UNIVERSITY OF JUBA

COLLEGE OF ENGINEERING AND ARCHITECTURE

DEPARTMENT OF ELECTRICAL ENGINEERING

(TELECOMMUNICATIONS)

OPTICAL COMMUNICATION SYSTEM PERFORMANCE OVER WIRELESS CHANNELS

BY

AGNES PAUL PETER

DENG ABENEGO CHANY

GUM THOMAS AJAK

SUPERVISOR

DR.JAMES OSURU MARK

A DISSERTATION SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENT
FOR THE AWARD OF BACHELOR OF SCIENCE (HONOURS) DEGREE IN ELECTRICAL
ENGINEERING (TELECOMMUNICATIONS).

DECEMBER, 2018
 Declaration
We declare that we have authored this thesis independently, that we have not used other than the
declared sources / resources, and that we have explicitly marked all materials which have been
quoted either literally or by content from the used sources.

By

Name Signature Date

Agnes Paul Peter _______________ ______________
Deng Abenego Chany ________________ ______________
Gum Thomas Ajak ________________ ______________

Supervisor
Name Signature Date
Dr. James Osuru Mark _______________ _______________

i
 Acknowledgement

First of all we would like to thank God for bringing us this far. We would be nowhere without
him. Wish to express our sincere appreciation to Dr. James Osuru Mark for his guidance,
advice, interest, encouragement and assistant in developing of this project.
Our most heartfelt appreciation goes to the staffs, at the University of Juba, College of
Engineering and Architecture for their stand on tasks with the challenges facing us.
We also registered our votes of thanks to every person who contributed to the making of this
project.
We would like to take this opportunity to give special thanks to all our classmates, your
continual support and trust in our abilities has not gone unnoticed.

ii
 DEDICATION
This project is dedicated to our loving parents, families, friends, colleagues,
lecturers who have been supporting us, your love; care throughout our life means
the world to us.

iii
 ABSTRACT

Investigations into the possible applications of Radio Frequency (RF) modulation and coding
techniques on Optical Wireless (OW) channels are important since they can enhance link
performance and signal transmission quality. This work performs the study of such applications
using the simulation framework developed as an essential part of it. In addition to the often used
MATLAB, it provides an efficient tool for researchers. Results of the simulations using the
matlab and c++ are in accordance with those reported in prior research in the field of optical
wireless. These results are presented in different a format, textual and graphical, which make
them easy to compare and study. The simulation consists of modules that implement different
functionalities of the communication channel, such as number generators, checksum calculators,
modulators/demodulators and coding schemes. Modulation schemes that are studied using this
simulator are optical On-Off Keying(OOK), Pulse-position Modulation (PPM), Binary Phase
Shift Keying (BPSK), Quadrature Phase Shift Keying (QPSK), M-ary Pulse Amplitude
Modulation (M-PAM), M-ary Quadrature Amplitude Modulation (M-QAM) and they can be
combined with channel coding techniques such as Bose Chaudhuri Hocquenghem (BCH) codes,
Reed-Solomon (RS) and Low Density Parity Check(LDPC) codes.

iv
 LIST OF ABBREVIATION / ACRYNOMS
3G….. ………………………………………………………….3rd generation
4G………………………………………………………………4th generation
AIM………………………………………………..Analogue intensity modulation
ALI…………………………………………………..Ambient light interference
AM……………………………………………………… Amplitude modulation
ANN…………………………………………………Artificial neural network
ANSI…………………………………….American National Standards Institute
APD……………………………………………Avalanche photodiode detector
ASK…………………………………………………….Amplitude shift keying
AWGN………………………………………….Additive white Gaussian noise
BER……………………………………………………….Bit error rate
BLL……………………………………………………….Beer–Lambert law
BPSK………………………………………………Binary phase-shift keying
CDMA…………………………………………Code division multiple access
CWT……………………………………………...Continuous wavelet transform
DC………………………………………………………. Direct current
DD………………………………………………………. Direct detection
DH…………………………………………… PIM Dual-header pulse interval
DVB…………………………………………………Digital video broadcast
DWT…………………………………………….. Discrete wavelet transform
EV…………………………………………………..Electron volt
FEC…………………………………………………Forward error correction
FLI……………………………………………..Fluorescent light interference
FOV……………………………………………………. Field of view
GaAs……………………………………………..Gallium arsenide
Ge……………………………………....................Germanium
HDD………………………………………………Hard decision decoding
HPF………………………………………………High-pass filter
IM………………………………………………..Intensity modulation
v
IR……………………………………………….. Infrared
LAN…………………………………………….. Local area networks
LD………………………………………………..Laser diode
LDPC…………………………………………….Low-density parity check
LED……………………………………………... Light-emitting diode
LOS………………………………………………Line of sight
LPF………………………………………………Low-pass filter
MIMO……………………………………….Multiple input multiple output
MISO………………………………………….Multiple input single output
ML…………………………………………….Maximum likelihood
NRZ………………………………………….Non-return to zero
OBPF………………………………………….Optical band pass filter
OFDM………………………Orthogonal Frequency Division Multiplexing
OOK……………………………………..On and off keying
OPP……………………………………..Optical power penalty
OWC…………………………………….Optical wireless communications
PAM……………………………………… Pulse amplitude modulation
PD…………………………………………Photodetector
PDF………………………………………..Probability density function
PER………………………………………Packet error rate
PPM……………………………………….Pulse position modulation
PSD…………………………………….. Power spectral density
PSK……………………………………. Phase shift keying
QAM………………………………….Quadrature amplitude modulation
QPSK……………………………………….Quadrature phase shift keying
RF…………………………………………..Radio frequency
RZ…………………………………………..Return to zero
SER…………………………………………Slot error rate
Si ……………………………………………Silicon
SNR…………………………………………Signal-to-noise ratio
UV…………………………………………Ultraviolet

vi
 TABLE OF CONTENTS
Declaration ....................................................................................................................................... i
Acknowledgement .......................................................................................................................... ii
DEDICATION ............................................................................................................................... iii
ABSTRACT ................................................................................................................................... iv
LIST OF ABBREVIATION / ACRYNOMS ................................................................................. v
TABLE OF CONTENTS .............................................................................................................. vii
LIST OF FIGURES ....................................................................................................................... ix
LIST OF TABLES .......................................................................................................................... x
Chapter One .................................................................................................................................. 1
INTRODUCTION ..................................................................................................................... 1
1.1 Background ........................................................................................................................... 2
1.2 Problems statements .............................................................................................................. 4
1.3 Main Objectives of Study ...................................................................................................... 4
1.4 Specific Objectives of Study. ................................................................................................ 4
1.4 Scope of work........................................................................................................................ 4
1.5 Dissertation Organization ...................................................................................................... 5
Chapter Two .................................................................................................................................. 6
LITERATURE REVIEW ............................................................................................................ 6
2.1 Optical wireless Channels .................................................................................................... 6
Figure 2.1: Optical channel block diagram ................................................................................. 6
2.1.1 INDOOR OPTICAL WIRELESS CHANNEL MODELS ................................................ 7
2.1.2 OUTDOOR OPTICAL WIRELESS CHANNEL MODELS ............................................ 9
2.2. Optical Sources and Detectors ........................................................................................... 10
2.2.1. Light sources ................................................................................................................. 10
2.2.3 LASER (Light Amplification Simulation Emission and Radiation) ................................ 11
2.2.3 LED (Light Emitting Diode) ............................................................................................ 12
Figure 2.2: Electromagnetic Spectrum ...................................................................................... 13
Table 1. Fiber Optic Transmission Windows............................................................................ 14
22.4 Photodetectors ................................................................................................................... 14
2.2.6 PIN PHOTODETECTOR : .............................................................................................. 15
Chapter Three ............................................................................................................................. 17
vii
SYSTEM MODEL AND ANALYSIS ....................................................................................... 17
3.1 Channel Modeling ............................................................................................................... 17
3.2.1 On-Off Keying ................................................................................................................. 20
3.2.3 M-ary Phase Shift Keying ................................................................................................ 23
3.2.4. Binary Phase Shift Keying ............................................................................................. 23
3.2.5. Quadrature Phase Shift Keying ....................................................................................... 25
3.2.6. M-ary Pulse Amplitude Modulation .............................................................................. 25
3.2.7. 4-PAM ............................................................................................................................ 25
3.2.8. Multi-Level Quadrature Amplitude Modulation (M-QAM) .......................................... 26
3.3.0. Comparison of modulation techniques ............................................................................ 27
3.3 Channel coding .................................................................................................................... 28
3.3.1 Bose, Chaudhuri and Hocquenghem (BCH) codes .......................................................... 29
3.3.2 Reed-Solomon (RS) block codes ..................................................................................... 30
2.3.3 Low-Density Parity-Check (LDPC) ................................................................................. 31
3.1.4. Bit error rate analysis ...................................................................................................... 32
3.1.5 Performance Analysis of OWC Modulation over Wireless ............................................. 34
Chapter Four ............................................................................................................................... 39
RESULTS AND DISCUSSIONS ............................................................................................... 39
4.1. Additive White Gaussian Noise (AWGN) ......................................................................... 39
4.1.1. Visualization in Matlab ................................................................................................... 40
4.1.2. Modulation schemes unclouded ...................................................................................... 41
4.1.1 Comparison between RF and optical ............................................................................... 41
4.1.2. Modulation schemes coded ............................................................................................. 43
Chapter Five ................................................................................................................................ 44
CONCLUSIONS AND RECOMMENDATIONS .................................................................... 44
5.1. CONCLUSIONS ................................................................................................................ 44
5.2. RECOMMENDATIONS ................................................................................................... 44
APPENDIX ................................................................................................................................... 46
REFERENCES. ............................................................................................................................ 48

viii
LIST OF FIGURES
2.1 Optical channel block diagram . . . . . . . . . . . . . . . . . . . . . . . . . . ………………….. ………6
2.2. Electromagnetic Spectrum…………………………………………………………………..13
2.3. PIN Photodetector Schematic diagram……………………………………………………...15

