AN ADAPTIVE MODULATION SELECTION POLICY IN ORTHOGONAL FREQUENCY DIVISION MULTIPLEXING BASED WIRELESS SYSTEMS IBRAHIM ISMAIL MOHAMMED AL-KEBSI THESIS SUBMITTED IN FULFILMENT FOR THE DEGREE OF DOCTOR OF PHILOSOPHY FACULTY OF ENGINEERING AND BUILT ENVIRONMENT UNIVERSITI KEBANGSAAN MALAYSIA BANGI 2010
256
Embed
THE APPLICATION OF ORTHOGONAL FREQUENCY DIVISION ... · an adaptive modulation selection policy in orthogonal frequency division multiplexing based wireless systems . ibrahim ismail
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
AN ADAPTIVE MODULATION SELECTION POLICY IN ORTHOGONAL FREQUENCY DIVISION MULTIPLEXING BASED WIRELESS SYSTEMS
IBRAHIM ISMAIL MOHAMMED AL-KEBSI
THESIS SUBMITTED IN FULFILMENT FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
FACULTY OF ENGINEERING AND BUILT ENVIRONMENT UNIVERSITI KEBANGSAAN MALAYSIA
BANGI
2010
POLISI PEMILIHAN PEMODULATAN ADAPTIF DALAM SISTEM TANPA WAYAR BERASASKAN PEMULTIPLEKSAN PEMBAHAGIAN
FREKUENSI ORTOGON
IBRAHIM ISMAIL MOHAMMED AL-KEBSI
TESIS YANG DIKEMUKAKAN UNTUK MEMPEROLEH IJAZAH DOKTOR FALSAFAH
FAKULTI KEJURUTERAAN DAN ALAM BINA UNIVERSITI KEBANGSAAN MALAYSIA
BANGI
2010
ii
DECLARATION
I hereby declare that the work in this thesis is my own except for quotations and
summaries, which have been duly acknowledged.
22 September 2010 IBRAHIM ISMAIL MOHAMMED AL-KEBSI P34762
iii
ACKNOWLEDGMENTS
First, I am grateful to ALLAH s.w.t, the omniscient, for guiding me to conceptualize, develop and complete my PhD thesis. Indeed, without His help and will, nothing is accomplished. I am indebted to my supervisor, Professor Dr. Mahamod Bin Ismail and my co-supervisor, Professor Dr. Kasmiran Bin Jumari, who have initiated ideas, encouraged and given support during the periods of research, simulation, and writing thesis. I wish to extend my warmest thanks to all those who have helped me with my work in the Department of Electrical, Electronic and Systems Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia (UKM). This research work was partially supported by PKT4/2008 research grant from Malaysian Communications and Multimedia Commission (MCMC). I would like to express my sincerest gratitude and appreciation to my family for their endless love and support during my life. Without their moral support, this thesis would never have been completed. I would like to express my gratitude to my scholarship sponsor from the Yemen's ministry of Higher education for financing of my study.
iv
ABSTRACT
Orthogonal Frequency Division Multiplexing (OFDM) is a technology used to compress a large amount of data into a small amount of bandwidth. Among challenges encounter by OFDM system is related to the selection policy of an appropriate modulation scheme for transmission in each subcarrier. In addition, identifying a compromise between Peak-to-Average Power Ratio (PAPR) improvement and reasonable Symbol Error Rate (SER) performance is the most important issue in OFDM system. Although Adaptive Modulation (AM) is a physical layer technique that can adjust modulation based on the channel condition, it is not good solution to enhance the system throughput. At each Signal-to-Noise Ratio (SNR) value, it cannot utilize more than one modulation scheme to map the data onto carriers, and this leads to constant system throughput if the channel condition does not change. This thesis proposes two models of modulation selection policy. The selection policy in the first model is controlled by SER. In this model, all available modulation schemes are utilized to map the input data at low SNRs but with different percentage of utilization based on SER that can be calculated from the last successful OFDM symbol transmission. However, in the second model, the selection policy is controlled by the values of SNR and SER but limited to two modulation schemes only. Furthermore, these two selection policy models are then combined with the clipping technique to produce an algorithm called Adaptive Modulation and Clipping (AMCl). The clipping technique in this algorithm serves to reduce the PAPR and to control the selection policy in each model by defining variant CRs depending on the modulation scheme order, SER, and SNR. AMCl also employs updating mechanism to define the value of CR at each new symbol transmission. AMCl in both models is built using Matlab code. The proposed models are tested and validated in three OFDM wireless systems; IEEE802.11g, 802.16e, and 4G. Simulation results show the ability of the proposed models to improve the PAPR, enhance the data rate at all SNRs, and provide good SER performance. For example, the OFDM-256QAM signal in IEEE802.16e at 10-3
is improved with 6 dB and 3.5 dB over normal system in PAPR and SER respectively. The data rate is enhanced from 7.9 to 38.1 Mbps at SNR of 5 dB.
v
ABSTRAK
Pemultipleksan Pembahagian Frekuensi Ortogon (OFDM) merupakan teknologi yang digunakan untuk memampatkan sejumlah data yang besar dalam lebar jalur yang sempit. Antara cabaran yang dihadapi oleh sistem OFDM adalah berkait dengan polisi pemilihan pemodulatan yang sesuai bagi penghantaran setiap subpembawa. Tambahan, isu terpenting dalam sistem OFDM adalah untuk mengenalpasti tolak ansur antara penambahbaikan Nisbah Kuasa Puncak-ke-Purata (PAPR) dan prestasi nilai Kadar Ralat Simbol (SER) yang sesuai. Walaupun Pemodulatan Adaptif (AM) merupakan teknik lapisan fizikal yang boleh mengubah pemodulatan berasaskan keadaan saluran, teknik ini bukanlah suatu penyelesaian yang baik bagi mempertingkatkan truput sistem. Ketika nilai Nisbah Isyarat-ke-Hingar (SNR) adalah malar, teknik ini tidak dapat menggunakan lebih dari satu skema pemodulatan bagi memeta data ke atas pembawa, dan hal ini akan menghasilkan truput sistem yang malar jika keadaan saluran masih tidak mengalami perubahan. Tesis ini mencadangkan dua model pemilihan pemodulatan. Polisi pemilihan dalam model pertama dikawal oleh SER. Dalam model ini, semua skema pemodulatan yang boleh disediakan akan digunakan bagi memeta data masukan pada nilai SNR yang rendah tetapi dengan peratusan penggunaan yang berbeza berdasarkan nilai SER yang boleh dikira dari simbol OFDM terakhir yang berjaya dihantar. Namun dalam model kedua pula, polisi pemilihan dikawal oleh nilai-nilai SNR dan SER tetapi terhad kepada dua skema pemodulatan sahaja. Di samping itu, kedua-dua model polisi ini kemudian digabungkan dengan teknik pengetipan bagi menghasilkan algoritma yang dinamakan Pemodulatan Adaptif dan Pengetipan (AMCl). Teknik pengetipan dalam algoritma ini bertujuan mengurangkan PAPR dan untuk mengawal polisi pemilihan dengan mentakrifkan pelbagai Nisbah Pengetipan (CRs) bergantung kepada tertib skema pemodulatan, SER dan SNR. AMCl juga menggunakan mekanisma pengemaskinian untuk mentakrifkan nilai CR pada setiap penghantaran simbol baru. AMCl pada kedua-dua model yang dibina menggunakan kod Matlab. Model yang dicadangkan telah diuji dan disahkan dalam tiga sistem tanpa wayar OFDM; IEEE802.11g, 802.16e dan 4G. Hasil penyelakuan menunjukkan model yang dicadangkan berupaya memperbaiki PAPR, mempertingkatkan kadar data pada semua nilai SNR dan memberikan prestasi SER yang baik. Contohnya, isyarat OFDM-256QAM bagi IEEE802.16e pada 10-3
telah ditambahbaik masing-masing sebanyak 6 dB dan 3.5 dB berbanding sistem normal dari segi PAPR dan SER. Kadar data meningkat dari 7.9 ke 38.1 Mbps pada SNR bernilai 5 dB.
vi
CONTENTS
Page
DECLARATION ii
ACKNOWLEDGMENTS iii
ABSTRACT iv
ABSTRAK v
CONTENTS vi
LIST OF SYMBOLS x
LIST OF ABBREVIATIONS xii
LIST OF FIGURES xiv
LIST OF TABLES xviii
CHAPTER I INTRODUCTION
1.1 Introduction 1
1.2 Orthogonal Frequency Division Multiplexing (OFDM) 3
1.3 Problem statement 5
1.4 Thesis objectives 7
1.5 Thesis contribution 8
1.6 Thesis scope and limitation 9
1.7 Thesis methodology 9
1.8 Thesis organization 11
CHAPTER II LITERATURE REVIEW
2.1 Introduction 12
2.1.1 The concept of multicarrier modulation 13 2.1.2 OFDM as multicarrier modulation 14 2.1.3 Evolution of OFDM 15 2.1.4 Applications of OFDM 17 2.1.5 OFDM system design issues 20 2.1.6 OFDM system design requirements 20 2.1.7 OFDM system design parameters 21
2.2 OFDM transceiver systems 22
2.2.1 OFDM modulation and demodulation 25 2.2.2 QAM mapping 28 2.2.3 OFDM implementation by FFT 32
vii
2.2.4 Intersymbol and intercarrier interference 35 2.2.5 Guard time and cyclic extension 37
2.3 Peak power problem 40
2.3.1 PAPR of OFDM 40 2.3.2 Reasons for reducing PAPR in OFDM 40 2.3.3 PAPR calculation in OFDM 42 2.3.4 Distribution of PAPR 43 2.3.5 Effects of PAPR 44
2.4 PAPR reduction techniques 44
2.4.1 Clipping and filtering technique 46 2.4.2 Significance of clipping technique 48
3.5 Complementary cumulative distribution function 84
3.6 Signal to noise ratio 87
3.7 Error vector magnitude 88
3.8 Symbol error rate 90
3.9 The proposed adaptive modulation selection models 92
viii
3.10 The proposed system models 93
3.10.1 The algorithm AMCl in model (1) 95 3.10.2 CR definition mechanism in model (1) 99 3.10.3 The algorithm AMCl in Model (2) 102 3.10.4 CR definition mechanism in model (2) 105 3.10.5 Data rate calculation in proposed models (1) and (2) 106 3.10.6 The complexity of AMCl algorithm 107
3.11 Summary 107
CHAPTER IV ADAPTIVE MODULATION SELECTION POLICY MODELS
4.1 Introduction 109
4.2 Performance of normal OFDM system 111
4.2.1 Order of modulation scheme 111 4.2.2 Cyclic prefix and number of subcarriers 115 4.2.3 Processing and multipath delay 124
4.3 Effects of clipping ratio (CR) 130
4.3.1 CR and the modulation schemes order 130 4.3.2 CR and percentage of the clipped samples 132 4.3.3 CR and SER performance 135
4.4 Out-of-band radiation 145
4.5 In-band distortion 148
4.5.1 Error performance of clipped passband OFDM signal 148 4.5.2 EVM of the clipped passband OFDM signal 149
4.6 The utilization percentage in model (1) 151
4.7 The utilization percentage in model (2) 154
4.8 Summary 159
CHAPTER V THE PERFORMANCE OF AMCl IN OFDM BASED WIRELESS SYSTEMS
5.1 Introduction 160
5.2 The selection policy of AMCl in model (1) 161
5.3 The selection policy of AMCl in model (2) 174
5.4 Validation of AMCl in model (1) 177
5.4.1 SER performance 177 5.4.2 Improvement in PAPR 187 5.4.3 Throughput enhancement 193
ix
5.5 Validation of AMCl in model (2) 202
5.5.1 SER performance comparison 202 5.5.2 PAPR reduction 207 5.5.3 Data rate enhancement 211
5.6 Cost function of algorithm AMCl 217
5.7 Summary 218
CHAPTER VI CONCLUSIONS AND FUTURE WORK
6.1 Introduction 222
6.2 Conclusions 222
6.5 Future work 225
REFERENCS 226
APPENDICES
A List of publications 234
x
LIST OF SYMBOLS
k Carrier number
l OFDM symbol number
m Transmission frame number
N Number of transmitted carrier
N Number of signaling symbols per subband s
T Symbol duration s
T Inverse of the carrier spacing U
T’ Duration of a data symbol
f Central frequency of the radio frequency (RF) signal c
k’ Carrier index relative to the center frequency
c Complex symbol for carrier k of the Data symbol no. n in frame number
m,n,k
α Minimum distance separating two constellation points
C Number of points in a constellation
τ Delay spread max
F OFDM frame
f Doppler frequency D
A Clipping level
z Peak threshold value
σ Root mean squared power of the unclipped OFDM signal
S Average signal power
P Total signal power
P BER performance in an AWGN channel G
pb Bit error probability
γ Signal to Noise Ratio (SNR)
l SNR thresholds between the modulation schemes i
M Number of the signal constellation states
B Average spectral efficiency
η Spectral efficiency
R Signal data rate
V Average voltage of the signal modulation constellation
xi
X Transmitted data symbol on the n-th subcarrier n
E Average energy per bit av
xii
LIST OF ABBREVIATIONS
ADC Analog-to-Digital Converter
ADSL Asymmetrical Digital Subscriber Line
AM Adaptive Modulation
AMC Adaptive Modulation and Coding
AMCl Adaptive Modulation and Clipping
AOFDM Adaptive OFDM
ATM Asynchronous Transfer Mode
AWGN Additive White Gaussian Noise
BPSK Binary Phase Shift Keying
BWIF Broadband Wireless Internet Forum
CCDF Complementary cumulative distribution function
CP Cyclic prefix
CR Clipping Ratio
CSI Channel State Information
DAB Digital Audio Broadcasting
DAC Digital-to-Analog Converter
DER Detection Error Ratio
DFFT Discrete Fast Fourier Transform
DTV Digital Television (based on MPEG-2 video signal)
DSP Digital Signal Processing
DVB Digital Video Broadcasting
DVB-T Digital Video Broadcasting-Terrestrial
DUT Device Under Test
ETSI European Telecommunication Standards Institute
EVM Error Vector Magnitude
FDD Frequency Division Duplex
FEC Forward Error Correction
FFT Fast Fourier Transform
HPA High Power amplifier
IBI Inter-block interference
IBO Input Back-Off
xiii
ICI Inter-Carrier Interference
IEEE Institute of Electrical and Electronics Engineers
where Tprefix is the guard time duration appending to the beginning of the symbol and
Tpost
is the guard time duration appending to the end of the symbol.
