Thesis 2

Modeling and Performance Evaluation of
a BPPM UWB System

(MSc thesis)
Hailiang Mei
(July 2003)
Delft University of Technology
Philips Research Laboratories Eindhoven

Modeling and Performance Evaluation of a BPPM UWB System MSc Thesis
Table of Contents
Table of Contents .................................................................................................................I
Table of Symbols .............................................................................................................. III
Table of Abbreviations....................................................................................................... V
Summary ...........................................................................................................................VI
Acknowledgements ..........................................................................................................VII
1 Introduction ................................................................................................................. 8
1.1 Overview of wireless communication systems ................................................... 8
1.1.1 3G and WLAN systems............................................................................... 8
1.1.2 For supporting higher transmission speed – UWB ..................................... 9
1.2 Problem Statement .............................................................................................. 9
1.3 Organization of this thesis................................................................................. 10
2 Overview of UWB .................................................................................................... 12
2.1 Definition of a UWB signal .............................................................................. 12
2.2 Features of UWB system................................................................................... 12
2.2.1 Extremely low signal power spectral density............................................ 12
2.2.2 Robust to multipath interference ............................................................... 13
2.3 Related work and recent progress ..................................................................... 14
2.3.1 Previous work............................................................................................ 14
2.3.2 FCC and ETSI regulation on UWB products............................................ 14
2.3.3 IEEE 802.15.3 standard............................................................................. 15
3 UWB Signal Analysis ............................................................................................... 16
3.1 Single pulse representation................................................................................ 16
3.1.1 Gaussian monocycle.................................................................................. 16
3.1.2 Raised Cosine pulse .................................................................................. 17
3.1.3 Pulse shape selection................................................................................. 18
3.2 Pulse train and pseudo random pulse location .................................................. 19
3.3 Modulation techniques ...................................................................................... 21
3.3.1 PAM .......................................................................................................... 21
3.3.2 OOK .......................................................................................................... 22
3.3.3 PPM........................................................................................................... 22
3.3.4 Antipodal Modulation (BPSK) ................................................................. 22
3.3.5 Other modulation schemes ........................................................................ 23
3.4 Selected modulation scheme ............................................................................. 23
3.5 Power analysis................................................................................................... 25
3.6 Transmission environment analysis and its effects on UWB signals................ 25
3.6.1 Multipath effects ....................................................................................... 25
3.6.2 Effects caused by related moving between Tx and Rx ............................. 26
3.6.3 Relocking on the most resolvable path(s) ................................................. 26
4 UWB Receiver Architecture ..................................................................................... 28
4.1 Correlation receiver (Matched filter) ................................................................ 28
4.2 Rake receiver..................................................................................................... 29
4.3 Processing gain factors...................................................................................... 31
4.4 Discussion ......................................................................................................... 32
4.4.1 The number of Rake fingers...................................................................... 32
I
4.4.2 Analog and digital circuit design .............................................................. 32

5 Pulse Detection and Synchronization Scheme .......................................................... 34
5.1 Problem discovered and existing methods for pulse detection ......................... 34
5.2 Proposed pulse detection method...................................................................... 34
5.2.1 Template signal match detection............................................................... 34
5.2.2 Received single pulse with a sinusoidal signal ......................................... 35
5.2.3 In-phase situation detection....................................................................... 37
5.2.4 Sinusoidal referenced pulse detection ....................................................... 39
5.2.5 Comparison between two detection methods............................................ 40
6 Evaluation by Simulation .......................................................................................... 42
6.1 UWB system simulation setup .......................................................................... 42
6.1.1 System simulation overview ..................................................................... 42
6.1.2 Variable sample rate simulation................................................................ 44
6.1.3 Factors related to the performance of simulation...................................... 44
6.2 Evaluation of pulse detection by simulation ..................................................... 45
6.2.1 Pulse detection simulation......................................................................... 45
6.2.2 Error analysis............................................................................................. 49
6.3 Evaluation of receiver architecture by simulation............................................. 51
6.3.1 Theoretical analysis................................................................................... 51
6.3.2 Effects of different Rake finger numbers.................................................. 52
6.3.3 Effects of timing errors ............................................................................. 53
7 Interference From 5 GHz WLAN System................................................................. 56
7.1 Review of 802.11a WLAN signal ..................................................................... 56
7.2 Analysis of UWB system performance with 802.11a interference................... 56
7.3 Solution to the interference problem................................................................. 57
8 Conclusions and Recommendations.......................................................................... 58
8.1 Conclusions ....................................................................................................... 58
8.2 Recommendations and further research topics.................................................. 58
9 Appendixes................................................................................................................ 60
9.1 Appendix A ....................................................................................................... 60
9.2 Appendix B ....................................................................................................... 61
9.3 Appendix C ....................................................................................................... 62
9.4 Appendix D ....................................................................................................... 62
References ......................................................................................................................... 64
II
Table of Symbols
f + B G% – The highest -10dB bandwidth frequency of UWB signal
f / B G% – The lowest -10dB bandwidth frequency of UWB signal
gm(t ) – Gaussian monocycle
GM (t ) – Fourier transformation of Gaussian monocycle
fF – Center frequency
rc (t ) – Raised Cosine pulse
RC (t ) – Fourier transformation of Raised Cosine pulse
f G% – The -6dB frequency of Raised Cosine pulse
p SHULRGLF (t ) – The periodic Gaussian monocycle pulse train
PSHULRGLF (t ) – Fourier transformation of the periodic Gaussian monocycle pulse
train
p UDQGRP (t ) – The randomly located Gaussian monocycle pulse train
PUDQGRP (t ) – Fourier transformation of the randomly located Gaussian
monocycle pulse train
p UDQGRP B %330 (t ) – The BPPM modulated Gaussian monocycle pulse train
(Transmitted UWB signal)
TI – The frame period
∆T M – The PN random position of the pulse within one frame
T330 B VKLIW – The BPPM time position shift (The length of BPPM time lag)
P7[ – Transmitted power of UWB signal
P5[ – Received power of UWB signal
SNRVLPXOWDQHR XV – Simultaneous signal to noise ratio of the received UWB signal at
the receiving antenna
r (t ) – The received UWB signal
s (t ) – The single received UWB pulse
aQ – The amplitude of the nth multipath component
τQ – The time delay of the nth multipath component
sWHPSODWH (t ) – The known UWB signal template pulse shape
Gain 5DNH – The processing gain of the Rake receiver
Gain 0DWFKHG B ILOWHU – The processing gain of the matched filter
RU B V (τ ) – Cross correlation between r (t ) and sWHPSODWH (t )
ref (t ) – The sinusoidal reference signal
fU – The frequency of ref (t )
cross _ corrV B UHI (τ ) – Cross correlation between s (t ) and ref (t )
corrU B UHI (t ) – Correlation between s (t ) and ref (t )
III
ref LGHDO B LQSKDVH (t ) – The ideal in-phase reference signal

