRSs. As
it is also illustrated in the figure 7.7 and 7.8 that with just 3dB power difference interference
occurs. This is the reason why we need some receiver IC algorithm to detect the signal
properly. Solid or red curve represents when receiver IC scheme is implemented. Dotted
dashed or green curve is for the case when interfering Macro is absent or for single Femto
cell. Single Femto cell without interference from Macro is used as reference for the
comparison purpose here. The performance of the system when IC is applied is almost close
to the performance of the reference single Femto cell. So, high gain can be achieved in terms
of SNR when IC is implemented in receiver.
System performance for QAM is shown in the figure 7.10 below for the same scenario as
mentioned above for QPSK.
-5 0 5 10 15 2010
-3
10-2
10-1
100
QPSK, Jakes Channel Model
Serving Cell SNR(dB)
Bit E
rro
r P
rob
ab
ility
IC
No-IC
Single-Cell(Femto)
58
Figure 7.10: BER when SNR is varying for QAM using IC scheme
From figure 7.10 it is shown when the data bits are modulated with QAM-16 and QAM 64. In
QAM-16 and QAM-64, for blue curve or no IC receiver BER is not converging at all. IC
receiver has BER at for 13dB SNR for QAM-16 whereas for QAM-64 BER is
converging close to for 19dB SNR. This is due to the fact that QAM-16 and QAM-64
has 4 bits and 6 bits in a symbol respectively which increases the probability of bit error
detection.
We have also analyzed system performance with block error probability (BLER) against
Serving cell SNR. For the block error rate calculation we have transmitted 1500 sub-frames.
Generally in OFDM system it is more reasonable to check the system performance with Block
error probability.
Figure 7.11: Block error rate for QPSK, 1500 sub-frames
-5 0 5 10 15 20 2510
-3
10-2
10-1
100
IC,QPSK, Jakes Channel Model
Serving cell SNR(dB)
Blo
ck E
rror
Pro
ba
bilit
y
IC
NO IC
Single cell (femto)
59
Here for the BLER the Macro RSRP is 9dB and Femto RSRP is 3dB. QPSK and QAM-16 are
used as modulation scheme. So it can be observed that for IC scheme block error probability
is decreasing as SNR is increasing whereas for no IC scheme Block error rate is constant for
all SNR. Moreover UE without interferer has BLER converging at lower SNR. Conclusion is
that UE with IC scheme can decode the signal at low SNR than UE without IC scheme.
Figure 7.12: Block error rate for QAM-16, 1500 Sub-frames
7.1.4 Conclusion based on IC scheme
In short, when IC is implemented in the receiver the performance of the system is much better
than the performance of the system without IC. Block error rate and Bit error rate of the
received signal for QPSK and QAM-16 indicate that interference is significantly reduced. In
the above figures 7.9 to 7.12, no IC curve is going almost straight so it is difficult to say
exactly how much gain in terms of SNR is achieved with respect to IC curve. But it is certain
that Interference is minimized up to large extent. Moreover, desired signal quality is as close
to received signal quality of system without interference.
-5 0 5 10 15 20 2510
-3
10-2
10-1
100
IC,QAM-16, Jakes Channel Model
Serving cell SNR(dB)
Blo
ck E
rror
Pro
ba
bilit
y
IC
NO IC
Single cell (femto)
60
Chapter 8
Conclusion and Future work
Conclusion 8.1
In this thesis, we have analyzed Femto cell UE in the Heterogeneous network of LTE, which
is in the cell extension region trying to get access from Femto cell. Since in LTE
heterogeneous network Macro and Femto cell operates with same set of frequencies therefore
there is always possibility of inter-cell interference especially when received signal power
from interfering cell is high. As we have mentioned in problem definition that with the usage
of ABS interference towards the data signal is eliminated however there is a chance that CRSs
collide and create interference. Creating this environment in the Matlab simulation we
implemented comIC algorithm to reduce the impact of colliding CRSs. If we see the IC
schemes based results in 7.1.3 graph shows that interference is minimized and brought it
down to acceptable range. Above all, quality of received signal is way better than the received
signal without IC algorithm.
As mentioned earlier cell ID and RSRP of both Macro and Femto cells are known to UE.