3.1. Optical Wireless Communication Channel………………………………………………….18

3.2 Random OOK signal [9] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . …………………………….21
3.3 Random PPM signal [9] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ………………… ………….22
3.4 BPSK constellation diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . ………………… …………24
3.5 QPSK constellation diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . …………………………...25
3.6 Random PAM signal [9] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . …………………………….26
3.7 Optical power gain over OOK vs bandwidth efficiency [13] . . . . . . . . . . . . ……………….28
3.8. BER vs SNR plot………………………………………………………………....................37

4.1.BER of OOK vs M-PAM modulation …………………………………………………….41

4.2. BER of OOK vs M-PAM modulation………………………………………………………42
4.3. BER of Multilevel PPM modulation…………………………………………………..........43

ix
 LIST OF TABLES

Table 1. Fiber optic transmission window………………………………………………………14

x
 Chapter One
INTRODUCTION
Optical wireless communications represent one of the most promising approaches for addressing
the increasing demand in the global broadband access market. Current popular broadband access
technologies include cable and radio-frequency links. Cable offers reliable weather-independent
links, although requires point-to-point infrastructure (cable from the company to house) which
results in high "last-mile" costs. Radio frequency (RF) links on the other hand can be mildly
affected by weather, and they do not require point-to-point infrastructure [2]. The issue with
current RF technology is the already high-demand in RF spectrum and license requirements that
are highly priced. Commonly referred to as optical wireless (OW), Free Space Optics (FSO)
offers low start-up and operational costs, high security, rapid deployment, high fiber-like
bandwidths, and doesn’t require licenses. The current drawback of these FSO links is that they
can be severely affected by weather conditions such as fog, clouds and atmospheric turbulence
[3]. To improve the performance of optical wireless links further, we investigate different
channel modeling techniques. More specifically, this work looks at performance that can be
achieved by deploying various modulation techniques together with channel coding techniques
on the transmission channel. These modulation techniques, aim to correction of possible errors
that can occur during the transmission. A significant amount of theoretical research and testing
of modulations with hardware implementation has already been done at the Institute of
Broadband Communications at the TU Graz. Techniques for signal modulations such as the
OOK (On-Off Keying) modulation and the PPM (Pulse Position Modulation) have been
thoroughly investigated and publish in different optical journals within the FSO links research,
such as [1],[5], and much more. However, limited work has been done regarding the possible
application of RF modulation schemes on optical wireless links and no simulator has been
developed in Matlab that can test these schemes.
In this work, we investigate signal processing techniques which are predominantly used in RF
communications and their possible deployment to FSO links, i.e. optical wireless channels. More
specifically, we look at techniques such as PSK (Phase Shift Keying) - BPSK (Binary PSK),
QPSK (Quadrature PSK), and M-PSK (M-ary PSK), followed by QAM (Quadrature Amplitude
Modulation) and PAM (Pulse Amplitude Modulation).

1
Optical wireless channels present many unique constraints not present in RF communications.
These are discussed in detail in further chapters. Therefore, typical signal processing techniques
from RF communications cannot be applied directly to the optical channel; they need to be
adapted to the optical channel requirements and used only in that form. Also only certain
combinations of modulation and coding techniques make the channel more reliable and resilient
against channel faults. Within this project the efficient solutions will be suggested and evaluated.
These can further be implemented and tested with corresponding hardware design.

1.1 Background
Optical communication systems date back to the 1790s, to the optical semaphore telegraph
invented by French inventor Claude Chappe. In 1880, Alexander Graham Bell patented an
optical telephone system, which he called the Photophone. However, his earlier invention, the
telephone, was more practical and took tangible shape. The Photophone remained an
experimental invention and never materialized. During the 1920s, John Logie Baird in England
and Clarence W. Hansell in the United States patented the idea of using arrays of hollow pipes or
transparent rods to transmit images for television or facsimile systems.

In 1954, Dutch scientist Abraham Van Heel and British scientist Harold H. Hopkins

separately wrote papers on imaging bundles. Hopkins reported on imaging bundles of unclad
fibers, whereas Van Heel reported on simple bundles of clad fibers. Van Heel covered a bare
fiber with a transparent cladding of a lower refractive index. This protected the fiber reflection
surface from outside distortion and greatly reduced interference between fibers.

Abraham Van Heel is also notable for another contribution. Stimulated by a conversation with
the American optical physicist Brian O'Brien, Van Heel made the crucial innovation of cladding
fiber-optic cables. All earlier fibers developed were bare and lacked any form of cladding, with
total internal reflection occurring at a glass-air interface. Abraham Van Heel covered a bare fiber
or glass or plastic with a transparent cladding of lower refractive index. This protected the total
reflection surface from contamination and greatly reduced cross talk between fibers. By 1960,
glass-clad fibers had attenuation of about 1 decibel (dB) per meter, fine for medical imaging, but
much too high for communications. In 1961, Elias Snitzer of American Optical published a
theoretical description of a fiber with a core so small it could carry light with only one
waveguide mode. Snitzer's proposal was acceptable for a medical instrument looking inside the

2
human, but the fiber had a light loss of 1 dB per meter. Communication devices needed to
operate over much longer distances and required a light loss of no more than 10 or 20 dB per
kilometer.

By 1964, a critical and theoretical specification was identified by Dr. Charles K. Kao for long-
range communication devices, the 10 or 20 dB of light loss per kilometer standard. Dr. Kao also
illustrated the need for a purer form of glass to help reduce light loss.

In the summer of 1970, one team of researchers began experimenting with fused silica, a
material capable of extreme purity with a high melting point and a low refractive index. Corning
Glass researchers Robert Maurer, Donald Keck, and Peter Schultz invented fiber-optic wire or
"optical waveguide fibers" (patent no. 3,711,262), which was capable of carrying 65,000 times
more information than copper wire, through which information carried by a pattern of light
waves could be decoded at a destination even a thousand miles away. The team had solved the
decibel-loss problem presented by Dr. Kao. The team had developed an SMF with loss of 17
dB/km at 633 nm by doping titanium into the fiber core. By June of 1972, Robert Maurer,
Donald Keck, and Peter Schultz invented multimode germanium-doped fiber with a loss of 4 dB
per kilometer and much greater strength than titanium-doped fiber. By 1973, John MacChesney
developed a modified chemical vapor-deposition process for fiber manufacture at Bell Labs. This
process spearheaded the commercial manufacture of fiber-optic cable.

In April 1977, General Telephone and Electronics tested and deployed the world's first live
telephone traffic through a fiber-optic system running at 6 Mbps, in Long Beach, California.
They were soon followed by Bell in May 1977, with an optical telephone communication system
installed in the downtown Chicago area, covering a distance of 1.5 miles (2.4 kilometers). Each
optical-fiber pair carried the equivalent of 672 voice channels and was equivalent to a DS3
circuit. Today more than 80 percent of the world's long-distance voice and data traffic is carried
over optical-fiber cables.

3
1.2 Problems statements
Since the use of wireless channel in the usual process of sending large information from one
point to another at high cost bring inconvenience to the customers, there is a need for fiber optic
communication for larger information carrying capacity at low cost that is less space and easy
installation and at the same time accurate way. This device must be accurate, user friendly, safe,
and easy to maintain. Long range transmission system requirements needed in analyzing links
are; the maximum (or possible) transmission distance, the data rate or channel bandwidth and the
bit rate error (BER). Based on these requirements, optical communication systems are designed
to ensure that the desired performance level can be maintained over the expected system lifetime
without over-specifying the component characteristics.

1.3 Main Objectives of Study

The main goal of this project is to determine an Optical communication performance system over
wireless channels.

1.4 Specific Objectives of Study.

The objectives of this research project are:
 To identify a suitable OWC system by determining suitable modulation
 To determine the level of bit rate and maximum achievable bit rate that supports certain
level of bit error rate performance.

1.4 Scope of work

The main incentive for this project is the investigation of using different modulation and coding
techniques, predominantly used in RF communications, in optical wireless communications, and
also to build a software simulator that will be easy to use and extend for further research and
investigation. The primary evaluation criteria is the channel error performance under different
combinations of modulation and coding techniques. In addition, we discuss the advantages and
disadvantages of the usage of those techniques in various scenarios.
The optical channel simulation is built in a modular manner which enables the implementation of
channel blocks or modules as separate entities. More specifically, it consist of separate, mostly
independent modules developed as a part of the communication channel, that are connected over
the common interface, whereby the output vector of the each module becomes the input vector of
the next, adjacent module in the order as they are listed. Due to this design, it is also easily
expandable and results in additional modules and/or functionalities being easy to add.

4
The result of the simulation represents BER (Bit Error Rate) values for the given SNR (Signal To
Ratio) values together with the Block error rates on the channel. This method of evaluating
results is chosen, as it is a standard method of evaluating error patterns and link performance of
optical wireless channel in order to research different signal processing methods for a better
signal reception. They are presented in the form of textual files and console output, and also
plotted using Matlab.

1.5 Dissertation Organization

After studying the basic introduction, literature review, we define the objectives in chapter II. In
chapter III, we investigate the optical communication system performance over wireless channel
in term of bit error rate (BER). In chapter IV, we practically investigate and validate the
performances of optical communication over wireless channels. We finally discuss conclusions
in chapter V and also the future work.