At the receiver side, CP is removed before any processing starts. As long as
the length of CP interval is larger than maximum expected delay spread τmax
SNR
, all
reflections of previous symbols are removed and orthogonality is restored. The
orthogonality is lost when the delay spread is larger than length of CP interval.
Inserting CP has its own cost, a part of signal energy is lost since it carries no
information. The loss is measured as
loss_cp= −10log10(1-TCP/Ts
)
(2.19)
40
Here, TCP is the interval length of CP and Ts
2.3 PEAK POWER PROBLEM
is the OFDM symbol duration. It
is understood that although part of signal energy is already lost, the fact that zero ICI
and ISI situation pay off the loss.
An OFDM signal has an approximately Gaussian amplitude distribution when the
number of subcarriers is large. Therefore, very high peaks in the transmitted signal
can occur. This property is often measured via the signal's PAPR.
2.3.1 PAPR of OFDM
Peak to Average Power Ratio (PAPR) is proportional to the number of sub-carriers
used for OFDM systems. An OFDM system with large number of sub-carriers will
thus have a very large PAPR when the sub-carriers add up coherently. Large PAPR of
a system makes the implementation of Digital-to-Analog Converter (DAC) and
Analog-to-Digital Converter (ADC) to be extremely difficult.
The design of RF amplifier also becomes increasingly difficult as the PAPR
increases. If the dynamic ranges of the A/D and D/A are increased, the resolution also
needs to be increased in order to maintain the same quantization noise level.
Therefore, an OFDM signal may require expensive A/Ds and D/As compared to many
other modulation formats.
Also, a large power back-off of the amplifiers is necessary. A major
disadvantage of an OFDM system is its large PAPR. As the number of sub-carriers N
increases, the maximum possible peak power becomes N times the average power (Li
& Cimini 1998).
2.3.2 Reasons for Reducing PAPR in OFDM
41
When the PAPR of an OFDM transmission is high, the D/A converters and power
amplifiers require a large dynamic range to avoid clipping of the given signal
mitigating undesirable consequences, such as signal distortion and spectral spillage.
Moreover, a large dynamic range implies increased complexity, reduced
efficiency and increased cost of the components. The reasons for reducing PAPR will
be elaborated in the following two sections.
a. Dynamic Range of D/A Converters and Power Amplifiers
Amplitude clipping of an OFDM signal causes several undesirable outcomes, such as
signal distortion and spectral regrowth (Li & Cimini 1998). It also causes in-band
noise, which results in SER performance degradation, and higher order harmonics that
spill over into out-of-band spectrum.
It has been shown that filtering after the power amplifier to remove this
spectral leakage is very inefficient with respect to power usage, thus making it an
undesirable solution (Krongold 2003). Therefore, the dynamic range of D/A
converters should be large enough to accommodate large peaks of signals, i.e. high
PAPR.
A high precision D/A converter supports high PAPR with reasonable
quantization noise, but may be very expensive for a given sampling rate of the system.
On the hand, low precision D/A converter would be cheaper, but quantization noise
will be significant which reduces signal SNR, when the dynamic range of the D/A
converter is increased to support high PAPR. Otherwise, the D/A converter will
saturate and clipping will occur (Tellado 2000).
Similarly, the dynamic range of the power amplifiers should also be large to
accommodate large PAPR. Otherwise, power amplifiers may saturate, resulting in
amplitude clipping. The component cost of the D/A converters and power amplifiers
increase in the dynamic range.
42
b. Power Savings
Power amplifiers with a high dynamic range exhibit poor efficiency, which is the ratio
of power delivered to the load and total power consumed.
For a given OFDM signal, the average input power needs to be adjusted such
that the peaks of the signal are rarely clipped. Therefore, the efficiency of the power
amplifiers is inversely proportional to the PAPR. Therefore, the net power savings is
directly proportional to the desired average output power and is highly dependent
upon the clipping probability level (Baxley & Zhou 2004).
2.3.3 PAPR Caculation in OFDM
An OFDM signal consists of a sum of independent signals modulated over several
orthogonal subcarriers of equal bandwidth. Therefore, when added up coherently, the
OFDM signal may exhibit large peaks, while the mean power remains relatively low.
Being a variant of OFDM signals suffer from this same problem.
By definition, the peak-to-average power ratio (PAPR) is the ratio of the peak
instantaneous power to the average power of a given signal, which characterizes the
envelope variations of the signal in time domain (Tellado 2000).
Recalling Equation (2.1), with Ns the number of subcarriers, di complex
modulation symbol, and s(t) one OFDM symbol starting at t=ts
, the PAPR formula is
given by:
{ }2
2
0
)(
)(max))(PAPR(
tsE
tsts Tt≤≤= (2.20)
where E{∙} denotes the expectation operator. Without loss of generality, the cyclic
extension can be neglected safely from analysis since it does not contribute to the
PAPR problem.
43
The continuous time PAPR of s(t) can be approximated using the discrete time
PAPR, which is obtained using samples of the OFDM signal s(n). It has been shown
that an oversampling factor of four is sufficient to estimate the continuous PAPR of a
BPSK system (Tellambura 2001; Yu & Wei 2003).
2.3.4 Distribution of PAPR
When the number of sub-carriers in an OFDM system is high, conventional OFDM
signals can be regarded as Gaussian noise like signals; their variable amplitude is
approximately Rayleigh distributed, and the power distribution has a cumulative
distribution function given by
F (z) = 1-e-z (2.21)
The probability that the PAPR is below some threshold is given by
P (PAPR ≤ z) = F (z)N = (1-e-z)N
(2.22)
where z is the Peak threshold value, and N is the total sub-carrier number. The
distribution is shown in Figure 2.13.
44
Figure 2.13 Cumulative distribution function of PAPR
Source: Hazy 2004
2.3.5 Effects of PAPR
The effects of the large Peak to Average Power ratio are:
a. The power amplifiers at the transmitter need to have a large linear range of
operation. When considering a system with a transmitting power amplifier, the
nonlinear distortions and peak amplitude limiting introduced by the High
Power amplifier (HPA) will produce inter-modulation between the different
carriers and introduce additional interference into the system. This additional
interference leads to an increase in the Symbol Error Rate (SER) of the system.
One way to avoid such non-linear distortion and keep low SER is by forcing
the amplifier to work in its linear region. Unfortunately such solution is not
power efficient and thus not suitable for wireless communication.
b. The Analog to Digital converters and Digital to Analog converters need to
have a wide dynamic range and this increases complexity.
2.4 PAPR Reduction Techniques
Numerous techniques have been proposed in the literature that attempt to reduce
PAPR of an OFDM signal. The PAPR can be reduced by amplitude clipping or
companding, by using a PAPR reduction code, by phase adjustments, or by precoding
(Du & Swamy 2010). These solutions include clipping and filtering (Li &
Cimini.1998), (Wang & Tellambura 2005), error control coding (Ahn et al. 2000),
(Jones et al. 1994), and constellation shaping techniques (phase, power, or both)
(Sezginer & Sari 2005; Krongold & Jones 2003; Han & Lee 2004).
The PAPR techniques can achieve a decrease in PAPR, but at the cost of
increased system complexity, reduced information rate, or degraded SER
45
performance. Moreover, due to non-contiguous subcarriers, the PAPR reduction
techniques proposed for the OFDM may need to be modified for reducing the values
of PAPR for OFDM signals.
Intentional or accidental clipping of the OFDM signal often occurs in practice.
The clipping of a received sample affects all subcarriers in the system. The sensitivity
to clipping effects is investigated by, e.g., (Gross & Veeneman 1993). At least three
concepts for reducing the peak-to-average power ratio have been proposed (Muller &
Huber, 1997; Narahashi & Nojima 1997; Jones & Wilkinson 1994; Van Nee 1996;
Tellado & Cioffi 1997). In the first concept one signal with a low peak-to-average
power ratio out of a set of signals is transmitted. In (Narahashi & Nojima
1997), for
instance, it is observed that by appropriately choosing the phase of each subcarrier the
peak-to-average power ratio can be reduced.