ref D B LQSKDVH (t ) – The approximate in-phase reference signal
corrD B LQSKDVH (t ) – Correlation between s (t ) and ref D B LQSKDVH (t )
θ – The threshold value for finding the in-phase pulse
EG – The difference signal energy
EE – Signal bit energy
N – The noise power
PH – Bit error rate
IV
Table of Abbreviations
3G – Third Generation Mobile Network
ADC – Analog Digital Converter
AWGN – Additive White Gaussian Noise
BER – Bit Error Rate
BPSK – Binary Phase Shift Keying
CDMA – Code Division Multiple Access
EIRP – Effective Isotropic Radiated Power
ETSI – European Telecommunications Standards Institute
FCC – Federal Communications Commission
GPR – Ground Penetrating Radar
GPS – Global Positioning System
IEEE – Institute of Electrical and Electronics Engineers,
ISI – Inter Symbol Interference
MAC – Media Access Control
OFDM – Orthogonal Frequency Division Multiplexing
OOK – ON OFF Keying
PAM – Pulse Amplitude Modulation
PAN – Personal Area Network
PHY – Physical Layer
PN – Pseudo-random Noise
PPM – Pulse Position Modulation
PRF – Pulse Repetition Frequency
PSD – Power Spectral Density
QoS – Quality of Service
UWB – Ultra WideBand
Wi-Fi – Wireless Fidelity
WLAN – Wireless Local Area Network
V
Summary
UWB radio is a revolutionary wireless technology that offers a high data rate within a
short range. Therefore, UWB is regarded as a good PHY layer solution to the PAN
applications, e.g. PDA, mobile or wireless USB standards. Since UWB is a relatively new
technology in the area of commercial wireless communication, many issues were not
clear in the beginning of the project. Within this project, we have tried getting a better
understanding about UWB concepts and the following works have been done.
• Analyzed UWB signal and related modulation schemes.
• Presented a UWB receiver architecture that reduces the multipath effects and
noise for its performance. Calculated the processing gains of different stages in
the architecture. Verified various properties by the simulation.
• Proposed a new algorithm for UWB pulse detection, which is an essential
problem in UWB receivers. Studied the performance of this new method and
compared it with the template signal match detection.
• Studied the interference problem between the UWB signal and existing wireless
systems.
• Implemented a UWB system simulation platform in Matlab. This platform can be
used for analyzing a future UWB system design.
VI
Acknowledgements
This thesis describes the result of a Master of Science assignment at Philips Research
Laboratory Eindhoven, The Netherlands. This assignment has been carried out from
November 2002 to July 2003 at the Mixed Signal Circuits & Systems group for
completing the master degree in Electrical Engineering in the section of Wireless Mobile
Communications group in Delft University of Technology, The Netherlands.
I am grateful to people who have contributed towards shaping this thesis.
At the outset, I would like to express my appreciation to Dr. Raf Roovers and Ir. Gerard
van der Weide (Philips) for their advice during my master research endeavor for the past
eight months. As my supervisors, they have constantly enforced me to be in focus
towards achieving this goal. Their observations and comments helped me to establish the
overall direction of the research and move forward with investigation in depth.
I greatly appreciate my supervisor Dr. Gerard Janssen (TU Delft) for his critical
comments and suggestion during the whole thesis work, and for his generous time in
proof reading. As my supervisor at the University, his continuous supervision helped me
to keep this research project on the right track and achieve this final research thesis.
I would like to thank my colleague Ir. Harish Kundur (Philips) who helped me to set up
the simulations. He generously shared his time and knowledge in my work regarding the
theory as well as programming skills. He plays a major role in making me understand the
concepts and coaching me on the thesis writing.
I would like to thank my other colleagues at Philips Research and Philips Semiconductors
for the pleasant working environment, for helping me solve various problems and for
sharing their knowledge that influenced the work and resulted in a satisfactory result.
I acknowledge many friends in Delft and Eindhoven who have made my two-year study
so cheerful.
Last, but not least, I would like to dedicate this thesis to my family in China, for their
love, patience, and understanding.
VII
Chapter 1 Introduction
1 Introduction
1.1 Overview of wireless communication systems
Nowadays, wireless communication technology is developing rapidly worldwide, and its
landscape is changing drastically due to the emergence of new standards driven by the
rapid growth in information services provided by the Internet, for example the
multimedia applications including MP3, broadband data streaming, in particular video.
Some existing or under development wireless systems (3G and WLAN) are designed to
support this kind of multimedia data services and low quality video transmission.
1.1.1 3G and WLAN systems
In a 3G system, e.g. UMTS or CDMA-2000, user data rates can be provided at up to 2