5
 Chapter Two

LITERATURE REVIEW

2.1 Optical wireless Channels

Communication channels are typically presented in a simplified manner using a block diagram.
Each block is a module or building block, which in most cases corresponds to a mathematical
model. For an optical communication channel, such a block diagram is presented in figure 2.1. It
consists of a transmission segment, transmission channel and a reception segment [1]. This is a
simplified presentation of the channel, as it can generally contain more or fewer blocks,
depending on the signal processing methods deployed in a specific case.
The transmission part of the channel consists of a random generator that generates statistically
independent numbers, followed by a channel encoder for the Forward Error Correction (FEC)
algorithms [5]. A modulator then modulates the bits onto the carrier signal, to which electrical to
optical converter is connected. Hence, until electro-optical conversion module, all transmission
modules are equivalent or similar to those in a RF channel.
The actual transmitter of the optical channel is a light source, usually a Light Emitting Diode
(LED) or a Laser Diode (LD). It sends out optical signals through the channel that will be
received at the reception segment. The channel itself introduces certain impairments and can be
modeled as an Additive White Gaussian Noise (AWGN) channel (as typical in communication
channel simulations), or as a Fading channel. Both of these will be described in the following
sections.

Figure 2.1: Optical channel block diagram

6
The reception segment contains modules in the opposite order from the transmission segment.
This begins with the light detector, which is in most cases is a Photo-diode (PD), followed by the
optical to electrical converter - together, they constitute the reception part of the optical channel.
Next in the sequence are a channel demodulator for demodulation of modulated signals and
decoder. In case an error correction method is used on the channel, an error detector module and
bit error rate calculation module are present. Therefore in a simple presentation of an optical
wireless channel the signal is electrically processed, converted to light, transmitted through the
wireless optical channel, received by a light detector, and finally converted back to an electrical
signal for further processing [3].
Within this project, both channel coding pair and modulation pair from the block diagram 2.1 are
included in the simulation, whereby the electric to optical (E/O) converter is implemented as a
part of the modulator module and optical to electric (O/E) converter as a part of the demodulator
module. The error detector block takes the bit sequence coming out of the decoder and compares
it to the transmitted bit sequence, which is the output of the random generator block in the
transmission segment. It then calculates the Bit Error Rate, which is number of different bits over
number of bits in the block. It also calculates the block error rate, by dividing the BER with the
number of blocks.
Different BERs are simulated for different channel characteristics and simulation settings; for
example, these can be only modulation without the coding block, or coding with some simple
form of modulation.

2.1.1 INDOOR OPTICAL WIRELESS CHANNEL MODELS

The wireless optical channel in an indoor environment was first described in [1] as
broadcast/multi-access channel with diffused optical radiation at 950nm wavelength and low-to-
medium transmission speed covering a distance of up to 50m [2]. Within the indoor
environment, various reflecting surfaces are modelled to be Lambertian reflectors for the first
time. The physical parameters of the channel are represented by reflection properties of the
materials of the indoor reflecting surfaces, overall system bandwidth and bandwidth limitation
due to multipath propagation, ambient noise interference and range and distribution of the optical
radiation field. The upper limit of transmission bandwidth of the system due to multipath is

7
calculated to be 260Mbit.m/s while accounting for the ambient background light noise, the
transmission speed reduces to 1Mbps. Also, the experimental indoor infrared links are
demonstrated at 125kbps for baseband PCM modulation and 64kbps for PSK.
The simulations in the model above are actually meant for a link budget analysis and therefore
they account only for first order reflections at the most to estimate time integral of the channel
impulse response i.e. the total power arriving at the receiver.
A more accurate impulse response analysis is presented in [2] accounting for any number of
reflections. Due to significant contribution of temporal spread of the transmission signal to
performance limitation in high-speed applications, this approach models multipath dispersion
with reasonable accuracy and serves by far as the most popular benchmark model. The indoor
optical wireless environment is divided into infinitesimal cells which act as point sources and
receivers accounting for multiple reflections of the transmitted signal. The impulse response for
the kth reflection is calculated as the convolution of the LOS impulse response and the impulse
response due to (k-1)th reflection. The computation time is roughly proportional to Nk, where k
is number of reflections and N is the number of elements in which the indoor optical wireless
environment is discretised into. As a result, for reflections of the order higher than three, this
model proves to be prohibitively expensive in terms of system memory requirements and total
computation time.
A different approach to modelling indoor optical wireless environment is presented in [3],
whereby it is claimed that realistic multipath infrared channels can be completely characterized
by two parameters, namely the optical path loss and the root mean square (RMS) delay spread.
The so called ceiling bounce model translates the effect of multiple reflections into loss of
optical power at the receiver and the RMS delay spread of the signal arriving at the receiver after
multiple reflections. It provides a comparatively less complex way to predict power requirements
for different modulation schemes given the dimensions of indoor optical wireless environment
and the locations of the receiver and the transmitter. However, this model does not account for
the presence of obstructions is not expected to be highly accurate.
In a later work [4], an iterative site based model was developed extending the multipath infrared
channel characterisation idea previously described above to estimate the impulse response of the
indoor wireless infrared channel. Instead of calculating the impulse response as a recursion as in
[2], the model uses the simply the addition of appropriately scaled and shifted versions of the (k-
1)th bounce impulse response to calculate the kth bounce impulse response. This method can

8
include any number of reflections and can account for all types of obstructions and shadowing
effects within the indoor environment. It is computationally more effective and claims to be 90
times faster than the recursive model in [2] for estimating the impulse response for reflections of
order three.

2.1.2 OUTDOOR OPTICAL WIRELESS CHANNEL MODELS

Outdoor optical wireless channels are, in most cases, LOS links and therefore free from
multipath propagation considerations. However, various environmental factors present a
considerable challenge to outdoor optical wireless transmission and therefore have to be
modelled accurately for calculating the channel impulse response. Atmospheric turbulence is the
main performance limiting factor which causes fluctuations in both the intensity and the phase of
the received infrared signal.
It is suggested in [8] that the atmospheric turbulence in an outdoor free space optical
communication channel can be modelled by Kolmogorov theory. The refractive index along the
length of the channel is split into two components, the first component being the average
refractive index and the second one being the fluctuation component induced by spatial
variations of pressure and temperature in the atmosphere. The received optical signal is then
analysed in terms of spatial and temporal coherence after passing through the turbulent channel
described by Kolmogorov distribution of the total channel refractive index.
In a related work [1], a single-step Markov chain model is introduced to model the fading
correlation arising out of atmospheric turbulence in an outdoor free space optical link employing
intensity modulation/direct detection. This model is used to derive two low-complexity
suboptimal MLSD algorithms which are shown to provide significant performance improvement
over symbol-by-symbol detection.
The reliability of free space optical transmission through atmospheric turbulence can be
increased by scaling down the distance between the transmitter and the receiver through
multihop routing. This technique is well known as relayed transmission in radio frequency
communication systems. The multihop outdoor free space optical links in the presence of strong
atmospheric turbulence are modeled by Gamma-Gamma distribution in [13] for systems
employing Amplify-and-Forward or Decode-and-Forward relays. This model factorizes the
received optical signal as the product of two independent random processes each one having a

9
Gamma probability distribution function. As a special case study, it is shown that the dual-hop
free space optical communication system outperforms the single-hop link in the presence of
atmospheric turbulence.

2.2. Optical Sources and Detectors

There are a number of light sources and photodetectors that could be for OWC systems The most
commonly used light sources used are the incoherent sources light emitting diodes (LEDs) and
coherent sources—laser diodes (LD) LEDs are mainly used for indoor applications However, for
short link (e g , up to a kilometre) and moderate data rates, it is also possible to use LEDs in
place of LDs Lasers, because of their highly directional beam profile, are mostly employed for
outdoor applications Particularly for long transmission links, it is crucial to direct the energy of
the information to be transmitted precisely in the form of a well-collimated laser beam This is to
limit the often still very large channel power loss between the transmitter and the receiver In
order to limit the beam divergence, ideally, one should use a diffraction-limited light source
together with a relatively large high-quality optical telescope At the receiving end, it is also
advantageous to use a high- directionality telescope not only to collect as much of the
transmitted power as possible but also to reduce the background ambient light, which introduces
noise and thus reduces the performance of the link As for detectors, both the PIN and the APD
photodetectors could readily be used This chapter discusses the types of lights sources, their
structures and their optical characteristics The process of optical direct detection as well as
coherent detection is also covered in this chapter Different types of noise encountered in optical
detection will be introduced and the statistics of the optical detection process is also discussed

2.2.1. Light sources

For optical communication systems, light sources adopted must have the appropriate wavelength,
linewidth, numerical aperture, high radiance with a small emitting surface area, a long life, a
high reliability and a high modulation bandwidth There are a number of light sources available
but the most commonly used source in optical communications are LEDs and LDs, both of
which rely on the electronic excitation of semiconductor materials for their operation [1] The
optical radiation of these luminescent devices, LED and LD, excludes any thermal radiation due
to the temperature of the material as is the case in incandescent devices Both LD and LED light
sources offer small size, low forward voltage and drive current, excellent brightness in the visible
wavelengths and with the option of emission at a single wavelength or range of wavelengths

10
Which light source to choose mainly depends on the particular applications and their key
features, including optical power versus current characteristics, speed and the beam profile Both
devices supply similar power (about 10–50 mW) [2]. LEDs/LDs can be fabricated to emit light
across a wide range of wavelengths (colours) from the visible to the infrared (IR) parts of the
electromagnetic spectrum. The first and a part of the second segment are adopted in both optical
fibre and optical wireless communications The green part of the visible spectrum, 495–570 nm,
is of particular interest in underwater OWC because of a low attenuation window within this
band; ultraviolet (UV) radiations are currently being explored for indoor applications while IR
has long been widely used for optical wireless applications.