The second concept reduces the peak-to-average power ratio by coding. Block
codes can accomplish this, see, e.g., (Jones & Wilkinson 1994), and in (Van Nee
1996) it is shown that complementary codes have good properties to combine both
peak-to-average power reduction and forward error correction. Finally, in (Tellado &
Cioffi
1997) impulse-like time-domain functions are iteratively subtracted from the
original signal to reduce the peaks. These time-domain signals are generated by a set
of reserved, unused symbols in the DFT domain. Subcarriers which are not used to
transmit data symbols are used to transmit symbols, chosen to generate a transmitted
signal with low peak-to-average power ratio.
The PAPR reduction techniques broadly fall into one of the three techniques
(Sesia, et al. 2009):
a) Signal Distortion
b) Coding
c) Scrambling and Carrier selection
The proposed models in this thesis employed the clipping technique that fall
into the signal distortion techniques because of its simplicity and it happens at the
transmitter side. Moreover by a proper definition of the clipping level, the
46
improvement in the PAPR can be achieved without serious degradation in the SER
performance.
2.4.1 Clipping and Filtering Technique
Clipping is a way of reducing PAPR by very simply limiting the maximum amplitude
of the OFDM signal to a desirable maximum. Amplitude clipping limits the peak
envelope of the input signal to a predetermined value or otherwise passes the input
signal through unperturbed, that is, Then, the real valued bandpass samples, x, were
clipped at amplitude A as follows:
>≤≤
−<=
) (if )- (if
) (if -
AxAAxAx
AxAy (2.23)
where x denotes the signal before clipping and y denotes the signal after clipping.
This is clearly seen in Figure 2.14, where the clipping level is equal to 50 V.
When the OFDM peaks are greater than 50 or less than -50 V, the signal will be
clipped. Since the probability of the occurrence of the high peak power is low,
clipping is effective for reducing the PAPR.
47
2 4 6 8 10 12 14
x 10-7
-50
0
50
Time response of original signal
Time (sec)
Am
plitu
de (v
)
2 4 6 8 10 12 14
x 10-7
-50
0
50
Time response of clipped signal
Time (sec)
Am
plitu
de (v
)
Figure 2.14 Effects of clipping on OFDM signal
Figure 2.15 is illustrated the clipping and filtering technique. The distortion
caused by amplitude clipping can be viewed as another source of noise. The noise
caused by amplitude clipping falls in-band and out-of-band. In-band distortion cannot
be reduced by filtering and results in error performance degradation, while out-of-
band radiation reduces spectral efficiency.
Figure 2.15 Clipping and filtering
Source: Ergen 2009
x(t) y(t)
48
Filtering after clipping can reduce out-of-band radiation, but may also cause
some peak regrowth so that signal after clipping and filtering will exceed the clipping
level at some points. To reduce overall peak regrowth, a repeated clipping and
filtering operation can be used. Generally, repeated clipping and filtering takes many
iterations to reach a desired amplitude level. When repeated clipping and filtering is
used in conjunction with other PAPR reduction techniques described here, the
deleterious effects may be significantly reduced.
But clipping also has some disadvantages in the sense that it acts on the whole
of the signal instead of only on those points above threshold. This actually reduces the
SER. Definition of clipping level is very important issue and needs long observation
to the envelope of OFDM signal with different modulation schemes. The envelope of
OFDM signal has high fluctuation between low and high peaks. This fluctuation
increases as the order of the modulation scheme is increased. Fortunately the high
peaks in OFDM samples rarely happen.
This leads to necessity to implement simulation to observe the behavior of
OFDM sample and search for these high peaks. Finding these high peaks will ease the
mission of choosing the appropriate CR for each modulation scheme. In fact the
appropriate value of clipping ratio (CR) can be chosen based on other performance
metrics in OFDM system such as SER and SNR. The degree of clipping (hard or soft)
can affect on the value of SER, for example clipping OFDM signal at low CR will
cause serious degradation in SER.
The situation of channel decides the level of clipping that can be chosen. It is
better to use high CR at high SNRs and vice versa. Some moderate values of CR can
be use at high SNRs only because it offers reasonable SER performance. This leads to
necessity to perform classification of the values of CRs based on the order of
modulation schemes, calculated SER, and estimated SNR. With this classification the
clipping technique can offer compromise between PAPR improvement and reasonable
SER performance.
2.4.2 Significance of Clipping Technique
49
One of the limitations of using OFDM is the high peak-to-average power ratio (PAR)
of the transmitted signal. A large PAR leads to disadvantages such as increased
complexity of the analog to digital converter (A/D) and reduced efficiency of the radio
frequency (RF) amplifier. If power amplifiers are not operated with large linear-power
back-offs, it is impossible to keep the out-of-band power below imposed limits. This
leads to very inefficient amplification, expensive transmitters and causing
intermodulation among the subcarriers and undesired out-of-band radiation. The PAR
reduction techniques are therefore of great importance for OFDM systems.
Partial transmit sequences (PTS), selected mapping (SLM), active
constellation extension (ACE), Golay sequences and Reed-Muller codes (GRC),
clipping signal scheme, selective scrambling (SLS) and windowing signal scheme
were recently proposed to reduce PAR. The side information (SI) must be transmitted
to the receiver when using PTS, SLM and SLS methods, and then the channel
efficiency drops a little.
The complexity of using GRC method is too high to employ at real-time
practice, and it is only efficacious when the subcarrier number equals to several
special values as well. The ACE scheme wastes the precious power, especially when
the mobile’s power is very limited. The clipping signal scheme is relatively simpler
than others.
And the performance of the clipping scheme is superior to that of the
windowing signal scheme in OFDM systems, because the windowing technique
distorts all signals, but the clipping technique distorts small portion signal where the
peak power exceeds the max-permitted power. In this thesis the modulation selection
policy is combined with the clipping technique to produce an algorithm that can
improve the performance of OFDM based wireless systems in terms of PAPR, SER,
and data throughput.
When using the clipping scheme, the phase information of the signal is
completely transmitted, and the amplitude of the signal is clipped if the power of the
signal exceeds the max-permitted power (Dukhyun & Gordon 2007).
50
2.5 ADAPTIVE MODULATION
Current and future demands in wireless communication for various high speed
multimedia data services entail a robust, high data rate transmission system.
Increasing numbers of users amid limited spectrum motivate research on technology
to expand the capacity and increase spectral efficiency.
Adaptive modulation is a technique that varies some transmission parameters
to take advantage of favorable channel conditions. Under bad channel conditions, a
robust signal transmission mode will be applied to ensure reliable data exchange.
While, in good channel, spectrally efficient mode that offer higher throughput is
applied. In other words Adaptive Modulation (AM) is a technique used to provide
more user capacity over the air during good propagation conditions, where the
modulation level of the radio link adapts dynamically to the conditions of the path.
In traditional point to point systems the modulation is fixed at a certain level,
delivering a constant throughput for a defined channel bandwidth. This mechanism
ensures the most efficient mode is always used based on certain criteria and
constraints. The varying parameters can be the symbol transmission rate, transmitted
power level, constellation size, SER, code rate or scheme, any combination of these
parameters (Chung & Goldsmith 2001).
Compared to the fixed modulation system, which was designed specifically for
the worst case channel conditions, this adaptive modulation offers higher spectral
efficiency, higher throughput and remarkable capacity enhancement without
sacrificing SER or wasting power (Catreux et al. 2002). An early work includes in
(Webb & Hanzo 1994), where W.T. Webb and L. Hanzo introduced variable rate
QAM.
The transmitter varies the signal constellation size from 1bit/symbol
corresponding to BPSK to 6 bits/symbol star 64-QAM. In a good quality channel, the
constellation size is increased, and as the channel quality become worst, i.e. as the
51
receiver enters a deep fade, the constellation size is decreased to a value, which
provides an acceptable SER.
2.5.1 The Conventional Adaptive Modulation for OFDM
Adaptive Modulation in OFDM system refers to the automatic modulation adjustment
that OFDM based wireless system can make to prevent weather-related fading from
causing communication on the link to be disrupted. The goal of adaptive modulation is
to choose the appropriate modulation mode for transmission in each subcarrier, given
the local SNR γn
, in order to achieve a good tradeoff between throughput and overall
SER. The acceptable overall SER varies depending on other systems parameters, such
as the correction capability of the error correction coding and the nature of the service
supported by this particular link.
Figure 2.16 shows the concept of the adaptive modulation, this concept of the
selection policy states that a general estimate of the channel conditions needed for
different modulation schemes. When heavy weather conditions, such as a storm, affect
the transmission and receipt of data and voice over the wireless network, the radio
system automatically changes modulation so that non-real-time data-based
applications may be affected by signal degradation, but real-time applications will
continue to run uninterrupted. As the range is increased, it must be step down to lower
modulations (in other words, 16-QAM), but as the range get closer, it should be utilize
higher order modulations like 256-QAM for increased throughput.
Figure 2.16 The Conventional modulation selection policy
16-QAM 256-QAM 64-QAM
52
Since OFDM signals are modulated, varying the modulation also varies the
amount of bits that are transferred per signal thereby enabling higher throughputs and
better spectral efficiencies. For example, 256-QAM modulation scheme can deliver
approximately four times the throughput of 4-QAM. It should be noted, however as a
higher modulation technique is used, a better SNR is needed to overcome interference
and maintain a tolerable SER level. The problem arises in the conventional selection
policy when the channel conditions still the same for long time and the value of SNR
still without change (especially the low SNRs). This will reflect in a constant
throughput. Sometimes the transmitter receives wrong reading of the estimated SNR
and this means poor performance at all. As discussed before, the adaptive system has
to employ appropriate techniques in order to fulfill the following requirements:
− Channel quality estimation,
− Choice of the appropriate modulation modes, and
− Signalling or blind detection of the modulation modes.
2.5.2 Chanel Quality Estimation
It is well known that the wireless channel causes an arbitrary time dispersion,
attenuation, and phase shift in the received signal. The use of OFDM and a cyclic
prefix mitigates the effect of time dispersion. However, it is still necessary to remove
the amplitude and phase shift caused by the channel if you want to apply linear
modulation schemes. The function of channel estimation is to form an estimate of the
amplitude and phase shift caused by the wireless channel from the available pilot
information. The equalization removes the effect of the wireless channel and allows
subsequent symbol demodulation.
A number of different algorithms can be employed for these modules. This
application note considers simple techniques that illustrate the feasibility of
implementation and showcase the design methodology. Although the guard time
which has longer duration than the delay spread of a multipath channel can eliminate
ISI completely, the received symbol is still interfered by its replicas received from
multipath components.
53
This corresponds to frequency-selective fading of the symbol. In order to
compensate this distortion, a one-tap channel equalizer is needed for each sub-carrier.
At the output of FFT on the receiver side, the sample at each sub-carrier is multiplied
by the coefficient of the corresponding channel equalizer.
The coefficient of an equalizer can be calculated based on the zero-forcing
(ZF) criterion or the minimum mean-square error (MMSE) criterion (Sari, et al 1995).
The ZF criterion forces ISI to be zero at the sampling instant of each sub-carrier. The
coefficient of a one-tap ZF equalizer is calculated as follows:
nHnc 1= (2.24)
Where Hn is the channel frequency response within the bandwidth of the n-th
sub-carrier. The disadvantage of the ZF criterion is that it enhances noise at the n-th
sub-carrier if Hn
To make the tradeoff between ISI and noise, MMSE criterion is used and the
coefficient of a one-tap MMSE equalizer is calculated as follows:
is small, which corresponds to spectral nulls.