Mbps in a stationary environment, while outside of a building or especially when moving
the supported data rates would be lower. Such throughput ability can support multimedia
data service or low quality video streams. The cell size of the 3G system is smaller than
the current 2G system, like GSM, and it is about 300 meters in urban area and can be
larger in rural area.
Compared to 3G, Wireless LAN (WLAN) may offer higher data throughput (Figure 1-1).
For example, Wi-Fi products (802.11b) are already available on the market offering user
data rates up to 11 Mbps theoretically and coverage up to 100 meters. Future WLANs
will provide data rates up to 54 Mbps theoretically (802.11a/g), and the new designed
MAC protocol can enable the system to support ad-hoc networking, synchronized
services, and dynamic link adaptation for QoS control. Hence, the entire WLAN system
can become a good platform for video transmission.
Figure 1-1: WLAN data throughput vs. range based on measurements[2]
8
1.1.2 For supporting higher transmission speed – UWB
In newer systems, obviously the data rate is going up and the applications of near-body
wireless communication are becoming important. However, the gap between the
transmission speed needs and the data rate that can be offered still exists. In Table 1, it is
shown that we need over 100Mbps wireless links capable of maintaining multiple
MPEG-2 streams or even higher quality video streams, which is the new requirement for
a home network or a personal area network (PAN). But, the existing systems like 3G or
WLAN cannot meet this specification. Therefore, a new wireless transmission standard is
required, which leads to the researched system of this thesis – UWB (Ultra WideBand)
system.
Display standards VGA SVGA XVGA SXVGA

Horizontal pixel count 640 800 1024 1280
Vertical pixel count 480 600 768 1024
Total pixel 307200 480000 786432 1310720
Total bits (16bit color) 4915200 7680000 12582912 20971520
Total bits (24bit color) 7372800 11520000 18874368 31457280
Mbps at a minimum animation of 30 147 230 377 629
frames per second (16 bit color)
Mbps at a minimum animation of 30 221 345 566 943
frames per second (24 bit color)
Mbps after compression 6 - 32 15 - 50 20 – 70 30 – 100
(Throughput)
Applications MPEG- Projector Laptop PC
2 DVD Projector monitor
Table 1: Desired data for video transmission
1.2 Problem Statement

The potential high data rates make UWB an interesting technology to investigate. The
UWB system is very different compared to conventional wireless communication
systems. Therefore, there are many challenges that have to be met when we are designing
such a system. Many of these challenges do not exist in conventional wireless systems,
and we see that issues like modulation techniques, synchronization methods and the
performance limits of the UWB receiver are important, and they will be highlighted in
this thesis.
• Modulation techniques. The choice of modulation scheme may influence the
signal spectrum properties and the performance of a UWB communication system.
Some schemes like PAM, OOK, PPM and Bi-polar have been proposed and can
be considered as potential candidates for this thesis project. What is a good
modulation scheme for a UWB system will be issued as an important problem.
9
• Synchronization method. Synchronization is one of the crucial tasks of a UWB

receiver, which obviously has its influence on the system performance. Generally,
we might have two solutions: analog and digital synchronization methods.
Although, the trend is to digitize the system as much as possible, it is clear that
pure digital solutions are still not applicable due to high computational
requirements and because a very high bandwidth ADC is needed which would
require too much power. Therefore, we need to design a simpler synchronization
scheme that can be implemented more easily. The required synchronization
scheme may utilize the features and advantages of both analog and digital
solutions.
• UWB receiver architecture and its performance evaluation. The performance of
the receiver is one of the primary considerations in the design of a communication
system. The performance of the UWB receiver architecture in a noisy
environment needs to be evaluated and compared with the theoretically maximum
receiver performance. Since it is possible that some other radio systems are
operating nearby the UWB system, an analysis of their interference is also
necessary since we need a robust system.
The topics listed above do not cover the entire picture of the UWB system. Issues like
channel measurements and modeling, security problems and UWB antenna design are
also very critical, but are out of the scope of this thesis.
1.3 Organization of this thesis