2.2.3 LASER (Light Amplification Simulation Emission and Radiation)

Laser diodes (LD) are used in applications in which longer distances and higher data rates are
required. Because an LD has a much higher output power than an LED, it is capable of
transmitting information over longer distances. Consequently, and given the fact that the LD has
a much narrower spectral width, it can provide high-bandwidth communication over long
distances. The LD’s smaller N.A. also allows it to be more effectively coupled with single-mode
fiber. The difficulty with LDs is that they are inherently nonlinear, which makes analog
transmission more difficult. They are also very sensitive to fluctuations in temperature and drive
current, which causes their output wavelength to drift. In applications such as wavelength
division multiplexing in which several wavelengths are being transmitted down the same fiber,
the stability of the source becomes critical. This usually requires complex circuitry and feedback
mechanisms to detect and correct for drifts in wavelength. The benefits, however, of high-speed
transmission using LDs typically outweigh the drawbacks and added expense. Laser diodes
(LDs) are a more recent technology which has grown from underlying LED fabrication
techniques. LDs still depend on the transition of carriers over the band gap to produce radiant
photons, however, modifications to the device structure allow such devices to efficiently produce
coherent light over a narrow optical bandwidth. As mentioned above, LEDs undergo
spontaneous emission of photons when carriers traverse the band gap in a random manner. LDs
exhibit a second form of photon generation process : stimulated emission. In this process,
photons of energy are incident on the active region of the device. In the active region, an excess
of electrons is maintained such that in this region the probability of an electron being in the

11
conduction band is greater than it being in the valence band. This state is called population
inversion and is created by the confinement of carriers in the active region and the carrier
pumping of the forward biased junction. The incident photon induces recombination processes to
take place. The emitted photons in this process have the same energy, frequency, and phase as
the incident photon. The output light from this reaction is said to be coherent [2][6]. In order for
this process to be sustainable, the double heterostructure is modified to provide optical feedback.
This optical feedback occurs essentially by placing a reflective surface to send generated photons
back through the active region to re-initiate the recombination process. There are many
techniques to provide this optical feedback, each with their merits and disadvantages. A Fabry-
Perot laser achieves photon confinement by having internal reflection inside the active region.
This is accomplished by adjusting the refractive-index of surrounding materials. The ends of the
device have mirrored facets which are cleaved from the bulk material. One facet provides nearly
total reflection while the other allows some transmission to free-space [9]. The operation of this
optical feedback structure is analogous to microwave resonators which confine electromagnetic
energy by high conductivity metal. These structures resonate at fixed set of modes depending on
the physical construction of the cavity. As a result, due to the structure of the resonant cavity
LDs emit their energy over a very narrow spectral width. Also, the resonant nature of the device
allows for the emission of relatively high power levels. Unlike LEDs which emit a light intensity
approximately proportional to the drive current, lasers are threshold devices. As shown in Figure
2.2, at low drive currents, spontaneous emission dominates and the device behaves essentially as
a low intensity LED. After the current surpasses the threshold level, stimulated emission
dominates and the device exhibits a high optical efficiency as indicated by the large slope in the
figure. In the stimulated emission region, the device exhibits an approximately linear variation of
optical intensity versus drive current.

2.2.3 LED (Light Emitting Diode)

LEDs are typically used in lower-data-rate, shorter-distance multimode systems because of their
inherent bandwidth limitations and lower output power. They are used in applications in which
data rates are in the hundreds of megahertz as opposed to GHz data rates associated with lasers.
These sources produce light at certain wavelengths depending upon the materials from which
they are made. Most fiber optic sources use wavelengths in the infrared band, specifically 850nm

12
 -9
(1 nm=10 m), 1300nm and 1550nm. For reference, visible light operates in the 400-700nm
range (Figure 2.2).

Figure 2.2: Electromagnetic Spectrum

2.2.3. Transmission Windows

They range of wavelengths at which the fiber operates best. Each range is known as an Operating
window. Each window is centered on the typical operational wavelength, as shown in Optical
fiber transmission uses wavelengths that are in the near-infrared portion of the spectrum, just
above the visible, and thus undetectable to the unaided eye. Typical optical transmission
wavelengths are 850 nm, 1310 nm, and 1550 nm. Both lasers and LEDs are used to transmit light
through optical fiber. Lasers are usually used for 1310- or 1550-nm single-mode applications.
LEDs are used for 850- or 1300-nm multimode applications.
Table 1 shows the transmission windows operating and their wavelengths.

13
Table 1. Fiber Optic Transmission Windows

Window Operating Wavelength

800 – 900 nm 850 nm
1250 – 1350 nm 1310 nm
1500 – 1600 nm 1550 nm

22.4 Photodetectors
The photodetector is a square-law optoelectronic transducer that generates an electrical signal
which is proportional to the square of the instantaneous optical field impinging on its surface
Thus, the signal generated by a photodetector is always proportional to the instantaneous
(received) optical power Since the optical signal is generally weak, having travelled through the
communication channel, the photodetector must therefore meet stringent performance
requirements such as high sensitivity within its operational range of wavelengths, a low noise
level and an adequate bandwidth to accommodate the desired data rate The effect of temperature
fluctuations on the response of the photodetector is required to be minimal and the device must
equally have a long operating life The wavelengths at which the detector responds to light
depend on the detector’s material composition Figure 2 19 shows the detector response curve for
different materials The ratio of the number of electron–hole (e–h) pairs generated aphotodetector
to the incident photons in a given time is termed the quantum efficiency η, which is Ideally the
detector should offer high sensitivity and fidelity with high optical to electrical efficiency
conversion, a large detection area, minimum noise, and short response times at low cost and high
reliability. Needless to say many of these requirements conflict and thus some compromise is
necessary. The choice of a photodetector for optical communications is generally between pin
photodiodes and avalanche photodiodes (APDs). The first two are usually the detectors of
choice for OW due to their widespread commercial availability. A pin photodetector is simple
both in structure and to employ but is about 10–15 dB less sensitive than an APD. The increased
power margin offered by APDs delivers a system that is more robust to pointing inaccuracy and
other losses. However, they need high bias voltages and exhibit highly nonlinear current gain and
sensitivity to reverse bias voltage giving rise to strong modulation effects from ambient light
interference. The other consideration is that of detector bandwidth, determined by the

14
capacitance of the device that limits the maximum speed of operation in conjunction with the
resistance presented by the receiver. However, to increase the amount of light in indoor OW
systems, large area detectors are to be preferred but this increases the capacitance. The trade-off
implied and methods to ameliorate the effect of the capacitance.

2.2.6 PIN PHOTODETECTOR :

The PIN photodetector consists of p- and n-type semiconductor materials separated by a very
lightly n-doped intrinsic region [2] In normal operating conditions, a sufficiently large reverse
bias voltage is applied across the device as shown in the schematic figure.

Figure 2.3. PIN photodetector schematic diagram

The reverse bias ensures that the intrinsic region is depleted of any charge carriers, For the
device to convert an incident photon into an electron/electric current, the energy of the incoming
photon must not be less than the band-gap energy of the semiconductor material The incident
photon uses its energy to excite an electron from the valence band to the conduction band,
thereby generating a free electron– hole pair in the process Normally, the incident light is
concentrated on the depleted intrinsic region The high electric field present in this depleted
region causes the generated charge carriers to separate and be collected across the reverse biased
junction. This gives rise to a current flow in an external circuit. there is one electron flowing for
every carrier pair generated The semiconductor material of the photodetector determines over
what wavelength range the device can be used. The upper cut-off wavelength in micrometres is
generally given by Equation.

15
 hc
C m  
1.24
 2.1
Eg Eg

The responsivity of a PIN photodetector is always less than unity and a graph showing typical
responsivity values for different PIN photodetectors is depicted, PIN photodetectors are capable
of operating at very high bit rates exceeding 100 Gbps. A high-efficiency waveguide PIN
photodiode with a bandwidth greater than 40 GHz at a responsivity of 0 55 A/W (external
quantum efficiency of 44%) has been reported in Ref. whereas In GaAs unitravelling- carrier
photodiode with 310 GHz bandwidth at 1550 nm wavelength is demonstrated in Ref. However
the commercially available devices only offer bandwidth up to 20 GHz, this is mainly due to
limitations of the packaging.

2.2.6. APD PHOTODETECTOR :

The APD is different from the PIN photodetector in that it provides an inherent current gain
through the process called repeated electron ionization This culminates in increased sensitivity
since the photocurrent is now multiplied before encountering the thermal noise associated with
the receiver circuit Hence, the expression for the responsivity of an APD includes a
multiplication (or gain) factor given by

M  IT / I p 2 .2

Where I is the average value of the total output current and Ip = is the primary unmultiplied
photocurrent.

16
 Chapter Three

SYSTEM MODEL AND ANALYSIS

3.1 Channel Modeling

When modeling the optical channel, there are certain conditions and constraints different from
Radio Frequency (RF) communication channels that need to be considered [3],[1],[5]. The
information in an optical channel is transmitted at the optical intensity I(t) of the transmitted
signal, which is defined as the optical power emitted per solid angle, in units Watts per steradian
[W/sr] (for more details, see [3]). This differs from a typical RF channel, where the information
is contained in a signal’s amplitude, frequency or phase. Since optical power can never take
negative values, the signal containing the information transmitted through the optical channel has
to remain non-negative at all times.
To design, implement and operate efficient optical communication systems, it is imperative that
the characteristics of the channel are well understood. the effects of channel distortions A
considerable amount of work has been published on the channel characterization, covering both
experimental measurement and modeling, these Two types of configurations are considered in an
optical wireless channel and consequently the path loss is easily calculated from the transmitter
beam divergence, receiver size and separation distance However, a non-LOS configuration, also
known as diffuse systems (mainly used in indoor environment), uses reflections off the room
surfaces and furniture These reflections could be seen as unwanted signals or multipath
distortions which make the prediction of the path loss more complex A number of propagation
models, for LOS and non-LOS are introduced that affects of link performance that can affect
characteristics of the propagating optical beam, thus resulting in optical losses and turbulence
induced amplitude and phase fluctuation There are a number of models to characterize the
statistical nature of the atmospheric channel, and these will be discussed practically test bed for
investigating the atmospheric effect on the free space optics link, as well as measured data. There
are a number of topologies that are commonly used for optical wireless applications, the
configurations can be classified according to (1) the degree of directionality of transmitter and
receiver. (2) the existence of the LOS path between the transmitter and the receiver, The drive
current of an optical source is directly modulated by the modulating signal m(t), which in turn

17
varies the intensity of the optical source x(t). The receiver employs a photodetector with a
response which is the integration of tens of thousands of very short wavelengths of the incident
optical signal that generates a photocurrent y(t) This photocurrent is directly proportional to the
instantaneous optical power incident on it, that is, proportional to the square of received electric
field. Optical wireless system has an equivalent baseband model that hides the high- frequency.