2
22
symbol
noisen
nn
H
Hc
σσ+
=∗
(2.25)
Where 2noiseσ is the noise variance and 2
symbolσ is the variance of source
symbols. A MMSE equalizer gives better performance than a ZF equalizer when
spectral nulls are present in the channel frequency response.
Equation 2.24 and 2.25 show that one needs to perform channel estimation in
order to obtain weights for equalizers on individual sub-carriers. Training symbols,
also known as pilot symbols, are also often used to perform channel estimation. In
OFDM, since equalization is performed in the frequency domain, it is the channel
frequency response that must be estimated. In the multipath environment, the
demodulated symbol Xn on the n-th subcarrier at the output of FFT without ISI and
ICI can be represented as:
54
( ){ } nn
L
lNnll
n NXjHY +
= ∑
−
=
1
0
2-exp)0( π (2.26)
Where L is the number of multipath components, Nn is the FFT of the additive
white Gaussian noise (AWGN) on the n-th subcarrier and Hl(0) is the channel
frequency response of the l-th multipath component at the zero-th frequency. To
estimate the channel frequency response, pilot symbols are inserted on the sub-carriers
in the frequency domain, i.e., they are inserted before IFFT operation at the
transmitter side. Let Hn be the channel frequency response experienced by Xn
, i.e.
∑−
=
=
1
0
2-exp)0(L
l
ln N
nljHH π (2.27)
The channel frequency response experienced by the pilot symbol P n
on the n
th subcarrier can be estimated as
n
nn
n
nn P
NH
PY
H +==∧
(2.28)
Since pilot symbols usually occupy a small amount of bandwidth n-th for
spectral efficiency, interpolation across frequency is required to estimate the channel
frequency response where pilot symbols are not located. The channel frequency
response at the m-th subcarrier mH∧
can be interpolated linearly as (Sampei & Sunaga.
1993)
∧∧∧
+
−= 211 pp H
NmH
NmH , p1≤m≤p2 (2.29)
Where 1pH∧
and 2pH∧
are the channel frequency responses estimated by the
pilot symbols on the p1-th and p2-th sub-carriers. Furthermore, if the multipath
channel is time-varying in nature, then interpolation across time may also require
tracking the channel. To determine the minimum pilot spacing in time and frequency
in OFDM, the bandwidth of the channel variation is necessary to be found in time and
frequency. These bandwidths are equal to the maximum Doppler frequency maxDf in
the time domain and the maximum delay spread maxτ in the frequency domain.
55
According to the sampling theorem, the pilot spacing in time ts and frequency fs is
(Hoeher, et al 1997).
sD
t Tfs
max2
1≤ (2.30)
F
s f ∆≤
max21
τ (2.31)
where sT is the OFDM symbol duration and F∆ is the frequency spacing between
two sub-carriers. Decreasing the pilot spacing improves the estimation of channel
frequency response but decreases bandwidth efficiency. On the other hand, increasing
the pilot spacing beyond the one specified by the sampling theorem decreases the
accuracy of the channel estimation but increases the bandwidth efficiency. Hence, the
chosen pilot density is a tradeoff between the performance of channel estimation and
bandwidth efficiency. Moreover, besides interpolating the channel frequency response
in the frequency and time domain separately, a two-dimensional interpolation can also
be applied in OFDM. More detail on the two-dimensional interpolation scheme can be
found in (Sari, et al 1995; Hoeher, et al 1997).
a. Channel Estimate Impact on Adaptive Modulation
In this section, the assumption of ideal channel knowledge will be removed that it was
held in the previous section. Here, the performance curves were found become even
more degraded. In the following plots, FFT channel estimation and SNR estimation
were incorporated into the system. The effectiveness of FFT estimation is based on the
pilot to information symbol ratio. There is a Nyquist condition that must be met
(Okamoto, et al. 1991).
Based on this criterion, the following relationship must be satisfied:
N
Tf sd 21. ≤ (2.32)
Where fd is the maximum Doppler frequency of the channel; Ts is the symbol period,
and N is the number of symbols in a frame and a frame is defined as a set of symbols
associated with a single pilot symbol.
56
The more pilots used per frame, the higher Doppler rate the estimator can
compensate for. In these simulations, one pilot was used per every fifteen information
symbols.
b. Propagation Delay Impact on Adaptive Modulation
Thus far, it was assumed that there has been no time lag or propagation delay when
the receiver relays control information back to the transmitter. In the system so far,
there has been instantaneous relay between receiver and transmitter. The delay will be
introduced in this system that will amount to two frames worth of time. In other
words, from the time the receiver transmits information back to the transmitter, two
more frames will be on route to the receiver from the transmitter and are not privy to
the latest information most recently sent from the receiver.
2.5.3 Choice of the Modulation Schemes
The two communicating stations use the open loop predicted channel transfer function
acquired from the most recent received OFDM symbol, in order to allocate the
appropriate modulation modes to the subcarriers. The modulation modes were chosen
from the set of BPSK, QPSK, and 16-QAM, as well as No Transmission, for which no
signal was transmitted.
These modulation modes are denoted by Mm
, where m ∈ (0, 1, 2, 4) is the
number of data bits associated with one data symbol of each mode. In order to keep
the system complexity low, the modulation mode is not varied on a subcarrier by
subcarrier basis, but instead the total OFDM bandwidth of 512 subcarriers is split into
blocks of adjacent subcarriers, referred to as subbands, and the same modulation
scheme is employed for all subcarriers of the same subband.
2.5.4 Conventional Modulation Selection Algorithm
57
a. Fixed Threshold Adaptation Algorithm
The fixed threshold algorithm was derived from the adaptation algorithm proposed by
Torrance for serial modems (Torrance, et al. 1999). In the case of a serial modem, the
channel quality is assumed to be constant for all symbols in the time slot, and hence
the channel has to be slowly varying, in order to allow accurate channel quality
prediction.
Under these circumstances, all data symbols in the transmit time slot employ
the same modulation mode, chosen according to the predicted SNR. The SNR
thresholds for a given long term target SER were determined by Powell optimization
(Torrance & Hanzo 1996).
Torrance assumed two uncoded target bit error rates: 1% for a high data rate
(speech) system, and 10−4 for a higher integrity, lower data rate (data) system. The
resulting SNR thresholds ℓn for activating a given modulation mode Mm in a slowly
Rayleigh fading narrow band channel for both systems are given in Table 2.5.
Specifically, the modulation mode Mm is selected if the instantaneous channel SNR
exceeds the switching level ℓn
.
Table 2.5 Optimized switching levels for adaptive modulation for the (speech) and (data) system, shown in instantaneous channel SNR (dB)
System type ℓ ℓ0 ℓ1 ℓ2
Speech
3
-∞ 3.31 6.48 11.61
58
Data -∞ 7.98 10.42 16.76
Source: Torrance & Hanzo 1996
This adaptation algorithm originally assumed a constant instantaneous SNR
over all of the block’s symbols, but in the case of an OFDM system in a frequency
selective channel the channel quality varies across the different subcarriers. For sub–
band adaptive OFDM transmission, this implies that if the subband width is wider
than the channel’s coherence bandwidth (Webb & Hanzo 1994). Then the original
switching algorithm cannot be employed. For these investigations, therefore it is
essential to employ the lowest quality subcarrier in the subband for the adaptation
algorithm based on the thresholds given in Table 2.5.
b. Sub-band BER Estimator Adaptation Algorithm
As was seen in previous section, the fixed switching level based algorithm leads to a
throughput performance penalty, if used in a subband adaptive OFDM modem, when
the channel quality is not constant throughout each subband. This is due to the
conservative adaptation based on the subcarrier experiencing the most hostile channel
in each subband.
An alternative scheme taking into account the nonconstant SNR values γj
across the Ns subcarriers in the j-th subband can be devised by calculating the
expected overall bit error probability for all available modulation modes Mn
in each
subband. For each subband, the mode having the highest throughput, whose estimated
BER is lower than a given threshold, is then chosen.
While the adaptation granularity is still limited to the subband width, the
channel quality estimation includes not only the lowest quality subcarrier, which leads
59
to an improved throughput. It can be seen that the BER estimator algorithm results in
significantly higher throughput, while meeting the BER requirements. The BER
estimator algorithm is readily adjustable to different target bit error rates. Such
adjustability is beneficial, when combining adaptive modulation with channel coding.
Table 2.6 lists a brief comparison between the most known conventional
modulation selection algorithms in AM. The two proposed models in this thesis
suggest other performance metrics such as SER and CR to be used together with SNR
to select the appropriate modulation technique among the utilized modulation
schemes.
AMCl algorithm in both models presents new modulation selection policy in
order to improve the PAPR, enhance the data throughput and maintains a reasonable
error level in OFDM based wireless systems. This policy tries to avoid the
disadvantages of the conventional selection policies of Torrance and Hanzo.
Table 2.6 Conventional modulation selection algorithms in AM
Modulation selection algorithm
Properties Disadvantages
Fixed threshold adaptation (Torrance Model)
1. The channel quality is assumed to be constant for all symbols in the time slot.
2. All data symbols in the transmit time slot employ the same modulation mode.
3. Modes are chosen according to the predicted SNR.
4. The SNR thresholds for a
given long term target SER were determined by Powell optimization.
1. It assumes a constant instantaneous SNR over all of the block’s symbols, but in OFDM system in a frequency selective channel the channel quality varies across the different subcarriers.
2. If the subband width is wider than the channel’s coherence bandwidth. Then the original switching algorithm cannot be employed.
3. The fixed switching level
based algorithm leads to a throughput performance penalty, if used in a sub-band adaptive OFDM modem.
60
Subband BER estimator adaptation (Hanzo Model)
1. The expected overall bit error probability is calculated for all available modulation modes in each subband.
2. The mode having the highest throughput, whose estimated BER is lower than a given threshold, is then chosen.
3. The BER estimator algorithm results in significantly higher throughput, while meeting the BER requirements.
4. The BER estimator algorithm is readily adjustable to different target bit error rates.
5. The adjustability is beneficial, when combining adaptive modulation with channel coding.
1. Delay that comes from the calculation process of the quality of the received OFDM signal. This delay increases as the order of modulation mode increases.
2. This algorithm should employs channel coding technique to maintain the error level under the target value, this leads to overhead in the transmitted data.
3. Due to the inverse relation between BER and PAPR, thus this algorithm cannot offer low peak powers performance in OFDM system.
2.5.5 Signaling and Blind Detection
The adaptive OFDM receiver has to be informed of the modulation modes used for the
different sub-bands. This information can either be conveyed using signaling
subcarriers in the OFDM symbol itself or the receiver can employ blind detection
techniques in order to estimate the transmitted symbols’ modulation modes.
a. Signaling
The simplest way of signaling the modulation mode employed in a sub–band is to
replace one data symbol by an M-PSK symbol, where M is the number of possible
modulation modes. In this case, reception of each of the constellation points directly
signals a particular modulation mode in the current sub-band. In this case, for four
modulation modes, and assuming perfect phase recovery, the probability of a
signaling error ps(γ), when employing one signaling symbol is the symbol error
probability of QPSK. Then the correct sub-band mode signaling probability is:
61
( )( ) ( )( )2,11 γγ QPSKbpsp −=− (2.33)
where pb, QPSK is the bit error probability for QPSK:
( ) ( )
==
2erfc.