Within this thesis, several topics of the UWB system will be covered based on the FCC &
ETSI regulations and research results achieved so far. Some new methods and system
level designs are driven by the characteristics of the UWB signal.
In Chapter 2, an overview of UWB systems is given. Topics like the definition of UWB,
the features and related legal issues will be covered. The UWB signal representation,
modulation schemes and the multipath channel are addressed in Chapter 3. Chapter 4
describes the UWB receiver structure. Chapter 5 describes a new pulse detection method
and the corresponding synchronization scheme. In Chapter 6, a UWB system simulation
is introduced and some simulation results are discussed. Chapter 7 gives conclusions and
recommendations for further research.
10
11
Chapter 2 The Overview of UWB
2 Overview of UWB
2.1 Definition of a UWB signal
The term UWB stands for Ultra Wide Band, and it refers to a system or signal with an
extremely large bandwidth. Originally, UWB signals are defined as signals having a
fractional bandwidth of at least 0.25 or occupying at least 1.5GHz spectrum. In recent
literature [3], these two factors are modified to 0.20 and 0.5 GHz respectively. The
fractional bandwidth η is defined as:
f + B G% − f / B G%
η = 2⋅ (2-1)
f + B G% + f / B G%
where f + B G% and f / B G% represent the highest and lowest -10dB bandwidth frequencies
of the signal spectrum respectively.
2.2 Features of UWB system

2.2.1 Extremely low signal power spectral density
Since the bandwidth of UWB signals is much wider than that of conventional wireless
systems, a higher channel capacity can be achieved even in a low SNR environment.
According to Shannon’s theorem:
S
C = B log (1 + ) (2-2)
N
where C is the channel capacity, B is channel bandwidth, and S/N is the signal to noise
power ratio at the input to the receiver. For example, a UWB system that utilizes a 2 GHz
spectrum operating in 0dB SNR, its channel capacity can be calculated as
C = 2 ⋅ log (1 + 1) = 2Gbps . Based on this result, we can observe that a UWB system with
low signal power may still maintain a high data rate, and this feature will allow UWB to
be a good PHY layer solution for PAN equipments.
Because of the low signal power and the available large bandwidths, UWB systems
perform like spread spectrum systems. However compared to the more common forms of
spread spectrum like frequency hopping and direct sequence systems, UWB does not rely
on a spreading sequence and a hopping sequence to generate the wide bandwidth signals.
Instead, it is the extremely short duration of the UWB basic pulse that gives the system
its ultra-wide bandwidth.1
Compared to other narrow band communication systems, which operate in the

bandwidth-limit regime, UWB works in the power-limit regime (Figure 2-1). Therefore,
UWB signal power in any single narrow frequency channel is very small and the
1
This property is verified in chapter 3
12
interference to any other existing products like 802.11a terminals and 3G mobile phones
can be ignored in principle.2
Transmitted power
Conventional radio signal,

e.g. 3G or WLAN
Conventional radio signal,
UWB signal
Frequency
Figure 2-1: Bandwidth-limit regime (conventional wireless system) vs. Power-limit
regime (Ultra wideband radio system)
2.2.2 Robust to multipath interference
UWB is robust to the effects of multipath interference. Because the signal bandwidth of
UWB is much larger than the coherence bandwidth of the multipath channel, any
frequency-selective fading only affects a small portion of the signal spectrum for any
channel realization. The interval time between two continuously transmitted UWB pulses
determines the maximum observable multipath delay. Hence the multipath components
with differential delay larger than half the pulse width and within one period of the
periodic pulse transmission can be measured unambiguously.
Since the time resolution is approximately equal to the reciprocal of the bandwidth, the
use of signals with GHz bandwidth means that multipath is resolvable down to path
differential delays in the order of a nanosecond or less, i.e., down to path difference in the
order of 0.3 meters or less. This significantly reduces fading effects even in indoor
environments. Lack of significant multipath fading may considerably reduce fading
margins in link budgets and allow for a low transmission-power operation. The capability
to highly resolve multipath combined with the ability to penetrate through materials
makes UWB technology valuable for high-quality, fully mobile short-range indoor radio
systems.
2
In case multiple UWB equipments are operating within the same small area, the accumulated UWB signal
power may have an effect upon other wireless systems. The exact influence level needs to be found out in
order to estimate the potential problem.
13
2.3 Related work and recent progress