Figure 3.1. Show the block diagram of optical wireless communication channel

Figure 3.2. Show the equivalent baseband model of an optical wireless system

where R is the photodetector, h(t) is the baseband channel impulse response and n(t) is the
signal-independent shot noise, modelled throughout the additive white Gaussian noise (AWGN)
with a double-sided power spectral density (PSD) of N0/2.
Non-LOS links, particularly in indoor applications, are subject to the effects of multipath
propagation in the same way as RF systems and these effects are more pronounced This type of
link can suffer from severe multipath-induced performance penalties, Multipath propagation
causes the electric field to suffer from severe amplitude fades on the scale of a wavelength. The
detector would experience multipath fading if the detector size proportional to one wavelength
or less Fortunately, OWC receivers use detectors with a surface area typically millions of square
wavelengths In addition, the total photocurrent generated is proportional to the integral of the
optical power over the entire photodetector surface; this provides an inherent spatial diversity.

18
Although indoor OWC links do not suffer from the effects of multipath fading, they do suffer
from the effects of dispersion, which manifests itself in a practical sense as the intersymbol
interference (IS). Dispersion is modelled as a linear baseband channel impulse response h(t). The
channel characteristic of an OWC link is fixed for a given position of transmitter, receiver and
intervening reflecting objects, The channel characteristic only changes when these components
are moved by distances of the order of centimeters, Due to high bit rates and the relatively slow
movement of objects and people within a room, the channel will vary only on the time scale of
many bit periods, and may therefore be considered as quasi-static. The equivalent baseband
model of an optical wireless link can be summarized by the following equations

yt   Rxt   ht   nt  3.1


  Rx ht   d  nt  3.2


The impulse response h(t) can be used to analyse or simulate the effects of multipath dispersion
in indoor OWC channels.
I t   3.3

The other important constraint rises from concerns about the eye and skin safety [4],[3]. This has
been addressed in numerous guidelines and standards by international regulatory bodies, such as
the International Electrotechnical Commission (IEC) in standard IEC60825-1 [5] and the
American National Standards Institute (ANSI) in their publications [4]. The limitation on the
optical signal power that can be emitted from the optical source is mathematically expressed as:

∫ ( )

This means that the average amplitude over the signal period, which is equal to the normalized
average optical power, is limited to some fixed value P according to the above mentioned
standards. These two constraints are considered in the whole simulation.
3.2 Modulation techniques

19
Modulation is the process of mapping one information signal onto another signal, called the
carrier signal, according to a given set of rules [1]. Its purpose is to convey the information
signal over a wireless medium and in order to do so, it has to be modulated and shifted to some
higher frequencies suitable for wireless transmission. The most suitable set of modulations for
the optical channel transmission, keeping in mind the above mentioned constraints, are digital
modulation techniques [1],[4].
In a digital modulation scheme, an information signal is a sequence of bits originating from a
digital information stream. The modulation scheme modifies a carrier signal based on a set of
modifications called the modulation alphabet. The modulation alphabet is a set of bits or groups
of bits, called symbols that represent the changes of the carrier signal used to carry the
information signal. In a typical RF system there are different signal values that can be
modified/modulated, such as amplitude, frequency and phase of the carrier signal. In optical
channels, the optical intensity is most commonly modulated and the signal is received with
photo-detectors that have a surface area usually many times larger than the optical wave-length
to be detected. Hence the processed signal comes as an average of thousands of received
wavelengths.
For simulations executed in this project, the popular modulations used for indoor and outdoor
optical wireless channels are described in the following sections.

3.2.1 On-Off Keying

On-Off Keying (OOK) together with Pulse-Position Modulation (PPM) (discussed in the next
section) are some of the most popular techniques used in optical communications. Both are often
used in commercial systems due to their power efficiency. OOK is one of the simplest binary
modulation schemes, with only two values for signal changes. This scheme employs two
different signal intensities, each with equal duration and probability during the symbol interval.
In its simplest form, one optical pulse represents one bit of the symbol sequence and the other bit
is represented by the absence of the pulse [4].
There are two line coding schemes — Return-to-zero (RZ) and Non-return-to-zero (NRZ) —
which are extensively used. In the NRZ line coding scheme, the pulse duration equals the bit
duration, whereas in the RZ line coding scheme, the pulse duration equals to the half of the bit

20
duration (the other half is equal to zero). The intensity of the pulse is 2P where P represents an
average optical power. A random OOK signal is presented in figure 3.3.

Figure 3.3. Random OOK signal [9]

In channel simulations with an Additive White Gaussian Noise (AWGN) channel (as typically
done in signal processing models for communication channels), the analytical/theoretical
probability of the bit error rate for OOK is given as:

 PRh 
Pb , OOK  Q  3.5
 R  2n 
 s 

where the probability error function Q(x) is given as:

   u2 
Q x  
1
2
x
exp
 2 
dx 3.6

Bit error rate (BER) is the number of erroneous bits received divided by the total number of bits
sent, and it is used in that form for all further uncoded and coded modulation schemes. A general
form of the equation is:
Ne
Pb  lim 3.7
N s  Ns

In the simulations, a finite number of bits Ns are used in a bit sequence. In case when a channel
coding method is implemented in addition to the modulation, another measure of error is also

21
used, namely the Symbol Error Rate (SER), which is calculated analogous to BER in 2.5, only
using symbols instead of bits.

3.2.2 Pulse-Position Modulation

Pulse-position modulation (PPM) is another popular technique used in optical communications.
It is basically the coded version of OOK. In M-ary PPM, each symbol interval is divided into a
series of M subintervals or chips. Information is transmitted by applying the positive optical
intensity to one chip while the other chips remain zero. For example, in 4-PPM there are four
symbols and each of them is represented by a different positive value for the chip duration in
every interval.
Since all chips are non-overlapping in time, each symbol is orthogonal to all other symbols [3]
and hence the symbols don’t interfere.

Figure 3.4: Random PPM signal [9]

The analytical probability of error, Pb,PPM (shown in equation 2.13), where M is the number of
different chips, R the bit rate and Q(·) the probability error function 2.4. It is an approximate
theoretical value [1].

M  M log 2 M 

pb , PPM  .Q P  3.8
2  2R 

PPM is more bandwidth efficient modulation scheme compared to the binary scheme OOK.
However, both schemes are considered to be spectrally inefficient in terms of the bandwidth
[1],[4].

22
In terms of the optical power needed for the transmission, they are considered as power efficient
schemes.

3.2.3 M-ary Phase Shift Keying

The following modulations are widely used in radio communications, hence their possible usage
and adaptation to optical wireless is investigated within this research work. While widely used
modulations in optical wireless communications, such as OOK and PPM, belong to a group of
baseband pulse modulation (since they operate in the baseband part of the spectrum), the
modulations presented in this section belong to the bandpass or passband modulations. Due to
constrain when working with optical signals (mentioned in equation 2.1 and 2.2) most of the
conventional RF modulation techniques cannot be directly applied to optical signals [5]. The
ones presented below, can be applied to optical signals and they are simulated. M-PSK stands for
M-ary Phase Shift Keying, where M is a number of symbols used in modulation, and in this case
the number of different carrier phases. Hence it groups bits into a finite number of symbols that
change the phase of the carrier signal. There are different types of PSK, depending on the
number of symbols in the symbol alphabet.
These techniques are used as Subcarrier intensity Modulations (SCM) in order to better exploit
the bandwidth of an analog optical signal. It means that a number of baseband signals are up
converted in frequency before modulating their characteristics, such as intensity, frequency or
phase of the optical carrier. In order to satisfy the requirements for the optical communications,
these signals have to be non-negative, so a certain DC-offset is added in order for them to
comply with the requirements [5] This has another drawback - its consuming more optical
power, hence systems with applied SCM become less power efficient. For example, in a typical
AWGN channel with direct detection, BPSK and QPSK both require around 1.5dB more optical
power than OOK or 2-PPM schemes.

3.2.4. Binary Phase Shift Keying

Binary Phase Shift Keying (BPSK) is the simplest PSK modulation, where the symbol alphabet
consists of only two values, bit 0 and bit 1 [2]. Signal phases are separated by the angle of 180$.
Each bit is mapped to one signal value or a symbol, which in the case of optical communications
can only be positive values.

23
 Figure 3.5. BPSK constellation diagram

BPSK values are shown on the constellation diagram in Figure 3.4. This diagram is used to
represent the symbols from the modulation alphabet in a two-dimensional plane at sampling
times, without taking into account the frequency of the carrier signal. It is convenient for
representing digital modulations in a simple way. It can clearly be seen that BPSK has only two
values that represent signal changes.
Theoretical value of the bit error probability for the BPSK modulation in digital systems is given
in equation 3.5, with erfc(·) as the complementary error function, and Eb/N0 signal to noise ratio
(analogue to SNR) [1].

Figure 3.6. QPSK constellation diagram

1  Eb 
Pb , BPSK  erfc 
 3.9
2  N o 

BPSK is not very spectral efficient scheme, since it doesn’t utilize the available bandwidth
efficiently. Its bandwidth efficiency can be improved when this scheme is applied to subcarrier
modulations.