21, λγγ QQPSKbp (2.34)
which leads to the expression for the modulation mode signaling error probability of
( )2
2erfc.
2111
−−=
γγsp (2.35)
The modem mode signaling error probability can be reduced by employing
multiple signaling symbols and maximum ratio combining of the received signaling
symbols Rs,n, in order to generate the decision variable R's
prior to decision:
*,.
1,' nsH
sN
nnsRR
∑=
= (2.36)
where Ns is the number of signaling symbols per sub-band, the quantities Rs,n
nsH ,ˆ
are the
received symbols in the signaling subcarriers, and represents the estimated
values of the frequency domain channel transfer function at the signaling subcarriers.
Assuming perfect channel estimation and constant values of the channel transfer
function across the group of signaling subcarriers, the signaling error probability for
Ns
signaling symbols can be expressed as:
( )2
2 erfc.
2111,'
−−=
γγ sNsNsp (2.37)
b. Blind Detection by SNR Estimation
The receiver has no a prior knowledge of the modulation mode employed in a
particular received sub–band and estimates this parameter by quantizing the de–faded
received data symbols nn HR ˆ in the sub–band to the closest symbol mnR ,ˆ for all
62
possible modulation modes Mm for each subcarrier index n in the current subband.
The decision directed error energy em
for each modulation mode is calculated
according to:
( )2,ˆ∑ −=
nmnnnm RHRe (2.38)
The modulation mode Mm which minimizes em
is chosen for the demodulation
of the sub-band. It can be seen that the detection performance depends on the number
of symbols per sub-band, with fewer sub-bands and therefore longer symbol
sequences per sub–band leading to a better detection performance. It is apparent,
however, that the number of available modulation modes has a more significant effect
on the detection reliability than the block length.
If all four legitimate modem modes are employed, then reliable detection of
the modulation mode is only guaranteed for AWGN SNR values of more than 15-18
dB, depending on the number of sub-bands per OFDM symbol. If only M0 and M1
are
employed, however, the estimation accuracy is dramatically improved. In this case,
AWGN SNR values above 5-7 dB are sufficient to ensure reliable detection.
c. Blind Detection by Multi-mode Trellis Decoder
If error correction coding is invoked in the system, then the channel decoder can be
employed to estimate the most likely modulation mode per sub–band. Since the
number of bits per OFDM symbol is varying in this adaptive scheme, and the channel
encoder’s block length therefore is not constant, for the sake of implementational
convenience a convolutional encoder at the transmitter must be chosen.
Once the modulation modes to be used are decided upon at the transmitter, the
convolutional encoder is employed to generate a zero–terminated code–word having
the length of the OFDM symbol’s capacity. This codeword is modulated on the
subcarriers according to the different modulation modes for the different sub–bands,
and the OFDM symbol is transmitted over the channel. At the receiver, each received
63
data subcarrier is demodulated by all possible demodulators, and the resulting hard
decision bits are fed into parallel trellises for Viterbi decoding.
2.5.6 Adaptation Benefits
The fundamental idea in AM is to shift to a lower modulation when the link budget of
a system is no longer sufficient, typically during a rain fade. This modulation shift
results in an improved link budget (receiver threshold is improved and transmit power
can be increased), allowing the system to still operate, but at a lower throughput. In
order for this to be valuable, the microwave system must be able to perform priority
queuing on the user traffic.
By engineering a network with adaptive modulation, the operator can perform
several optimizations. For an existing operator with an existing network and frequency
licenses, the operator can use adaptive modulation to get much more throughput out of
the existing channels. For example, a channel of 28 MHz may have only been
engineered to 16QAM before, with about 80 Mbps throughput, but with adaptive
modulation, could now run at 256QAM during regular operation at close to 200 Mbps.
This can enable a two to three times capacity increase without having to
acquire additional or new licenses, or change antenna sizes. For a new deployment,
adaptive modulation can be used to minimize the antenna sizes of the link, while still
being able to transport high capacities. This will reduce antenna equipment costs, as
well as installation costs. More significantly, this will also reduce ongoing tower lease
costs, which are a significant portion of operating costs for many service providers.
Alternatively, in a new deployment, adaptive modulation may used to optimize
spectrum usage, and minimize annual frequency lease costs. This is especially
significant in Europe, where spectrum costs are quite high.
2.5.7 Adaptation Boundaries
Adaptive modulation is a powerful technique for maximizing the data throughput of
subcarriers allocated to a user. Adaptive modulation involves measuring the SNR of
64
each subcarrier in the transmission, then selecting a modulation scheme that will
maximize the spectral efficiency, while maintaining an acceptable SER. Adaptive
modulation is utilized to take advantage in the randomness of the channel. When the
channel is in good condition, the transmission is performed with higher data rates, and
when the channel is poor, the transmission rate is lowered with small constellation and
low rate codes. The channel side information is feeded to transmitter in order to
control transmit power, transmit constellation, and the coding rate (Ergen 2009).
Using adaptive modulation in a wireless environment is much more difficult as
the channel response and SNR can change very rapidly, requiring frequent updates to
track these changes. Adaptive modulation in wireless environment has not been used
extensively (Kim et al. 2004; Wong et al. 1999; Falahati et al. 2004; Tang & stolpman
2004), since the channel response and SNR can change very rapidly, and requires
frequent updates to track these changes. In 1999 Wong and Cheng investigated the
effectiveness using an adaptive subcarrier, bit and power allocation and use of
adaptive modulation and adaptive user allocation reduced the transmit power by 10 dB
(Zhang 2004; Shen et al. 2003).
It is important to know how to change the modulation scheme. In other words,
there is need to a way for the system to decide which modulation scheme is best suited
for the present conditions. Pons and Dunlop (1998) claimed that SER at the receiver
would be a good channel metric to decide switching. However, the decision was made
to use the metric that Pons and Dunlop rejected, which is to estimate the SNR of the
link. Reliable SER estimation is difficult over short periods and thus would restrict
adaptation rate. Therefore the selection policy in the adaptive modulation is based on
the estimated SNR. In other words the best modulation scheme will be selected based
on the estimated SNR. The SNR threshold can be defined based on the AWGN
performance of each modulation scheme. SNR can be defined as the signal power
divided by noise power. This resultant signal power is the instantaneous received
signal power and can be compared directly to the noise power, thus allowing the
system to consider the SER in an AWGN channel. From Torrance and Hanzo (1996)
the probability of bit error of QPSK, 16QAM, and 64QAM as follows:
65
)()( γγ QQPSKP = (2.39)
+
+
=
5213
5541)(16
γγγγ QQQQAMP (2.40)
−
+
+
−
−
+
+
+
+
+
+
+
+
+
=
13119
753
753
75364
21121
21121
2161
2161
2141
2141
2131
21121
21121
2161
2161
21212121121)(
γγγ
γγγγ
γγγγ
γγγγγ
QQQ
QQQQ
QQQQ
QQQQP QAM
(2.41)
In Equations 2.39 – 2.41, γ is the SNR, and Q(.) is the Q function,
dxexQx
x
∫=∞ −
2
2
21)(π
(2.42)
Refer to Table 2.7 and at an operating SER of 10-3
(as example) there is no
modulation scheme that gives the desired performance at an SNR below 10dB.
Therefore, QPSK was chosen as it is the most robust.
Table 2.7 AWGN switching thresholds
Modulation scheme SNR Threshold
QPSK (4-QAM) SNR < 17 dB
16-QAM 17 dB ≥ SNR ≥ 23 dB
64-QAM SNR > 23 dB
Source: Chan 2003
2.5.8 Limitations of Adaptive Modulation
66
There are several limitations with adaptive modulation. Overhead information needs
to be transferred, as both the transmitter and receiver must know what modulation is
currently being used. Also as the mobility of the remote station is increased, the
adaptive modulation process requires regular updates, further increasing the overhead.
There is a tradeoff between power control and adaptive modulation. If a remote station
has a good channel path the transmitted power can be maintained and a high
modulation scheme used (i.e. 64-QAM), or the power can be reduced and the
modulation scheme reduced accordingly (i.e. QPSK). Distortion, frequency error and
the maximum allowable power variation between users limit the maximum
modulation scheme that can be used. The received power for neighboring subcarriers
must have no more than 20 - 30 dB variation at the base station, as large variations can
result in strong signals swamping weaker subcarriers. Inter-modulation distortion
results from any non-linear components in the transmission, and causes a higher noise
floor in the transmission band, limiting the maximum SNR to typically 30 - 60 dB.
Frequency errors in the transmission due to synchronization errors and Doppler shift
result in a loss of orthogonality between the subcarriers. A frequency offset of only 1-
2 % of the subcarrier spacing results in the effective SNR being limited to 20 dB
(Moose 1994). Moreover the limited SNR restricts the throughput of the system.
The problem in the conventional policy is the system throughput at low SNRs.
In other words, the best modulation scheme should be selected based on SNR, this
means the system throughput will be kept constant as long as the channel conditions
does not change. This is because each modulation scheme is assigned to a group of
SNRs called SNR thresholds. Therefore if the estimated value of SNR is located
within the boundaries of one group, OFDM system will utilize only one particular
modulation scheme.
Another problem when wrong estimated value of SNR is received and this
problem can be worse if this wrong value of SNR is low. All these problems lead to
constant throughput for specific period of transmission time. The long observation of
the received OFDM signal shows the possibility of happen a sudden improvement in
SER even the transmitter still receiving the same SNR or value belongs to the same
group. This means that the system loses chance to change the order of modulation
scheme for some subcarriers that offer SER.
67
Unfortunately this leads to degradation in the data throughput of OFDM
system. It is better to use another performance metrics to make decision about the
order of modulation scheme in AM instead of SNR. This thesis will present models of
modulation selection policies that can offer huge enhancement in the system
throughput and introduce a compromise between the two important performance
metrics that are SER and PAPR.
2.5.9 Theoretical Performance of Adaptive Modulation
It is essential to discuss the theoretical performance of adaptive modulation, both in
terms of SER and spectral efficiency, and refer to Torrance and Hanzo (Torrance &
Hanzo 1996) for an analysis of adaptive modulation. First, PDF of the fluctuations of
the received, instantaneous, Rayleigh amplitude, s should be defined.
The envelope of a Rayleigh fading channel has a distribution of:
( )Ss-exp2),(S
sSsF = (2.43)
where, S is the average signal power. Next, it is necessary to determine the SER of
each modulation scheme. The SER can be determined analytically by:
dsSsFNsP
NSP G ),().()(
0∫=∞
γ (2.44)
where Pγ is the Rayleigh channel SER, and PRGR is the SER performance in an AWGN
channel. With the above two equations in addition to Equations 2.39-2.41, it is easy to
find the SER performance of adaptive modulation as:
68
∫+
∫+
∫
= −
dsSsFNsP
dsSsFNsP
dsSsFNsP
BNSP
l
lQAM
l
lQAM
l
lQPSK
A
),().(4
36
),().(3
24
),().(2
12
.)(
64
161 (2.45)
where the li are the SNR thresholds between the modulation schemes and B is the
average spectral efficiency. The values of li
can be inferred from Table 2.5. The value
of B is computed as follows:
∫ ∫ ∫++=2
1
3
2
4
3),(.6),(.4),(.2
l
l
l
l
l
ldsSsFdsSsFdsSsFB (2.46)
Having established the mathematical foundation behind adaptive modulation,
it is better to investigate the results in graphical form.