2.3.1 Previous work
Due to its perfect path resolving and ranging properties, UWB signals have been used in
Radar Technology for a long time already. The techniques for generating UWB signals
have existed for more than three decades. Perhaps it is more readily known to the radar
community within its time domain description as “baseband carrierless short pulse”
techniques. A comprehensive reference of early works in this area can be found in [4].
In the late 90’s, some research institutes and companies started to apply the knowledge of
UWB (or impulse radio) to the wireless communication area. Much work has been done
so far and some early results have been achieved. The representation of the UWB signal
and its modulation techniques are described in paper [12] and [14]. The pulse shape that
can be used in the UWB system is discussed in [6]. The results of indoor and outdoor
channel measurement are shown in papers [17], [18] and [19].
2.3.2 FCC and ETSI regulation on UWB products
For the legal use of UWB products, FCC released initial rules on February 14, 2002. The
main concern is about interference with existing radio systems. The spectrum that the
FCC has proposed for UWB devices now is also being used by 5GHz WLAN, GPS
(Global Positioning System) and GPR (Ground Penetrating Radar). Therefore, the
allowed transmission power for UWB equipment is restricted to a very low level to avoid
interference to those systems, as shown in Figure 2-2. As one might expect, when the
FCC announced that it was considering the rules for UWB to operate within a spectrum
that had already been licensed, a number of companies raised objections. The problem of
coexistence will be addressed in detail during chapter 6. UWB proponents believe that
UWB pulses will only cause negligible interference with narrowband applications,
because the pulses operate at a low power level.
14
Figure 2-2: FCC and ETSI UWB specification
2.3.3 IEEE 802.15.3 standard
The IEEE 802.15 Task Group 3 (TG3a [7]) for Wireless Personal Area Networks
(WPANs) is defining a standard to provide a higher speed UWB PHY enhancement
amendment to 802.15.3 for applications, which involve imaging and multimedia. Based
on the FCC regulation, 23 alternative proposals of UWB PHY layer solution to this new
standard were made by different UWB companies and research groups till 5th May 2003,
and the finalized standard is expected at the end of 2004.
15
Chapter 3 UWB Signal Analysis
3 UWB Signal Analysis

After discussing basic characteristics of UWB and its properties in the time and
frequency domain, we will continue with the UWB signal analysis in this chapter, where
the pulse shape, modulation techniques and the multipath effects are highlighted.
3.1 Single pulse representation

In order to comply to the FCC specification, two kinds of pulse shapes are investigated:
the Gaussian pulse and the Raised Cosine pulse. The derivation and the comparison of
these two pulse shapes are given below.
3.1.1 Gaussian monocycle
A Gaussian monocycle is a wide-bandwidth signal, with its center frequency and its
bandwidth related and dependent on the monocycle's width. In the time domain, the
Gaussian monocycle is mathematically similar to the first derivative of the Gaussian
function [2]. It has the form:
t −τ
W

gm(t ) = e (3-1)
τ
where τ is a time constant that determines the monocycle’s duration3. In the frequency
domain, a Gaussian monocycle’s Fourier transformation is:

GM ( f ) = − j ⋅ π ⋅ π ⋅ f ⋅ τ ⋅ e −π I τ
(3-2)
The center frequency f F should satisfy:

2 f Fτ π = 1 (3-3)
We can see that f F is proportional to the inverse of τ and the –3dB bandwidth is about
116%4 of f F . Thus, for example for a τ = 0.033ns monocycle, the center frequency is
about 6.85 GHz and its 3dB5 bandwidth is approximately 7.5 GHz. Figure 3-1 shows a
typical waveform of the Gaussian monocycle pulse and its spectrum.
3
Pulse width is defined as ≈ 5τ , which is 0.165ns when τ = 0.033ns.
4
The property that the center frequency is proportional to the inverse of the pulse duration and the value of
116% are proved in Appendix A.
5
The pure Gaussian pulse does not fit the FCC rules well. Additional modification is required in the
realistic implementation, e.g., filtering the Gaussian pulse before sending it into the air.
16
Figure 3-1: Gaussian monocycle pulse and its spectrum
3.1.2 Raised Cosine pulse
In the FCC specification, the allowed PSD mask for UWB signal is a rectangular shape.
It is obvious that the Gaussian shape pulses do not match with this rule perfectly.
Therefore, the Raised Cosine pulse is introduced to provide a better matching with the
FCC mask, which is as shown in Figure 3-2. The Raised Cosine pulse can be described in
the frequency domain as:
 1, f < f
   π ( f − f  
  f < f < B
H ( f ) =  1 + cos  , (3-4)
    2 f  
∆
f >B
 0,
where B is the absolute bandwidth. f and f ∆ are given as:
f ∆ = B − f G%
f = f G% − f ∆
where f G% is the -6dB frequency of the raised cosine pulse. For the aim of utilizing the
whole 7.5 GHz FCC approved bandwidth, the value of f G% is set to 3.75 GHz.
Its corresponding time domain waveform is calculated as:

sin 2πf G% t  cos 2πf ∆ t 
h(t ) = F − [H ( f )] = 2 f G% ( )  (3-5)
2πf G% t 1 − (4 f ∆ t ) 
Since h(t ) is the low frequency band signal ( − f G% , + f G% ), it needs to be shifted to the
desired frequency band. For example, if the Raised Cosine pulse spectrum utilizes the
whole FCC approved frequency band, the baseband pulse should be up converted to the
central frequency f F (6.85GHz). Therefore, the transmitted pulse will be
rc (t ) = h(t ) ⋅ cos(2πf F t ) (3-6)
17
rc (t ) and its energy spectrum RC ( f ) are shown in Figure 3-2.
1
2 f
Figure 3-2: Raised Cosine pulse and its energy spectrum
3.1.3 Pulse shape selection
By comparing the two proposed pulse shapes, we need to select one to be used for the
UWB system. We discover that although the Raised Cosine pulse might have nice
spectral properties, which fit FCC approved rectangular mask very well, it is hard to
generate by a simple circuit. Because in the time domain, Raised Cosine pulses have a
non-realistic side lobe and building such a pulse is too difficult for the circuits. On the
other hand, the Gaussian monocycle is relatively simpler to generate, and therefore it is
chosen as the pulse shape used in this project.
At the receiver’s end, pulse shaping caused by an antenna that can be considered as a
filter with impulse response matched to the transmitted pulse. If the transmitted pulse is
the Gaussian monocycle (1st derivative Gaussian pulse), we can expect the 2nd derivative
Gaussian pulse as the ideal received pulse (Figure 3-3). In the following chapters of this
thesis, the 2nd derivative Gaussian pulse s (t ) is used to represent the received UWB pulse
and the template pulse at the receiver side.
18
Figure 3-3: Second (received) derivative Gaussian pulse and its spectrum
3.2 Pulse train and pseudo random pulse location