24
3.2.5. Quadrature Phase Shift Keying
QPSK stands for Quadrature PSK, and its alphabet consists of four symbols. Each symbol is
represented with two bits and it results in the following alphabet: 00, 01, 10, 11. QPSK is less
noise-prone than BPSK and its usually used to double the data rate compared with BPSK at the
same bandwidth, or to keep the data rate the same at half of the bandwidth of BPSK. However,
in optical scenario, it acts similar to BPSK - hence it requires twice of the bandwidth than OOK
and the optical power requirement is higher.
Theoretical bit error probability for QPSK modulation (from [5]), can be approximated to:
 Es 
Pb , QPSK  erfc 
 3.10
 2 No 

with erfc(·) as a complementary error function.

Figure 2.5 shows a typical constellation diagram of QPSK, with four symbol values equidistant
from each other.

3.2.6. M-ary Pulse Amplitude Modulation

M-ary Pulse Amplitude Modulation is modulation in which the amplitudes of pulses are varied
according to the information signal that is going to be transmitted over the channel. Applied to
an optical channel, it can be seen as the generalization of OOK to a set of M symbols [3]. Every
PAM symbol with duration Ts is created by scaling the signal amplitude byM different factors. If
the input data is uniformly distributed (signals with the same amplitude value equally
distributed), the symbols are equally probable and the average power requirement, i.e. the power
intensity for the eye-safety for the optical signal, is fulfilled.
PAM modulations are more bandwidth efficient than PSK modulations, however they are not
used a lot in free space optical channels because of their sensitivity to channel loss and lower
power efficiency [4].

3.2.7. 4-PAM
This is a special case of PAM, where a two-bit modulator is used and it maps the signal
amplitude to one of four possible levels. In figure 2.6 a random PAM signal is presented, with

25
four different values for four different symbols that represent different pulse signals. Pulse
amplitude levels can be any value as long as they meet optical channel constrains.

Figure 3.7. Random PAM signal [9]

Symbol error rate for 4-PAM (as in [5]) with symbols belonging to the alphabet +1,−1, +3,−3.
Assuming equal probability of all the symbols and using the Gaussian probability distribution (as
in [3]) results in the equation 2.9:

3  Es 
Ps , 4 PAM  erfc 
 3.11
4  5 No 

3.2.8. Multi-Level Quadrature Amplitude Modulation (M-QAM)

In Quadrature amplitude modulation two carriers are shifted in phase by 90 degrees comprising
of in-phase and quadrature components. Then they are modulated using the Amplitude shift
keying (ASK) digital modulation. Therefore, the resulting output is a combination of two
modulation techniques, phase and amplitude modulation, PSK and ASK. PAM scheme and its
implementation are basically the same for an optical channel as for the RF channel, ensuring that
optical channel constraints are met.
The conventional way to adapt QAM to an optical channel is to add a DC bias to make the
negative values become non-negative (so the channel requirement in 2.1 is met) and ensures that
an average power stays within the certain limits (eye-safety constraint specified by channel
requirement in 2.2). This method is bandwidth efficient, but its drawbacks are that different
states of modulation signal are closer together so the whole signal is more susceptible to noise.

26
Both, PAM and QAM modulation schemes have more spectral efficiency due to transmitting
more information per symbol, but they are less power efficient [3].

3.2.9. 16-QAM
This higher-order modulation transmits four bits per symbol, and an example of its alphabet is
given in [5]. Since it is a higher-order modulation, symbols are closer together. For analytical
symbol error calculation, we use the following alphabet:

{±1 + ±1j, ±1 + ±3j, ±3 + ±3j, ±3 + ±1j}(3.12)

Assuming the Gaussian probability distribution of noise (additive white noise model) and that all
symbols are equally likely, the symbol error rate can be calculated using equation 2.11:

3  Es 
Ps , 16QAM  erfc 
 3.12
2  10 No 

3.3.0. Comparison of modulation techniques

When comparing modulation schemes in most of the cases two factors are considered; the optical
power efficiency and spectral efficiency (i.e. how effectively the available bandwidth is utilized)
[3]. The optimal solution for the wireless optical channel would be to maximize both of them.
Binary level modulations offer the best power efficiency while on the other hand, they have
lower bandwidth efficiency. Multilevel modulation schemes offer higher bandwidth efficiency;
they have lower power efficiency.
In Figure 2.7 the comparison between different modulation schemes is given by taking into
consideration the optical power gain over the simple OOK modulation scheme and the
bandwidth efficiency of the given scheme [4]. As described later (in the discussion of results),
typically phase-key shifting modulation techniques, such as BPSK and QPSK, require more
optical power than OOk, about 1.5db. PAM and QAM as multilevel modulation schemes offer a
better bandwidth efficiency, but they have lower power efficiency and increased sensitivity to the
channel loss.
27
 Figure 3.8: Optical power gain over OOK vs bandwidth efficiency [13]

3.3 Channel coding

Channel coding or Forward Error Correction (FEC) includes a set of methods used to improve
performance of the channel by introducing redundant bits for detection and correction of possible
channel errors. As more redundant bits are added, less information is sent through the channel,
which leads to lower data throughput. Therefore, depending on the system requirements, there is
always a trade-off between data throughput that aims at less delay on the channel and faster
communication versus better error correction directed towards less errors and better channel
performance.
There are several types of FEC codes, depending on the classification of the same. They can be
linear, using linear algebra and polynomial arithmetic, and non-linear. They can operate on
chunks of data, in which case they are called block codes, or on stream of data bits continuously
inserting redundant bits for error detection and correction; in which case they are denoted as
convolutional codes. Another characteristic sometime used in a description of some block codes
is the cyclic nature of the code. Cyclic block code means that any cyclic shift of a codeword is

28
also a codeword. Hence, when codewords are added together and shifted circularly, the result is
still a codeword. This is a very practical characteristic.
There are various channel coding algorithms together with decoding algorithms, each of them
has its own advantages and disadvantages for optical wireless communications. In this section
three methods are presented and they are to be implemented in the graphical simulation; these
are binary Bose, Chaudhuri and Hocquenghem (BCH) code, Reed-Solomon (RS) error
correction code, Low Density Parity Check (LDPC) code. The use of RS together with a higher-
order PPM scheme has shown an improvement in channel performance for strong attenuations
[4] (due to fog or weather conditions).

3.3.1 Bose, Chaudhuri and Hocquenghem (BCH) codes

Bose, Chaudhuri and Hocquenghem (BCH) codes are powerful random error correcting codes
[1],[3],[4]. They belong to a group of cyclic block error correcting codes and they are able to
correct multiple errors in the transmitted bit sequence. Decoding of BCH codes is possible in an
easy way, using an algebraic method called syndrome decoding [4],[2],[3]. There are two
versions of this code: binary – that are built upon binary fields with elements 0, 1; and non-
binary – that are built upon binary fields with more than two elements. A widely used non-binary
BCH code is Reed-Solomon (RC) code. Codeword belong to a finite field called Galois field that
contains a finite number of elements as well as other properties that are convenient for building
good codes [2]. Information bits are represented in the form of a polynomial, and the codewords
are built by taking the remainder after division of that polynomial by a chosen generator
polynomial. The generator polynomial is selected so it gives the code it cyclical characteristics –
all codewords are multiplies of the generator polynomial. BCH codes are typically denoted as
BCH(n, k) with n as the total number of bits, and k as the number of information bits. For
example, BCH(31,16) has codeword of 31 bits length, 15 check bits, and it is able to correct 3
errors, which means a minimum distance between codewords is 7. Information bits get zeros
appended, in the number equals to the degree of the generator polynomial (here 15). Then they
are divided by the generator polynomial (XORing elements of both of them) and the remainder
of polynomial division becomes the sequence of check bits. The codeword is built by appending
that sequence to the information bits. In this way the information and check bits can be
recognized in the resulting codeword; and this is called systematic encoding. By dividing the

29
codeword by the generator polynomial again, it can be checked if there were errors in
transmission. If the remainder is zero, there are no errors in the received codeword; if the
remainder is non-zero, the received codeword is erroneous. This remainder is called the
syndrome and it is used in the decoding algorithm for the location of the erroneous bits and
correction of the same.
Decoding is performed in a couple of steps. First the syndrome is computed from the received
codeword and the generator polynomial. Then the error location polynomial is found from a set
of equations derived from the syndrome. And lastly, the error location polynomial is used to
detect the erroneous bits and correct them. Further details about the BCH decoding algorithm can
be read in [3].

3.3.2 Reed-Solomon (RS) block codes

Reed-Solomon block codes are non-binary BCH codes. They are capable of correcting burst
errors as well as random symbol errors. In RS codes, information bits are gathered in blocks, that
have a fixed length and can further be divided into symbols, which are m-bits long [2]. Since this
code uses symbols (information bits), error correction information is added in the form of parity
symbols. A RS coded word consists of information symbols together with attached parity
symbols.
The notation RS(n,k) means n is the total number of symbols, or one codeword, and k is the
number of data or information symbols being encoded. Maximum number of errors that can be
corrected by the RS code is t = [n − k]/2 within one code block [2]. If more than t errors occur,
the code will not perform correctly and symbols may be decoded incorrectly, hence the original
information can be lost. RS codes are based on finite fields, also called Galois fields GF(2m).
Originally, k message symbols are viewed as coefficients of a polynomial p(x) over a finite field.
The field has a property that any defined operation on the finite field element results again in
another element of the field.
In addition, it has a finite number of elements, for example the finite field of integers. Detailed
explanation of RS codes can be read in [2],[5]. The error probability for the RS decoded symbol
is given in equation 3.12:

30
 m
1
 2 m  1
2

Ps 1  Ps 2 m1 j
1
Ps , RS  m
2 1
 j
j t 1 

j  3.13


where Ps stands for the symbol error probability of the deployed modulation scheme. As
previously stated, RS codes can correct up to t symbols, no matter how many bits get corrupted
in any of these t symbols. That makes them suitable for burst error correction. However, if the
number of erroneous symbols exceeds t, even if for only one bit error t+1, the codeword will not
be correctly decoded and the RS algorithm fails. Hence using RS codes on symbols has its
advantages, but also disadvantages. RS codes have gained a lot of attention in optical wireless
communications, especially in FSO links [2]. There are low-complexity hardware RS decoders
available that are well tested and can achieve a high data rate [4], which makes them one of the
most popular channel codes for optical wireless applications.