2.5.10 Parameters Adaptation
Different transmission parameters can be adapted to the anticipated channel
conditions, such as the modulation and coding modes.
Adapting the number of modulation levels in response to the anticipated local
SNR encountered in each subcarrier can be employed, in order to achieve a wide
range of different tradeoffs between the received data integrity and throughput.
Corrupted subcarriers can be excluded from data transmission and left blank or used
for example for Crest–factor reduction. A range of different algorithms for selecting
the appropriate modulation modes are investigated below. The adaptive channel
coding parameters entail code rate, adaptive interleaving and puncturing for
convolutional and turbo codes, or varying block lengths for block codes (Webb &
Hanzo 1994). Based on the estimated frequency–domain channel transfer function,
spectral predistortion at the transmitter of one or both communicating stations can be
invoked, in order to partially of fully counteract the frequency–selective fading of the
69
time dispersive channel. Unlike frequency-domain equalization at the receiver, which
corrects for the amplitude and phase errors inflicted upon the subcarriers by the
channel but cannot improve the SNR in poor quality channels spectral predistortion at
the OFDM transmitter can deliver near constant signal-to- noise levels for all
subcarriers and can be thought of as power control on a subcarrier-by-subcarrier basis.
In addition to improving the system’s SER performance in time dispersive
channels, spectral predistortion can be employed in order to perform all channel
estimation and equalization functions at only one of the two communicating duplex
stations. Low cost, low power consumption mobile stations can communicate with a
base station that performs the channel estimation and frequency domain equalization
of the uplink, and uses the estimated channel transfer function for pre–distorting the
downlink OFDM symbol. This setup would lead to different overall channel quality
on the uplink and downlink, and the superior downlink channel quality could be
exploited by using a computationally less complex channel decoder having weaker
error correction capabilities in the mobile station than in the base station. If the
channel’s frequency–domain transfer function is to be fully counteracted by the
spectral pre-distortion upon adapting the subcarrier power to the inverse of the
channel transfer function, then the output power of the transmitter can become
excessive, if heavily faded subcarriers are present in the system’s frequency range.
In order to limit the transmitter’s maximal output power, hybrid channel pre–
distortion and adaptive modulation schemes can be devised, which would deactivate
transmission in deeply faded subchannels, while retaining the benefits of pre-
distortion in the remaining subcarriers.
a. Signaling the Parameters
Signaling plays an important role in adaptive systems and the range of signaling
options is summarized in Figure 2.17 for both open loop and closed loop signaling, as
well as for blind detection. If the channel quality estimation and parameter adaptation
have been performed at the transmitter of a particular link, based on open loop
adaptation, then the resulting set of parameters has to be communicated to the receiver
70
in order to successfully demodulate and decode the OFDM symbol. If the receiver
itself determines the requested parameter set to be used by the remote transmitter the
closed loop scenario then the same amount of information has to be transported to the
remote transmitter in the reverse link. If this signaling information is corrupted, then
the receiver is generally unable to correctly decode the OFDM symbol corresponding
to the incorrect signaling information.
Unlike adaptive serial systems, which employ the same set of parameters for
all data symbols in a transmission packet (Torrance, et al 1999), adaptive OFDM
systems have to react to the frequency selective nature of the channel, by adapting the
modem parameters across the subcarriers. The resulting signaling overhead may
become significantly higher than that for serial modems, and can be prohibitive for
example for subcarrier-by-subcarrier modulation mode adaptation. In order to
overcome these limitations, efficient and reliable signaling techniques have to be
employed for practical implementation of adaptive OFDM modems. If some
flexibility in choosing the transmission parameters is sacrificed in an adaptation
scheme, like in the sub–band adaptive OFDM schemes described below, then the
amount of signaling can be reduced. Alternatively, blind parameter detection schemes
can be devised, which require little or no signaling information, respectively.
Actually, it is not fair to make any comparison between the tested modes in the
same models, because they are utilizing a different number of modulation schemes.
However it is essential to investigate the ability of each mode to provide the verified
OFDM system with the best performance that it needs. Moreover the comparison
between the two proposed models can give a good vision of suitability of them in
some OFDM based wireless systems in terms of SER, PAPR, and data throughput.
Table 5.34
provides a comprehensive comparison between the two proposed models
of selection policy in adaptive modulation.
Table 5.34 Comparison between the two proposed modulation selection policy models
Comparison Model (1) Model (2) Modulation selection policy
Based on SER Based on SNR and SER
CR updating mechanism Defined based on SNR Defined based on target SER
Freedom in CR definition Can define three different values of CR to meet the channel condition
Can define only one value of CR that maintain SER under target value
Utilization percentage AMCl utilizes Two modulation schemes or more at SNRs below 20 dB
AMCl utilizes one or two modulation scheme as maximum option at all SNRs.
SER performance Comparable to normal OFDM without adaptive modulation
Comparable to normal OFDM with adaptive modulation
SER performance Stability Non stable Stable
Data throughput High enhancement in the data throughput at all SNRs below 20 dB compare to normal OFDM with conventional AM
Less enhancement in the data throughput compared to the proposed model (1)
PAPR improvement High improvement compared to normal OFDM system
Low improvement compared to model (1)
Suitability to OFDM application
More suitable to 4G and DVB-T in terms of data throughput and PAPR reduction.
More suitable to IEEE 802.11g and IEEE 802.16e in terms of SER performance.
CHAPTER VI
CONCLUSIONS AND FUTURE WORK
6.1 INTRODUCTION
This chapter concludes the presentation of the work by resuming the achieved results,
and directions for further studies are also indicated.
6.1 CONCLUSIONS
In order to solve the problems in the selection policy of AM that are related to fixed
and low throughput at most SNRs, two models of modulation selection policy are
proposed namely model (1) and model (2). Each model employs number of modes
that utilize different number of modulation schemes. These models are combined with
the CR updating mechanism to introduce an algorithm called Adaptive Modulation
and Clipping (AMCl).
The results of the tested modes in model (1) show the ability of AMCl to
reduce the PAPR and enhance the data rate of the OFDM system at all SNRs. The
updating mechanism of CRs in AMCl shows great ability to offer suitable distribution
in utilization all available modulation schemes at all SNRs. This ability can be noticed
in the SER performance of the tested modes in model (1).
223
The SER performance at SNRs below 8 dB is better than the SER performance
of the normal (unclipped) OFDM with high order modulation schemes such as 64-
QAM, and 128-QAM. However, the SER performance of the tested modes in model
(1) is better than normal OFDM with 256-QAM at all SNRs. Moreover the SER
performance of these modes is superior to the SER performance of the clipped OFDM
system. AMCl algorithm in model (1) shows better improvement in the PAPR
compared to the normal OFDM system with different modulation schemes. Moreover
the proposed modulation selection policy in model (1) offers an enhancement in the
throughput of the OFDM system at all SNRs compared to the normal OFDM that
employs the conventional selection policy of AM. As a result of this increment in the
data rate of the verified systems, the required time to transfer these data can be
reduced dramatically at all SNRs.
Despite of clipping the OFDM samples, AMCl algorithm in model (2) offers
reasonable SER compared to the normal OFDM system. Because of AMCl defines the
values of CR based on the target SER, the SER performance is comparable to the
normal (unclipped) OFDM with conventional selection policy especially when
clipping OFDM signal at high CR. The tested modes in model (2) have superior SER
performance compared to the SER performance of the clipped OFDM system. The
effectiveness of the selection policy in the proposed model (2) over the conventional
policy can be noticed in the throughput of the OFDM system that is enhanced at all
SNRs. This increment in the data rate comes from using two modulation schemes
together with different utilization percentage at most SNRs. Definition an appropriate
value of CR based on target SER is essential in model (2) to define the degree of
enhancement in the data rate. Results shows that clipping OFDM samples at low CR
causes increment in the utilization percentage of low order modulation scheme and
vice versa. All tested modes in model (2) offers huge enhancement in data throughput
when clipping OFDM samples at high CRs, whereas these modes offer less PAPR
improvement. The SER performance of model (2) is better than the SER performance
of model (1). This is because of AMCl in model (2) utilizes only one or two adjacent
order of modulation schemes (as a maximum option) that meet the values of SNR and
SER, whereas in model (1) two schemes or more are utilized at SNRs below 20 dB
with different percentages depends on SER.
224
Moreover AMCl in model (1) employs updating mechanism of CR that defines
and updates the values of CR based on estimated SNR, whereas in model (2) constant
value of CR is defined based on the target SER. All this reasons make the SER
performance of model (2) better than model (1) and comparable to normal OFDM
with conventional selection policy. As a result of the excellent performance of SER in
the tested modes in model (2), and because of the reciprocal relationship between
PAPR and SER, the data rate enhancement in model (1) is better than the
enhancement in model (2) at most SNRs.
AMCl algorithm in both models offers improvement in the PAPR compared to
normal OFDM system. However the PAPR improvement in modes of model (1) is
better than model (2) because of AMCl defines CR in model (1) based on SNR,
whereas CR in model (2) is defined based on target SER. The dependence of
definition the CR on SER restricts using low CR in the tested modes of model (2),
whereas the updating mechanism of CR in modes of model (1) has more freedom to
select low, moderate, or high CRs of each modulation scheme depends on SNR. To
summarize the overall results, the proposed AMCl algorithm in models (1) and (2)
shows its ability to provide OFDM system with high data rate transmission at all
SNRs.
The improvement in PAPR is clearly noticed in all tested systems. Despite
these models uses clipping technique, they can offer comparable SER performance
compared to the normal OFDM system. The results prove suitability of AMCl
algorithm in two models to all verified OFDM systems in terms of SER, PAPR, and
data throughput. For more accurate explanation the tested modes in model (1) are
more suitable to OFDM system that needs better improvement in PAPR and huge data
throughput at all SNRs with reasonable SER performance at low SNRs such as DVB-
T and 4G systems. However OFDM system that needs better and stable SER
performance with reasonable reduction in PAPR and high data throughput at all SNRs
such as IEEE 802.11g and IEEE 802.16e, the tested modes in model (2) can be the
best choice.
225
6.5 FUTURE WORK
This thesis focuses only on the effects of joint performance of the modulation
adaptation and the PAPR clipping technique. It is essential in any upcoming work to
analyze the joint performance of rate adaptation and power adaptation with PAPR
reduction mechanisms in OFDM systems. The analysis should cover three possible
cases, the first case when the transmit power is constant with only rate adaptation, the
second one when the rate is constant with only power adaptation, and the last case
with both power and rate adaptation. It is recommended to obtain the cumulative
distribution numerically of the PAPR in each of the above cases. Moreover, it is
important to find an optimization for the algorithm of the link adaptation mechanism.
The envelope of OFDM signal has large fluctuation in the time domain. This
fluctuation becomes more worse when OFDM system utilizes high order modulation
schemes such as 64-QAM, 128-QAM, and 256-QAM. The current expression of
PAPR meets OFDM that utilize low order modulation schemes such as 4-QAM, and
16-QAM. Long observation in this thesis shows that the amplitude of peaks in
OFDM-256QAM signal is larger than other schemes. Moreover the measured value of
PAPR is larger than theoretical value. This leads to necessity of developing the
analytical expression of the PAPR.