After introducing some UWB pulse shapes, the transmission of pulses – the pulse train
needs to be taken into account. Initially, we can generate the Gaussian monocycles at
repetitive timing points, which is shown in Figure 3-4 and this pulse train can be
described as:
Q
p SHULRGLF (t ) = ∑ gm(t − ( j − 1)T ) I

(3-7)
M =
where T is the frame period which is equal to the interval between two continuous
I
pulses, and gm(t ) is the Gaussian monocycle. Such a periodic pulse train has a fixed
pulse repetition frequency 1 / T . Based on the energy spectral density of the Gaussian
I
monocycle GM ( f ) (3-2) and the power spectral density (PSD) formula for
cyclostationary pulse signal in [20], we can obtain the PSD for the periodic pulse train:

σ µ +∞
k k
∑

PSHULRGLF ( f ) = GM ( f ) + GM ( ) δ ( f − ) (3-8)
I
T T I = −∞
N
T I
T I
where µ and σ are the mean and variance of the available amplitude of the pulse, in
the case of periodic pulse train, they are 1 and 0 respectively. Therefore, we have:

+∞
1 k k
PSHULRGLF ( f ) =
T N
∑
= −∞
GM ( ) δ ( f − )
T T
(3-9)
I I I
We observe that in the frequency domain, the highly regular monocycle pulse train
produces power spikes at regular intervals of 1 / T . This is also verified by the simulation
I
as shown in Figure 3-4. Thus, the already limited low UWB signal power spreads among
the spectral lines. But if the PSD is flat, then the transmission power can be maximized
under the FCC specification. We can conclude that this periodic pulse train does not
utilize the spectrum resource efficiently and needs to be improved.
19
Figure 3-4: Periodic pulse train and its spectrum
For smoothing the spectral lines, which is the key to optimize the spectrum usage, we
need to remove the obvious correlation of the periodic pulse train in the time domain. A
good method for it is making the pulses appear randomly in time, thus too much
correlation can be avoided in the pulse train and the spectral lines can be suppressed. This
method is shown in Figure 3-5 and the pulse train with randomly located Gaussian
monocycles can be described as:
Q
p UDQGRP (t ) = ∑ gm(t − jT − ∆T )
I M
(3-10)
M =
where ∆T is the random number within (0, T ). The PSD of this pulse train can be
M I
obtained as:
1
PUDQGRP ( f ) = GM ( f ) (3-11)
T I
It is shown that PUDQGRP ( f ) is just a scaled version of the energy spectrum of the Gaussian
monocycle, and there are no spikes in it. This is because all pulses appear purely random
and there is absolutely no correlation information in the pulse position. Such a property is
verified by the simulation in Figure 3-5.
Figure 3-5: Random positioned pulse train and its spectrum
20
∆T is a relatively large time offset (many nanoseconds) that can be applied to each pulse.
M
The dataset of { ∆T } can be designed as a Pseudorandom Noise (PN) codes. In a

M
multiple access system, each user could have a unique PN code sequence. Only a receiver
operating with the same PN code sequence can decode the transmission. Without
knowledge of the unique time-hopping code, the signal is virtually undetectable, even at
very close proximity to the transmitter.
3.3 Modulation techniques