2.3.3 Low-Density Parity-Check (LDPC)

Low-density parity-check (LDPC) codes are a linear error-correcting code, which means that
they can be decoded in a time that is linearly proportional to their block length. They are called
low density due to the fact that their parity-check matrix contains only few 1’s compared to the
number of 0’s [1],[3]. Also their performance is very close to the channel capacity of many
different channels. When evaluating bit error performance on the channel, these codes exhibit a
typical characteristic abrupt drop of the BER curve at a certain level of SNR (signal to noise
ratio). The codeword is built with a help of the low-density parity-check matrix H(n−k)⇥k,
where n is the length of the codeword and k is the number of information bits. From this matrix,
a new generator matrix is created according to the G = [Ik|P]. By multiplying the generator
matrix with the sequences of all possible bit combinations of k bits in the length of 2k bits, new
codewords are generated. For each sequence a new codeword in the length of n bits is generated.
LDPC codes usually work with very long codewords, which makes processing them (encoding
and decoding) very slow. However, with improved hardware development and decoding
algorithms, processing times have reduced and it is possible to design more time efficient LDPC
codes. They are adopted as a standard for Digital Video Broadcast - Satellite (DVB-S2)
communications, where there can reach data rates close to 1Gbit/s [3]. Their usage in such an

31
environment together with their aforementioned advantages makes them interesting for
investigation and a candidate for usage in the optical wireless communications.

3.1.4. Bit error rate analysis

We only consider a direct LOS of FSO link; thus, only the background radiation modelled as an
AWGN is considered Assuming IID data transmission, the total probability of error Pec
conditioned on the received irradiance is given by

PeIO  PeI1
1 1
Pec  3.14
2 2

where P(e | 0) is the conditional bit error probability for receiving a ‘1’ provided a ‘0’ was sent.
Noise signals {nx(t), ny(t)}, including the background noise and the quantum noise,
Can be expressed as.

n x t   n xp t cos IF t  t t   n xq t sin  IF t  t  3.15

n y t   n xp t cos IF t  t t   n yq t sin  IF t   t  3.16

where {nxi(t), nxq(t)} and {nyi(t), nyq(t)} are the phase and quadrature components,
respectively, having a normal distribution with a zero mean and a variance of σn

c x t   K  n xi t cos IF t  t t   n xq t sin  IF t  t t  3.17

32
Given m(t) = 0 and the baseband outputs Vx(t) and Vy(t) for the upper and the lower arms (Figure
3.9), respectively, are given as

Vx t   K  nxi t   n 2 xq t 
2
3.18

V y t   n yi t   n yq t 
2 2
3.19

Vx(t) and Vy(t) have fixed mean values and the same variance given by

E V x t   k

 
E V y t   0

 2 x   y 2   n2 3.20

With ωIF << ωr, the PDF of Vx(t) and Vy(t) can be described by the Rice and the Rayleigh
probability functions, respectively.


V  KV   V 2  K 2 

pVx   x2 Io 2x  exp x  3.21
   2 n  
 n  n 
2


 V y2 
p V y   2 exp  
Vy
 3.22
n  2 2 
 n 

where I0 is the zero-order modified Bessel function of the first kind [7]
The conditional BER for m(t) = 0 can be derived as

33
 
  
V  KV    2Vx2  K 2  
Pec   pVx   pV y dVy  dVx    x2 Io 2x
 
 exp  dVx 3.23
0 n   n 2 n 
2
0 vx    

By invoking changes of variables and substituting into (3.21 into 3.22), Pec now becomes


1
2
 
Pec  e  n 2  mIomne  m 2  n 2  2dm 3.24
0

Pec is represented as

1  n2  1  K2 
Pec  exp    exp  
2 
3.25
2  2  2  4 n 

The electrical SNR at the output of the BPF is defined as

  
SNRPr   R dc 1   dc2 Pr PIo / 2 n 2
3.26

Pec can be expressed in terms of the SNR by substituting (3.23) into (3.24):

  
SNRPr   R dc 1   dc2 Pr PIo / 2 n 2
3.27

This result is the same as the BER expression of FSK With regard to the system sensitivity,
OOK and PPM techniques have complete equivalence.

3.1.5 Performance Analysis of OWC Modulation over Wireless Channels:

We now consider the performance of the modulation techniques discussed in the previous
chapter when used over AWGN channels. There are two performance criteria of interest first one
is the probability of error which defined relative to either symbol or bit errors, and second is the
outage probability which also defined as the probability that the instantaneous signal-to-noise
ratio falls below a given threshold. Wireless channels may also exhibit frequency selective
fading and Doppler shift, which gives rise to intersymbol interference (ISI), causes an irreducible
34
error floor in the received signal. Doppler causes spectral broadening, which leads to adjacent
channel interference and to an irreducible error floor in signals with differential phase encoding.
This chapter describes the impact on modulation performance. The performance of the OOK and
PPM in the presence is investigated to improve the link performance possible mitigation
techniques using high-pass filtering, equalization. In this section we define the bit error rate
(BER). We then examine the error probability on AWGN channels for different modulation
techniques as parameterized,
To compare different modulation schemes under different channel conditions, two key
performance indicators are used as defined below:

1. Normalized optical power requirement (NOPR): The NOPR of a system is calculated

by normalizing the optical power required to achieve the desired bit/slot error probability
ξ in the interfering channel with that of OOK system at 1 Mbps in an ideal AWGN
channel without interference.

2. Optical power penalty (OPP): The OPP of a system is calculated by normalizing the
optical power required to achieve the error probability of ξ in the interfering channel with
that of the ideal AWGN channel without interference.

The input bits, assumed to be i. i. d. and uniform on {0, 1}, are passed to a transmitter filter,
which has a unit amplitude rectangular impulse response p(t), with a duration of one bit Tb. The
pulses are then scaled by the peak transmitted optical signal power 2Pavg, where Pavg is the
average transmitted optical signal power.

35
 The optical signal x(t) is then subjected to the multipath distortion, hence the channel output
φ(t) in the absence of noise is given by

 t   xt   ht  3.28

Where the symbol ʘ denotes convolution.

The receiver front consists of the unit energy filter r(t), followed by a sampler and a threshold
detector. The input signal to the matched filter is given by

Z t   R t   nt  3.29

Note that this filter is optimum only when there is no multipath dispersion. The output of the
matched filter is sampled at the bit rate, and the samples are passed to a threshold detector, which
assigns a one or zero to each bit depending on whether the sampled signal is above or below the
threshold level, thus generating an estimate of the transmitted bit sequence.

The system can be modelled using the discrete time model. The discrete-time impulse
response ci of the cascaded system is given as

Ci  pt   ht   rt t  iTb 3.30

with the normalization Σici = 1 and sampling times are shifted to maximize the zero-sample h0.
The noise samples ni with zero mean and variance σ2 are given by

ni  nt   r t t  iTb 3.31

Unless the channel is no dispersive, ci contains a zero tap, a single precursor tap and possibly
multiple post cursor taps. The magnitude of the zero tap is larger than the magnitudes of the
other taps in a nondispersive channel, the optimum sampling point, that is, that which minimizes

36
the probability of error occurs at the end of each bit period. However, in dispersive channels, the
optimum sampling point changes as the severity of ISI changes. In order to isolate the power
penalty due to ISI, two assumptions are made. Firstly, perfect timing recovery is assumed. This
is achieved by shifting the time origin so as to maximize the zero tap, c0 . Secondly, an optimal
decision threshold is assumed. For OOK, basic symmetry arguments can be used to deduce that
the optimum threshold level lies midway between expected one and zero levels, regardless of the
severity of ISI.
Suppose that ci contains ζ taps. Let ai be an m-bit sequence, and aj,m−1 the value of the (m −
1)th slot in the sequence ai, where aj,m−1 ∈{0,1}. Unless the channel is nondispersive, for a ai
sequence of m slots, ci will contain m taps (slots): a single precursor tap, a zero tap, which has
the largest magnitude, and (m – 2) postcursor taps. All slots within a given sequence ai are
affected by dispersion of pulses appearing within Si as well as adjacent sequences ai – 1, or
ai+1, except for the penultimate slot, which is only affected by pulse dispersion appearing within
the sequence ai . Sequences that fall outside the boundaries of ai will not contribute to the
dispersion on the penultimate slot of ai. Therefore, when calculating the optical power
requirement, only the penultimate slot will be considered for each sequence.

Figure 3.9 Discrete-time impulse response ck for a sequence of six taps.

Let ai be an ζ-bit sequence and ai be the value of the penultimate bit in that sequence, where ai
∈ (0, 1). Let yi denote the receiver filter output corresponding to the penultimate bit, which, in
the absence of noise, is given by

yi  2RPavg ai  hk k   3.32

37
where ζ is the number of channel taps and ai = {a1, a2 . . . aζ} is the ζ bit sequence.
The average BER is approximated by averaging overall possible bits sequence of length ζ, and is
given by

 0.5RPavg Tb  yi  1  yi 0.5 RP Tb 
1  
Pe 
2
 Q
 N / 2

 2  Q

avg

N0 / 2 
3.33
iSo
 o  iS1
 

Where s0 and s1 correspond to bit ‘0’ and ‘1’, respectively.