REFERENCES
Agilent Technologies. 2000. Characterizing Digitally Modulated Signals with CCDF Curves. 3Thttp://www.home.agilent.com 3T. [13 November 2009].
Ahn, H., Shin, Y. M. & Im, S. May 2000. A block coding scheme for peak to average ratio reduction in an orthogonal frequency division multiplexing system. In Proc. 51st IEEE veh. Technol. Conf.-Spring 1: 56-60. Tokyo, Japan.
Al-Dhahir, N. & Cioffi, J. M. 1996. Optimum finite-length equalization for multicarrier transceivers. IEEE Transactions on Communications 44(1): 56- 64.
Alexander, T. 2007. Optimizing and Testing WLANs: Proven Techniques for Maximum Performance. ISBN 978-0-7506-7986-2. MA: Elsevier.
Alouini, M., Tang, X. & Goldsmith, A. 1999. An Adaptive Modulation Scheme for Simultaneous Voice and Data Transmission over Fading Channels. IEEE Journal on Sel. Areas in Comm. (JSAC) 17(5): 1T 837-850.1T
Antognetti, P. 1986. Power Integrated Circuits: Physics, Design and Applications,New York: McGraw-Hill.
Armstrong, J. 1999. Analysis of new and existing methods of reducing intercarrierinterference due to carrier frequency offset in OFDM. IEEE Transactions on
Communications 47(3): 365 – 369.
Bahai, A. R. S. & Burton R. Saltzberg. 1999. Multi-Carrier Digital Communications. Kluwer Academic/Plenum Publishers.
Baxley, R. J. & Zhou, G. T.. 2004. Power savings analysis of peak-to-average powerratio reduction in OFDM. IEEE Trans. Consumer Electron. 50: 792-798.
Bing Leung Patrick Cheung. 2002. Simulation of Adaptive Array Algorithms for OFDM and Adaptive Vector OFDM Systems. MA thesis, Virginia Polytechnic
Institute and State University.
Bonaccorso, M., Mhirsi, K., Maouche, K. & Buzenac, V. 1999. Reducing the Peak-to- Average Power Ratio in OFDM Systems. Gretsi: 781-784.
Catreux, S., Erceg, V., Gesbert, D. & Health, R.W. 2002. Adaptive Modulation and MIMO Coding for Broadband Wireless Data Networks. IEEE Commun Mag.:
108-115.
Chan, R. 2003. Channel Prediction for Adaptive Modulation in WirelessCommunications. MA thesis, Dept. of Electrical Engineering, Univ. ofVirginia Polytechnic and State, USA.
Chang, R.W. 1966. Synthesis of Band-Limited Orthogonal Signals for Multichannel Data Transmission. Journal of Bell Syst. Tech. 45: 1775-1796.
Chuang, J. & Sollenberger, N. 2000. Beyond 3G: Wideband Wireless Data AccessBased on OFDM and Dynamic Packet Assignment, IEEE Commun. Mag.
Chung, S. T. & Goldsmith, A. J. 2001. Degrees of Freedom in Adaptive Modulation: A Unified View. IEEE Trans. Commun. 49: 1561-1571.
Dukhyun, K. & Gordon L. S. 2007 “Clipping Noise Mitigation for OFDM by Decision-Aided Reconstruction,” IEEE Communications Letters. 3(1): 4-6.
DTG. 2002. Final Report of DTT (DVB-T) Transmission Mode Trials. http://www.dtg.org.uk/publications/books.html. [15 December 2009].
Du, K. L & Swamy, M. N. S. 2010, Wireless Communication Systems from RF subsystems to 4G Enabling Technologies. ISBN 978-0-521-11403-5. Cambridge: Cambridge University press.
Ergen, M. 2009. Mobile Broadband Including WiMAX and LTE. ISBN 978-0-387- 68189-4 New York: Springer Science.
ETSI. 2000. Digital Video Broadcasting (DVB).http://www.etsi.org/broadcast/dvb.htm. [15 October 2009].
ETSI. 2008. OFDM modulation for microwave digital terrestrial television.http://www.etsi.org. [20 May 2008].
ETSI EN 300 421. 1997. Framing Structure, channel coding and modulation for11/12 GHz satellite services. http://www.etsi.org. [18 August 2009].
Falahati, S., Svensson, A., Ekram, T. & Sternad, M. 2004. Adaptive Modulation Systems for Predected Wireless Channels. IEEE Trans. on Communications 52(2).
Flarion Technologies Inc. 2003. OFDM for Mobile Data Communications. http://www.qualcomm.com/innovation/research/papers.html. [12 may 2008].
Gross, R. & Veeneman, D. 1993. Clipping distortion in DMT ADSL systems, Electtronics Letters. 29(24): 2080-2081.
Han, S. H. & Lee, J. H. 2004. Peak-to-average power ratio reduction of an OFDM signal by signal set expansion. Proc. Paris IEEE Int. Conf. Commun.
2: 867-971. Hara, S. & Prasad, R. 2003. Multicarrier Techniques for 4G Mobile Communications.
ISBN 1-58053-482-1. 2. London: Artech House.
1THazy, L. 2004. Introduction to Orthogonal Frequency Division Multiplexing (OFDM).1T
Hirosaki, B. 1981. An orthogonally multiplexed QAM system using the discrete Fourier Transform. IEEE Trans. Comm. COM-29(7): 982-989.
Hoeher, P., Kaiser, S., & Robertson P. 1997. Two-dimensional Pilot-symbol- aided Channel Estimation by Wiener Filtering. IEEE International Conference on Acoustics, Speech, and Signal Processing 3: 1845-1848.
IEEE Std. 802.16e-2005 and IEEE Std. 802.16-2004/Cor 1-2005, (Amendment andCorrigendum to IEEE Std. 802.16-2004). 2006. Air interface for fixed andmobile broadband wireless access systems amendment 2: Physical andmedium access control layers for combined fixed and mobile operation inlicensed bands and corrigendum 1. IEEE standard for local and metropolitanarea networks part 16.
IEEE 802.11g-2003. 2007. Further higher data rate extension in the 2.4 GHz band http://standards.ieee.org/getieee802/download/802.11g-2003.pdf. [12 August 2009].
Jeon, W. G., et al. 2001. An equalization technique for orthogonal frequency divisionmultiplexing systems in time-variant multipath channels. IEEE Transactions on Communications 47(1): 27-32.
Jones, A. E., Wilkinson T. A. & Barton, S. K. 1994. Block Coding Scheme for Reduction of Peak to Mean Envelope Power Ratio of Multicarrier Transmission Schemes. Electronics Letters 30(22): 2098-2099.
Kim, H. J., Cho, S. C., Oh, H. S. & Ahn, J. M. 2003. Adaptive clipping technique for reducing PAPR on OFDM systems. VTC 2003-Fall 3: 1478-1481.
Kim, H. S., Kwak, J. S., Choi, J. M., & Lee, J. H. 2004. Efficient Subcarrier and Bit Allocation Algorithm for OFDM System with Adaptive Modulation. IEEE Vehicular Technology Conference 59: 1816-1820.
Krongold, B. S. 2003. New techniques for multicarrier communication systems. Ph.Dthesis, University of Illinois at Urbana-Champaign, Urbana, Illinois.
Krongold, B. S. & Jones, D. L. 2003. PAR reduction in OFDM via active constellation Extension. IEEE Trans. Broadcast. 49: 258-268.
Lee, G. W. 2008. The basic principles of OFDM. http://my.com.nthu.edu.tw/~jmwu/com5195/OFDM.pdf. [23 October 2009].
Lenart, T. & Owall, V. 2003. A 2048 complex point FFT processor using a novel data scaling approach. Circuits and Systems (ISCAS ’03): 45-48.
Proc. Bangkok IEEE International Symposium on
Li, X. & Cimini, L. J. 1998. Effects of clipping and filtering on the performance of OFDM. IEEE Commun. Lett. 2: 131-133.
Litwin, L. 2003. An introduction to multicarrier modulation Communications 19(2): 36-38.
, IEEE Transactions on
Lu, W. W. & Berezdivin, R. 2002. Commun. 9(2) : 104-105. Matic, D. 2001. OFDM Synchronization and Wideband Power Measurements at 60 GHz for Future Wireless Broadband Multimedia Communications. Ph.D thesis, Aalborg University, Denmark.
MATLABP
®P Help. 2008. Version 7.6.0.324 (R2008a), Licence Number: 161051.
1T Massachusetts: MathWorks Inc.
Mishra, A. R. 2010. Technologies for Emerging Markets: 2G, 3G, and Beyond. ISBN 9780470779477. Chichester: Wiley.
Mohr, W. 2002. Heterogeneous Networks to Support User Needs with Major
Technologies for 4G Mobile. IEEE Wireless
Challenges for New Wideband Access Systems. Wireless PersonalCommunications (Kluwer) 22(2): 109–137.
Moiser, R. R. and Calbaugh, R. G. 1958. Kineplex, a bandwidth efficient binarytransmission system. American Institute of Electrical Engineers76: 723-728.
(AIEE) Trans.
Moose, P. 1994. A Technique for Orthogonal Frequency Division Multiplexing Frequency Offset Correction. IEEE Transactions on Communications 42, (10): 2908-2914.
Muschallik, C. 1996. Improving an OFDM reception using an adaptive Nyquist Windowing. IEEE Transactions on Consumer Electronics 42(3): 259 – 269.
Muller, S. H., Bauml, R. W., Fischer, R. F. H., & Huber, J. B. 1997. OFDM with Reduced Peakto-Average Power Ratio by Multiple Signal Representation Annals of Telecommunications 52(1-2): 58-67.
Nanda S., Balachandran, K. & Kumar, S. 2000. Adaptation Techniques in Wireless Packet Data Services. IEEE Communications Magazine: 54 – 64.
Narahashi, S. & Nojima, T. 1997. A new phasing scheme for multitone signal systems to reduce peak-to-average power ratio. Electronics and Communications in Japan 80(1): 89-97.
Nassar, C. R. et al. 2002. Multi-carrier Technologies for Wireless Communication. Kluwer Academic Publishers.
NTT DoCoMo Advertisement. 2000. The Path to 4G Mobile. IEEE Communications Magazine: 38-41.
Ochiai, H. & Imai, H. 2002. Performance analysis of deliberately clipped OFDM signals. IEEE Transactions on Communication 50: 89-101.
Okamoto, E. , Li, H. & Ikegami, T. QAM Using FFT. IEEE Transactions on Vehicular Technology 48(5): 1079-
1991. Rayleigh Fading Compensation for 16
1082.
Oppenheim, A. V. & Schafer, R. W. 1989. Discrete -Time Signal Processing. New Jersy: Prentice Hall.
Peled, R. & Ruiz, A. 1980. Frequency Domain Data Transmission Using Reduced Computational Complexity Algorithms, Proc. of the Denver IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP’80: 964-967.
Pollet, T. et al. 1995. BER Sensitivity of OFDM Systems to Carrier Frequency Offset and Wiener Phase Noise. IEEE Trans. Comm. 43(2-3-4): 191-193.
Pons, J. & Dunlop, J. 1998. Bit Error Rate Link Adaptation for GSM. The Ninth IEEE International Symposium on Personal, Indoor and Mobile Radio Communications 3:1530 -1534.