In order to transmit information, additional processing is needed to modulate the
monocycle pulse sequence. The desired modulation technique needs to provide the best
error performance (BER) for a given energy per bit. As we compare different modulation
schemes, we examine the power efficiency, the effect of the modulation on the PSD and
also its robustness against interference of wireless links. Specifically, some modulation
approaches produce spectral lines, thus further limiting total transmit power to stay
within FCC limits on signal PSD, leading to a lower average power solution.
Several modulation techniques can be used to create UWB signals, some are more
efficient than others. In its formative years, some of the most popular methods to create
UWB pulse streams used mono-phase techniques such as Pulse Amplitude Modulation
(PAM), Pulse Position Modulation (PPM), On-Off Keying (OOK) or Antipodal
Modulation (BPSK). In these techniques, a “1” is differentiated from a “0” either by the
amplitude of the signal or its time of arrival – but all the pulses have the same shape.
The following modulation methods, PAM, OOK, PPM, BPSK, are investigated in the
following paragraphs.
3.3.1 PAM
PAM works by separating the “large” and the “small” amplitude pulses. By varying the
amplitude the receiver can tell the difference between “1” and “0,” and thereby decode
the data from the received signal.
Figure 3-6: PAM modulation [9]
But in the wireless channel, attenuation is a significant problem. Due to that, the receiver
will need the attenuation gain control. Therefore, this modulation is not very popular for
implementation.
21
3.3.2 OOK
In OOK (On Off Keying), a “1” is a pulse and an absence of a pulse is a “0”. Actually, it
is a special case of PAM where the amplitude of zero represents a “0”.
Figure 3-7: OOK modulation
3.3.3 PPM
In PPM, all pulses (both “1”s and “0”s) are of the same amplitude. The receiver
distinguishes between a “1” or a “0” by its time of arrival, or the time lag between pulses.
In this case, a positive time lag could mean a “1” and a negative time lag a “0”. Since the
multipath channel is causal in time, the difference in time lags is kept by such an
integrality. PPM modulation will also smooth the spectrum of the signal, thus making the
system less likely to interfere with conventional radio systems. PPM allows the use of an
optimal matched filter receiving technique. The receiver uses a correlator to find the
signal well below the ambient noise level.
Figure 3-8: PPM modulation
Compared to OOK and PAM, PPM has its advantages on synchronization and good
ability against attenuation effects. But because the transmitted information is contained in
the time lag, a higher timing accuracy is required for the system.
3.3.4 Antipodal Modulation (BPSK)
Antipodal modulation is that a binary “1” is represented by a positive pulse and binary
“0” is represented by a negative pulse. There is a 180 degrees polarity shift between
pulses of “0” and “1”. From the constellation diagrams [10], we can see a significant
property of BPSK. A BPSK is an antipodal signaling technique and has the greatest
distance for equal bit energy. This difference leads to a 3dB benefit in efficiency: to
achieve the same bit error rate, PPM or OOK must use a double bit energy, or a 3dB
higher EE .
22
Figure 3-9: BPSK modulation
In the multipath scenario, some of the pulses arrive at the receiver with reflections in the
channel. Therefore, they may have a polarity shift of 180 degrees compared to the
original transmitted pulses. Thus, the transmitted data “0” could be detected as “1” and
vice versa. If the transmission environment contains too many reflections, the
communication can only work with precise channel estimation, which is complicated for
a high speed UWB signal.
3.3.5 Other modulation schemes
Bipolar Modulation is similar to OOK, but a binary “1” is represented by alternately

positive or negative pulses. Because of the alternating polarity pulses for “1”s, this
scheme has error detection capability. But on the other hand, a long string of “0”s may
result in loss of synchronization. Chirp Modulation is also proposed by some researchers,
but the critical point of it is the design and realization of relevant SAW devices with
adequate production tolerance and temperature drift. Therefore, this modulation
technique is not covered by this thesis.
3.4 Selected modulation scheme

After introducing the several different modulation techniques such as PAM, OOK, PPM
and BPSK, we need one scheme that will be used for the system simulation and receiver
architecture design. PPM will be chosen as the modulation scheme within this project
although it is not the most power efficient method compared to BPSK in the constellation
diagram.
There are three concerns in terms of making this selection. Firstly, because the multipath
channel is causal in time, PPM can utilize this feature very well to be tolerated with the
reflective wireless channel and as well has good spectral properties. On the other hand,
BPSK modulation encodes the transmitted data in the pulse polarity that can be distorted
by the reflective wireless channel easily. Secondly, since Philips is working on accurate
timer implementation, which is important for building a real system, the problem of
timing accuracy can be more or less solved. The third reason of choosing PPM is that it is
also used by Time Domain Inc, one of first companies who are working on commercial
UWB products. Time Domain Inc already made some successful UWB communication
chipsets based on this modulation scheme, which indicates that PPM is acceptable as a
mature modulation scheme.
Binary-PPM modulation uses long sequences of monocycles for communications, and

data modulation and channelization is accomplished by varying the pulse position in each
23
frame. When transmitting such sequences, it is vital to ensure that the spectral integrity of
the transmissions remains intact.
T I
T I
T I
∆T1 ∆T2 ∆T j
t
Reference pulse
“0” pulse
“1” pulse
T330 B VKLIW T330 B VKLIW
Figure 3-10: Random positioned pulse train with BPPM modulation
The Gaussian monocycle pulse train modulated by BPPM can be described as:
Q
p UDQGRP B %330 (t ) = ∑ gm(t − ( j − 1)T I − ∆T M + d ⋅ T330 B VKLIW ) (3-12)
M =
T is the frame period. ∆T is the PN random position of the pulse within one frame,
I M
which is in the range of ( T330 B VKLIW , T − T330 B VKLIW ). d is the transmitted binary data,
I
“0”(can be “-1” too) or “1”. T330 B VKLIW is the time lag or time position shift constant. The
referenced pulse position is given by p UDQGRP (t ) . These parameters can be seen in Figure
3-10.
More specifically, the value of T indicates the symbol rate. If T is 100 ns, the symbol
I I
rate will be 10 Megabits per second. By decreasing T , the symbol rate can be increased,
I
but meanwhile the possibility of ISI (Inter Symbol Interference) will also increase.
Consequently, the value of T cannot be chosen too small, unless the distance between
I
Tx and Rx is very small too. Instead of adjusting the symbol rate to change the
transmission speed, we can also increase the data rate by using M-ary PPM. Thus, more
information bits can be included in one symbol.
{ ∆T } is a PN code sequence. For one communication link, the pair of transmitter and
M
receiver is operating with the same PN code. Thus the receiver will know when to detect
the right data (pulse) based on the PN code, and other receivers who do not know this PN
code can not retrieve the data from the link.
We find that once a pulse is detected by the receiver precisely, the following pulses can
be grasped by the correlator based on the PN code easily. Therefore, the key to solve the
24
problem of UWB receiver design is how can we detect the right time position of the pulse
accurately enough.
3.5 Power analysis