The effect of multipath propagation on OOK–NRZ signaling with a normalized delay spread DT
of 0.4, where variations in signal amplitude due to preceding pulses can easily be seen. It is
intuitive to see that for a given severity of channel-induced distortion higher data rates will cause
further variation in signal amplitudes. It can also be noted from the figure that the best sampling
point would be at the end of the bit period. Based on the assumption of the perfect timing
recovery and channel delay, the program to calculate the error probability of OOK–NRZ in a
multipath channel.

38
 Chapter Four

RESULTS AND DISCUSSIONS

The application of different signal processing techniques on optical wireless channels offer
different channel performance and error rates. Bit Error Rate (BER) simulations are usually
executed with varying combinations of modulations and coding techniques in order to investigate
the best signal performance. In this chapter some of these results are presented and compared.
Each of the combinations has its own advantages and disadvantages; it all comes up to the trade-
off between different modulation and coding schemes on the channel. Matlab is used to create
plots for results from the simulation framework due to it being the de facto tool for doing so and
the visual ease by which the results can be consumed in this form.
First the results of simulations employing only modulation techniques are presented and
compared, followed by the results of simulations employing channel coding schemes together
with modulation schemes. A brief discussion is provided as to why some schemes are preferred
in practical use over others.

4.1. Additive White Gaussian Noise (AWGN)

In a typical wireless communication system, the channel between the transmitter and the receiver
section is simulated with the Additive Gaussian White Noise (AWGN) channel model [1]. This
well known mathematical model is suitable for the simulation of the optical wireless channel and
according to it the main source of the noise in the transmission channel is the noise produced in
the receiver, which is caused by thermal movements of the electrons in the receiver equipment.
This effect of model is perfectly described using the Gaussian probability distribution for the
amplitude values that represent a linear addition of wideband or white noise with the constant
spectral density [3].
In the simple case of BPSK modulation, variance of the additive white noise is calculated
according to the equation 4.1

1
 4.1
E / No
2.10 s
10

39
where Es/N0 represents Energy per Symbol to Noise Power Spectral Density ratio, defined as
[3][2].
With Eb/N0 as the Energy per Bit to Noise Power Spectral density ratio, and CR is the Code
Rate of the channel coding method used in the simulation.

4.1.1. Visualization in Matlab

Results of the simulations are also presented graphically in the form of Matlab plots, where they
can be viewed better and also conclusions about their behavior can be made. For this purpose, a
small Matlab is explained in equations (3.23) and (3.24) and plots the BER from values the
simulation and the theoretical BER vs the SIS SNR values. An example of such a plot is given in
Figure 4.0

Figure 4.0: BER vs SNR plot

40
4.1.2. Modulation schemes unclouded
The most commonly used modulations techniques are the binary modulation schemes that
consist of only two different intensity levels to represent the binary information. These schemes
have become very popular (e.g., OOK and PPM). On the other hand, multilevel schemes tha wst
utilize more than two levels of a certain signal parameter can achieve higher data rates than
binary schemes, but they are not widely used in FSO due to their bad performance at fading
events. In the simulations performed with the simulator it is assumed that the modulator has a
fixed bandwidth in GHz, which means that it can produce rectangular pulses of fix duration.

4.1.1 Comparison between RF and optical

Multilevel modulation schemes, such as M-PAM are known to improve spectral efficiency,
especially with increasing M by utilizing the bandwidth more efficiently (than for example
simple OOK modulation). On the other hand they require higher signal to noise ratio at the
receiver, which means an increased optical power. This graph is explained in equation (3.4)

Figure 4.1: BER of OOK vs M-PAM Modulation

41
In Figure 4.1, simple OOK is compared with PAM for M = 2, 4, 8, 16. It can clearly be seen that
in order to achieve the same bit error rate employing the higher order PAM, more optical power,
approximately 1.8 dBs, is needed at the receiver. This graph is explained in equation (3.24).

Figure 4.2: BER of OOK vs M-PAM modulation

Phase shift keying modulations, such as BPSK and QPSK in figure 4.2 require a receiver with
higher sensitivity. Due to their BER curve more signal power is needed at the receiver, than
when using optical intensity modulations (OOK, PPM). They both require approximately 2dB
more optical power.
PPM and PAM modulation are compared using the same symbol rate and bandwidth, which
makes the comparison more accurate. They are multilevel modulation techniques, which simply
mean that more bits are encoded into symbols, and bandwidth can be utilized more efficiently.
From figure 5.3 it can be seen that the SNR level necessary for the error-free reception reduces
when the signal constellation size M increases.

42
4.1.2. Modulation schemes coded
For the additional channel performance gain by employing error correcting methods, popular
codes such as BCH, RS and LDPC can be used. They are well described in chapter 2 and their
implementations can be found in previous work and Matlab simulations. Basic declarations for
these coding schemes have been provided and can be implemented easily. This graph is
explained in equation (3.30).

Figure 4.3: BER of multilevel PPM modulation

43
 Chapter Five

CONCLUSIONS AND RECOMMENDATIONS

5.1. CONCLUSIONS
Optical wireless (OW) communications is a growing popular field, hence for outdoors and
indoors applications a lot of research work is being done. This simulator has practical application
for testing of channel schemes that are being investigated. OW links are mostly attractive
because of their high data rate, unlicensed bandwidths, and low operational costs, however they
cannot replace RF links completely, especially since they can be severely affected by inclement
weather.
Therefore OW links are best when used together in combination with RF links in some kind of a
hybrid structure [5]. In this case, each system can undertake a role that best corresponds to their
capabilities, and they will both operate in conjunction with each other. Hybrid systems are
currently under a lot of research and this kind of a simulator where RF schemes are applied to
OW field aims to support that research.
The simulator framework created in this work utilizes advanced concepts of the Matlab, such as
object-oriented modular approach for the simulation of communication channel. These channels
are built as a composition of a sequence of building blocks that can be connected to each other
more or less independently. Creating the framework in this manner (as a composition of building
blocks), was one of the main challenges of this work, but allows for easy development and
inclusion of new modules/blocks. As a result, this framework can be a useful tool for deploying
different signal processing techniques on the channel, and for examination and analysis of the
results.

5.2. RECOMMENDATIONS
Further work on the simulator can be done by adding the modules that implement different
channel coding techniques together with modulation techniques, and also applying the newly
developed subcarrier modulation and multiplexing techniques. Also in the Framework, the
technique for switching modules (i.e. taking decisions on which modules are going to be used in
current simulation) can be enhanced and adapted for different usages. An additional channel

44
model such as fading channel can also be implemented in addition to the existing AWGN
channel model used.
As already mentioned, future work with the simulator could involve deeper investigation into the
hybrid RF and OW schemes, specifically determining and implementing a transition point when
one scheme should be used over the other.

45
 APPENDIX

PSD OF DH-PIM
(A) 1.1 Fourier Transform of DH-PIM

The mathematical model of a DH-PIM pulse train has been presented in Section 3. The truncated
transmitted signal xN(t) of N symbols is given as
N 1 
 2t  Tn  1   2t  Tn  3  
x N  V  rect     hn rect    A.1
n 0   Ts 2  Ts 2 

Where V is the pulse amplitude, hn {0, 1} indicating H1 or H2 respectively, n is the

instantaneous-symbol number and Tn is the start time of the nth symbol defined.

The Fourier transform of the truncated signal (4 B 1) can be written as

N 1    2t  Tn  1   2t  Tn  3    jxdt

X N    V   rect     hn rect     A.2
n 0   Ts 2  Ts 2 

Thus, the Fourier transform of the truncated DH-PIM signal can be written as

    
N 1
X N   
V  jTo
1    js / 2  1  hn   jTs / 2 e  jTs n  1e
 jTs
 A.3
j n 0

(B) 1.2 POWER SPECTRAL OF DH-PIM

The power spectral density of the signal can be obtained by averaging over a large number of
symbols N and then performing the limiting operation as given in

E X N  . X N  
P   lim N  A.4
E TN  T0 

46
Where E[x] is the expected value of x and XN(_) is the complex conjugate of XN(ω), the
expected value of (TN − T0) is given by

 2 m1  1
ETN  To   NTs 1     A.5
 2 

And
 N 1 N 1 
 
S N    E  1  hn e  jTs / 2 1  hq e jTs   A.6
 n 0 q 0 

Expression is the best evaluated by splitting it into three regions:

S N 1  , where qn
and then summing them up as S
S N 1    S N 1    S N 2    S N 3   A.7

Here, we set out to find the expressions for SN1(ω), SN2(ω) and SN3(ω) as outlined
In
1. Taking q <n

   
N 1 N 1
S N 1     E 1  hn e  jTs / 2 1  hq e jTs / 2 e  jTs A.8
n 1 q  0

47
 REFERENCES.

[1]. Z. Ghassemlooy, W. Popoola, S. Rajbhandari, Optical

WirelessCommunications System and Channel Modelling with MATLAB, Taylor

& Francis Group, 6000 broken Sound Parkway NW, Suite 300, Boca Raton,Book

Number-13: 978-1-4398-5235-4, 2013.

[2]. ShlomiArnon, John R. Barr, George K. Karagiannidis, Robert Schober, Murat

Uysal, Advanced Optical Wireless Communication Systems,Cambridge University

Press, New York, ISBN 978-0-521-19787-8 hardback, 2012.

[3]. NevenaDjaja, optical wireless channel simulation, Institute for Broadband

Communication, Graz University of Technology, October 2012.

[4]. C++ reference manual. http://www.cplusplus.com/reference/.

[5]. InternationalElectrotechnicalComission. Safety of laser products (iec608251).

[6]2007.http://webstore.iec.ch/webstore/webstore.nsf/ArtNum/037864.

48