Popovic, B. M. 1991. Synthesis of Power Efficient Multitone Signals with Flat Amplitude Spectrum. IEEE Transactions on Communications 39(7): 1031- 1033.
Porter, G. C. 1968. Error distribution and diversity performance of a frequency differential PSK HF modem. IEEE Trans. Comm. COM-16: 567-575.
Proakis, J. G. 1989. Digital Communications. ISBN 0070509379. 2nd edition. York: McGraw-Hill.
New
Proakis, J. G. & Salehi, M. 2000. Contemporary Communications Systemsusing MATLAB. CA: Brooks/Cole.
Quatech. 2009. Prallel communication overview. http://www.quatech.com/support/comm-over-parallel.php. [14 October 2009].
Reimers, U. 1998. Digital Video Broadcasting. IEEE Communications Magazine:104-110.
Robertson, P. and Kaiser, S. 1999. The Effects of Doppler Spreads in OFDM(A) Mobile Radio Systems, Proc. of IEEE VTC’99-Fall.
Rohling, H. et al. 1999. Broad-band OFDM Radio Transmission for Multimedia Applications. Proc. of the IEEE 87(10): 1778–1789.Rohling, H. & May, T. 1997. Comparison of PSK and DPSK Modulation in a Coded OFDM System. Proc. Arizona IEEE VTC: 5–7.
Salzberg, B. R. 1967. Performance of an efficient parallel data transmission system. IEEE Trans. Commun. Technol. COM-15: 805-813.
Sampei, S. & Sunaga, T. 1993. Rayleigh Fading Compensation for QAM in Land Mobile Radio Communications. IEEE Trans. on Veh. Technol. 42(2): 137-147.
Sari, H., Karam, G. & Jeanclaude, I. Terrestrial TV Broadcasting. IEEE Comm. Mag. 33(2): 100–109.
1995. Transmission Techniques for Digital
Sezginer, S. & Sari H. 2005. Peak power reduction in OFDM systems using dynamic constellation shaping. Proc. 13th Antalya Eur. Signal Process. Conf.
Shen, Z., Andrews, J. G., & Evans B. L. 2003. Optimal Power Allocation in Multiuser OFDM Systems. Proc. IEEE Asilomar Conf. on Signals, Systems, and Computers 1: 1147-1151.
Stallings, W. 2007. Data and Computer Communications. ISBN 0-13-243310-9. 8. New Jersey: Prentice Hall.
Steendam, H. & Moeneclaey, M. 1999. Analysis and Optimization of the Performance of OFDM on Frequency-Selective Time-Selective Fading Channels. IEEE Transactions on Communications 47(12): 1811-1819.
Sesia, S., Toufik, I., & Baker, M. 2009, LTE The UMTS Long Term Evolution From Theory to Practice. ISBN 9780470697160. Chechester: Wiley.
Suraweera, H. A. & Armstron, J. 2004. Noise bucket effect for impulse noise inOFDM. Electronics Letters 40(18): 3536-3540
Tarokh, V. 2009. New Directions in Wireless communications Reserch. ISBN 978- 14419-0672-4 Cambridge MA: Springer.
Tang, C. & Stolpman, V. J. 2004. Multiple users Adaptive Modulation Schemes for MC-CDMA. Nokia (7th August 2004).
Tellado, J. 2000. Multicarrier Modulation with low PAR: Applications to DSL and Wireless. Massachusetts: Kluwer Academic Publishers.
Tellado, J & Cioffi, J. M. 1997. PAR Reduction in Multicarrier Transmission Systems, T1E1.4 contribution number 1997-367, Szcramento, CA.
Tellambura, C. 2001. Computation of continuous time PAR of an OFDM signal with BPSK subcarriers. IEEE Commun. Letters 5: 185-187.
Tjelta, T., Nordbotten, A., Annoni, M, Scarrone, E., Bizzarri, S., Tokarchuk, L., Bigham, J., Adams, C., Craig, K. & Dinis, M. 2001. Future Broadband Radio Access Systems for Integrated Services with Flexible Resource Management. IEEE Communications Magazine: 56-63.
Torrance, J. M. & Hanzo, L. 1996. Upper bound performance of adaptive modulationin a slow Rayleigh fading channel,” IEEE Electronics Letters 32: 718.
Torrance, J. M., Hanzo, L. & Keller, T. 1999. Interference aspects of adaptivemodems over slow rayleigh fading channels. IEEE Tr. on Veh. Techn. 48:1527-1545.
Channels.
Uysal, M, Al-Dhahir, N & Georghiades, C. N. 2006. A Space-Time Block-Coded OFDM Scheme for Unknown Frequency-Selective Fading
IEEE Communications Letters, 5(10): 393-396.
Van Nee, R. 1996. and Error Correction.
OFDM Codes for Peak-to-Average Power Reduction Proc. London IEEE GLOBECOM'96 1: 740-744.
Norwood, MA: Artech House. Van Nee, R. & Prasad, R. 2000. OFDM for Wireless Multimedia Communications,
Wang, H., Kondi, L., Luthra, A. & Ci, S. 2009. 4G Wireless Video Communications, Chichester, UK: Wiley.
Wang, J., Shan, X. & Ren, Y. 2005. A New Approach for Evaluating Clipping Distortion in DS-CDMA Systems. IEICE Transactions on Communications. E88-B2.
Wang, L. & Tellambura, C. 2005. A simplified clipping and filtering technique for PAR reduction in OFDM systems. IEEE Signal Process. Letters. 12: 453-456.
Wang, Z. & Giannakis, G. B. 2000. Wireless Multicarrier Communications. IEEESignal Processing Magazine: 29-48.
Watkinson, J. 2007. The art of digital audio. 3rd edition, ISBN 0-240-51587-0. MA: Elsevier.
Webb, W. T. & Hanzo, L. 1994. Modern Quadrature Amplitude Modulation: Principles and Applications for Wireless Communications. ISBN 0-7273- 1701-6. IEEE Press-Pentech Press.
Weinstein, S. B. & Ebert, P. M. 1971. Data Transmission by Frequency Division Multiplexing using the Discrete Fourier Transform. IEEE Trans. on Commns. 19: 628-643.
WiMAX forum. 2006. Mobile WiMAX -Part I: A Technical Overview and Performance Evaluation.
Wong, C. Y., Cheng, R. S. 1999. Multiuser OFDM with adaptive Subcarrier, bit and power Allocation. IEEE Journal on Selected Areas in Communications 1747-1758.
17(10):
Wong, C. Y., Cheng, R. S, Letaief, K. B. & Murch, R. D. 1999. Multiuser Subcarrier Allocation for OFDM Transmission using Adaptive Modulation. IEEE Journal
on Selected Areas in Communications (Special Issue on Broadband Wireless Techniques)
Yagoobi, H. 2004. Scalable OFDMA Physical Layer in IEEE 802.16 WirelessMAN”,
.
Intel Technology Journal. 8: 201-212.
Yu, H. & Wei, G. 2003. Computation of the continuous time PAR of an OFDM signal..Proc. Hong Kong IEEE Int. Conf. Acoust., Speech, Signal Process.: 529-531.
Yunjun, Z., Yongacoglu, A., & Chouinard, J. Y. 2000. Othogonal Frequency Division Multiple Access Peak-to-Average Power Ratio Reduction Using Optimized Pilot Symbols. Proc. Beijing ICCT 2000: 574-577.
Yunsong, L., Yong, H. and Xianci, X. 1999. approach for arbitrary bandpass signal within baseband sampling rate. Journal
A new quadrature sampling
of electronics. 16(4): 305-310.
Zhang, G. 2004. Subcarrier and Bit Allocation for Real time Services in Multiuser OFDM Systems. Proc Paris IEEE International Conference on Communications. 5:2985-2989.
Zhao, Y. & Haggman, S. 2001. Intercarrier interference self-cancellation scheme for OFDM mobile communication systems. IEEE Transactions on Communications 49(7): 1185 – 1191.
Zimmerman, M. S. & Kirsch, A. L. 1967. The AN/GSC-10 (KATHRYN) variable rate data modem for HF radio. IEEE Trans. Comm. 15: 197-205.
233
APPENDICES
APPENDIX A
LIST OF PUBLICATIONS
JOURNALS
Ibrahim Ismail Al-kebsi, Mahamod Ismail, and Kasmiran Jumari. 2008. The Impact of Modulation Adaptation on PAPR Clipping Technique in OFDM of 4G System. Asian Network for Scientific Information Science (JAS). 8(15): 2776-2780.
(ANSI), Journal of Applied
Ibrahim Ismail Al-kebsi, Mahamod Ismail, Kasmiran Jumari, and T. A. Rahman. 2009. Mobile WiMAX Performance Improvement Using a Novel Algorithm with New Form of Adaptive Modulation. International Journal of Computer Science and Network Security
Ibrahim Ismail Al-kebsi, Mahamod Ismail, Kasmiran Jumari, and T. A. Rahman. 2009. Throughput Enhancement and Performance Improvement of the OFDM Based WLAN System.
(IJCSNS). 9(2):76-82.
International Journal of Computer Science and Network Security
Ibrahim Ismail Al-kebsi, Mahamod Ismail, Kasmiran Jumari, and T. A. Rahman. 2009. Eliminate the Effects of Clipping Technique on the SER performance by Recovering the Clipped Part of the OFDM Signal.
(IJCSNS). 9(4):138-148.
International Journal of Computer Science and Network Security
(IJCSNS). 9(7):37-45.
PROCEEDINGS
Ibrahim Ismail Al-kebsi, Mahamod Ismail and Kasmiran Jumari. 2008. The Impact of Adaptive Modulation and Power Control on Peak to Average Power Ratio in OFDM of 4G system. Proc. IEEE of 6th National Conference on Telecommunication Technologies and 2nd Malaysia Conference on Photonics(NCTT-MCP). PutraJaya, Malaysia. 2: 295-299.
Ibrahim Ismail Al-kebsi, Mahamod Ismail. 2008. The Impact of Link Adaptation and Power Control on PAPR Clipping Technique in OFDM system. Proc. of Engineering Postgraduate Conference (EPC). Kajang, Malaysia. 1:17-21.
234
Tharek A. Rahman, Mahamod Ismail, and Ibrahim Ismail Al-kebsi. 2008. The Impact of PAPR and ICI on the Performance of the Mobile WiMAX OFDMA System. Proc. of Malaysian Communications and Multimedia Commission (MCMC) colloquium. 18-19 ISBN: 978-983-42563-2-6.
Ibrahim Ismail Al-kebsi, Mahamod Ismail, Kasmiran Jumari, T. A. Rahman, and Ayman A. El-Saleh. 2009. A Novel Algorithm with a New Adaptive Modulation Form to Improve the Performance of OFDM for 4G Systems. Proc. IEEE of the International Conference on Future Computer and Communication (ICFCC), Kuala Lumpur, Malaysia. 1: 11-16.
Ayman A. El-Saleh, Mahamod Ismail, Mohd Alauddin Mohd Ali, and Ibrahim Ismail Al-kebsi. 2009. Capacity Optimization for Local and Cooperative Spectrum Sensing in Cognitive Radio Networks. Proc. IEEE of the International Conference on Future Computer and Communication (ICFCC). Kuala- Lumpur, Malaysia. 1:145-150.