Based upon the investigation on Macro and Femto cell RSRP difference level and Matlab
simulation results proves that when the RSRP of Interfering cell is higher or equal to the
RSRP of serving cell Interference is detected, due to which Femto UE may not access Femto
cell at all. For the case when Macro RSRP is greater than Femto RSRP by 3dB, simulation
result shows that receiver with IC has system performance almost close to the system without
interference. However, for receiver with no IC, Bit error rate is not converging meaning that
system is not performing well.
Furthermore, quantity of noise in the system makes a huge difference in a quality of the
received signal. To see the effect of noise we have simulated the system without IC algorithm
under 5, 10, 15, 20 dB SNR values. We plot the graph of BER versus Femto RSRP/Macro
RSRP in dB. Here the Femto RSRP is desired signal power and Macro RSRP is interfering
signal power. The graph shows received signal quality gets better and better under 5, 10, 15,
20 dB SNR values respectively.
In LTE, PDSCH channel (data and multimedia) is modulated with QPSK, QAM-16 or QAM-
64 as high data rate is expected due to which we have also simulated with QPSK and 16
QAM. Received signal quality is better for QPSK than QAM-16 as QPSK has lower number
of states then QAM. For better detection of signal we have appended 24 bit CRC followed by
convolutional encoder. For a single system without interference convolutional encoder
provides around 7dB gain in terms of SNR, Which is shown in the figure 6.3. As interference
is the cause of power difference between two cells, to vary the signal power we have included
RAPP power amplifier model.
61
Future Work 8.2
We have studied the case when there is only one interfering eNB; it is interesting to
see for interfering cell more than one.
It would be clearer picture of the system if the result is compared with non-colliding
CRS case.
Here we have used LS and LMMSE for channel estimation, Implementation of IC
algorithm with different channel estimation technique can be interesting.
Interesting to see what will happen when two Femto UEs are in cell extension region.
62
References
[1] Beatriz Soret, Yuanye Wang, Klaus I. Pedersen, "CRS Interference Cancellation in
Heterogeneous Networks for LTE-Advanced Downlink," in International Workshop on
Small Cell Wireless Networks, 2012.
[2] Huang, W. X. Ming, "Macro-Femto Inter-Cell Interference Mitigation for 3GPP LTE-A
Downlink," pp. 75-80, 2012.
[3] 3GPP, R1-104036, "Investigation on CRS Interference Transmitted to Downlink control
Channel," NTT DOCOMO, july 2010.
[4] Lopes-Perez D; Chu Xiaoli, "Inter-Cell Interference Coordination for Expanded Region
Picocells in Heterogeneous Networks," pp. 1 - 6, 2011.
[5] Motorola Inc., "Long term evolution(LTE)," Technical white paper, p. 2, 2007.
[6] Ergen, Mustafa, Mobile Broadband Including WiMAX and LTE, Berkeley, CA: 2009.
[7] Telesystem Innovations Inc., "The Physical Layer WHITE PAPER," LTE in a Nutshell,
2010.
[8] Daniel Larsson, "Analysis of channel estimation methods for OFDMA," Master of
Science Thesis , Stockholm, Sweden, 2006-12-19.
[9] Prasad, R. van NeeR, OFDM for Wireless Multimedia Communications, London: Artech
publishers, January 2000.
[10] Erik Dahlman, Stefan Parkvall & Johan Skold, 4G: LTE/LTE-advanced, Academic Press,
2011.
[11] Anritsu Corporation, "LTE resource guide".
[12] Jim Zyren, Dr. Wes McCoy, Technical Editor, "Overview of the Long TermEvolution
Physical Layer," vol. Document Number: 3GPPEVOLUTIONWP, Rev 2007.
[13] "LTE, Evolved Universal Terrestrial Radio Access (E-UTRA),Physical channels and
modulation," vol. 3GPP TS 36.211 version 10.0.0 Release 10, Jan 2011.
[14] Hisham El Shear, "Interference Management in LTE-Advanced Heterogeneous Networks
Using Almost Blank Sub-frame," KTH, 2012.
[15] K. R. Wondmenech A.M, "''Load Balancing in Heterogenous LTE-A Networks'',"
Linkoping University, Sweden , Jun 2012.
[16] "Mobile data traffic surpasses voice," Ericsson, March 2010. [Online]. Available:
http://www.ericsson.com/news/1396928.
[17] Damnjanovic A., Wei, Yongbin, "A SURVEY ON 3GPP HETEROGENEOUS, Wireless
Communications," vol. 18, no. 3, June 2011.
63
[18] LTE,3GPP TS 36.211, " Evolved Universal Terrestrial Radio Access (E-UTRA);
Physical channels and modulation," vol. version 10.0.0 Release 10, 2010.
[19] 3GPP, RP-080671, "Modification to the RSRP definition," in Nokia, Nokia Siemens,
Warsaw, Poland, July 4, 2008.
[20] 3GPP, RP-100205, "Modification of RSRQ definition," in NTT DOCOMO, Valencia,
Spain, January 18 – 22, 2010.
[21] "3GPP, Evolved Universal Terrestrial Radio Access (E-UTRA); User Equipment (UE)
procedures in idle mode," vol. 8.0.0, p. Tech. Spech. 36.304, December 2007.
[22] 3GPP, R1-100701, "Importance of serving cell selection in heterogeneous networks," in
Qualcomm Inc, Jan 2010.
[23] 3GPP, "EUTRA and EUTRAN overall description," Stage 2, TS 36.300, vol. 8.10.0.
[24] Clercks, Claude Oestges and B., "MIMO Wireless Communications:From Real-World
Propagation to Space-Time Code Design," Elsevier, 2007.
[25] Asad Mehmood, Waqas Aslam Cheema, "Channel Estimation for LTE Downlink,"
Blekinge Institute of Technology, Master Thesis Report, Sweden, September 2009.
[26] Rappaport T., Wireless Communications, principles and Practice, NJ, USA: Prentice-
Hall, Englewood Cliffs, 1996.
[27] Goldsmith, Andrea, Wireless Communication, 2005.
[28] Irshad, Yasir, On Some continuous-time modeling and estimation problems for control,
Karlstad, Sweden: Karlstad University, 2013.
[29] Jakes, W.C., Microwave Mobile Communications, New York: Wiley, 1974.
[30] Ove Edfors, Magnus Sandell, Jan-Jaap van de Beek, Sarah Kate Wilson and Per Ola
Börjesson, "OFDM Channel Estimation by Singular Value Decomposition," IEEE
TRANSACTION ON COMMUNICATIONS, Vols. 46, NO 7, July 1998.
[31] Ye (Geoffrey) Li, Leonard J. Cimini and Nelson R. Sollenberger, "Robust Channel
Estimation for OFDM Systems with Rapid Dispersive Fading Channels," IEEE
TRANSACTION ON COMMUNICATIONS, Vols. 46, NO 7, July 1998.
[32] Jan-Jaap van Beek; Ove Edfors; Magnus Sandell; Sarah Kate Wilson; Per Ola Börjesson,
"On Channel Estimation in OFDM Systems," In proceedings of Vehicular Technology
Conference (VTC’95), vol. 2, pp. 815-819, September 1995.
[33] al, Jose Araujo et, "Self Triggered Control over Wireless Sensor and Actuator
Networks," KTH Royal Insitute of Technology, Technical report, Stockholm.
[34] Regmi, Raju, "Effects of Quantization Noise in Telecommunication Systems," Karlstad
University , Karlstad, Sweden, 2012-10-12.
64
[35] P. Mu˜noz, I. de la Bandera, R. Barco, F. Ruiz, M. Toril and S. Luna-Ram´ırez,
"Estimation of link-layer Quality parameter in system level LTE link simulators,"
University of M' alaga, Communication dept., Spain.
[36] Ho-Chul Jung, Chang-Ju Kim, Hyung-Rae Park and Jong-Ho Kim, "A comparative
performance analysis of STBC-OFDM system under Rayleigh fading environments,"
School of Avionics and Telecommunications Engineering, Hankuk Aviation University.
[37] Peterson, W. W. and Brown, D.T, "Cyclic Codes for Error Detection.," In Proceedings of
the IRE, p. 228–235, January 1961.
[38] Costello, S. Lin and D. J., Error Control Coding, Englewood Cliffs, NJ: Prentice Hall,
1982.
[39] Gokceoglu, Ahmet Hasim, "Performance of Adaptive feed forward methods in wideband
power amplifier linearization," Tampere University of Technology, Oct 2009.
65
Appendix A
A.1 LTE resource grid for LTE minimum bandwidth (1.4 MHz)
Figure A.1: LTE Resource grid for 1.4MHz bandwidth