In the FCC specification, the power limit for the frequencies from 3.1 GHz to 10.6 GHz
is –41.25dBm/MHz. Since the working frequency band is 7.5GHz bandwidth in total, the
upper limit of the transmission power P7[ of UWB system can be calculated as:
P7[ = −41.25dBm / MHz ⋅ 7500 MHz = −2.5dBm = 0.56 milliwatts (3-13)
Based on the value of transmitted power, we can calculate the received power P5[ in
UWB receiver by using the UWB channel model. A practical UWB channel model was
proposed by Philips Research Redhill based on channel measurements. It indicates that
while increasing the distance between Tx and Rx, the received UWB signal power will
decay by e to the power –1.9 for LOS environment and e to the power of –3.4 for non-
LOS environment:
For LOS: P5[ = e − G ⋅ P7[

For non-LOS: P5[ = e − G ⋅ P7[

where d is the distance between Tx and Rx in meters.
The power budget of a UWB system is shown in Table 3-1, and its detailed derivation
can be referenced in Appendix B.
UWB signal bandwidth 7.5 GHz (3.1 GHz – 10.6 GHz)

FCC signal power limit -41.25 dBm/MHz
Maximum transmitted power -2.5 dBm (0.56 milliwatts)
Distance between Tx and Rx 10 meters
Received power (Using –1.9 model) -115.0341 dB
Noise power -98.2900 dB
Duty cycle factor 0.01 (20 dB gain for received power)
SNRVLPXOWDQHR XV 3.2559 dB
Table 3-1: Power budget of UWB system
3.6 Transmission environment analysis and its effects on UWB

signals
3.6.1 Multipath effects
Multipath is one problem that is faced by almost all wireless communication systems. It
is the result of the receiving antenna picking up reflected signals from the transmitter as
well as the signal that travels directly to the receiver. If a single UWB pulse (Gaussian
monocycle) is transmitted, the received signal will contain a direct path signal and
25
reflected signals. Such a received signal r (t ) without the consideration of noise can be
represented by:
/
r (t ) = ∑
Q
a Q s (t − τ Q )
=
(3-14)
where τ < τ < ….< τ Q . The parameters τ Q and a Q are the time delay and the amplitude
of the nth multipath component, among which τ and a stand for the direct path. The
waveform s (t ) denotes the single received UWB pulse. L is the number of multipath
signals.
As we know, the pulse width of s (t ) is typically less than 200ps. Therefore, the multipath
components with differential delays larger than the pulse width can be measured
unambiguously. We can conclude that UWB system can be very robust to the effects of
multipath interference.
However the long time delayed multipath components a Q s( t − τ Q ) may cause ISI
problem especially when τ Q > T . To prevent the ISI problem, T cannot be chosen as a
I I
too small value when the transmitting pulse train is built.
3.6.2 Effects caused by related moving between Tx and Rx
A normal indoor office can be simulated as a high-density multipath environment, and

one problem we are going to face is the effect on the propagation channel by the moving
objects.
The highest speed of a moving object can be set as 2m/s, which is quite reasonable in
office. By simple geometric analysis, we can see the change in path lengths led by a 2m/s
moving object within 100ns should be smaller than 4e-7 meters. Since the speed of
microwave signal is 3e8m/s, the time shift within 100ns will be 4e-7/3e8, which is around
1e-15 seconds, 1e-6ns. Compare to the pulse width of 0.1ns, the time shift resulted by
moving object in office can be neglected in a short time period.
If we observe the system in a long time period, the moving object can cause the problem
that the locking on one or several resolvable UWB signal paths is lost at the receiver.
Therefore, a mechanism for path tracking is introduced in next section.
3.6.3 Relocking on the most resolvable path(s)
If the wireless transmission channel makes a significant change, which means the
resolvable path was lost tracking. The pulse detection for locking on the right path needs
to be redone. The problem here is how to detect loss of lock to the strongest path.
A good suggestion would be doing the pulse detection in parallel with the demodulation.
When the pulse detection observes the strongest path shifted too much, for example
exceeds a certain threshold value, it will start adjusting the timing information of input
26
UWB pulses to assure that the template pulse at the receiver always overlapping with the
strongest input pulse (most resolvable path). Later, the demodulation will work with the
new template signals, and importantly, still under the monitoring of the pulse detection.
27
Chapter 4 UWB Receiver Architecture
4 UWB Receiver Architecture

Building the UWB receiver is a complex task, which involves the UWB receiving
algorithm design as well as the circuit implementation. This chapter mainly focuses on
the UWB receiver architecture.
4.1 Correlation receiver (Matched filter)

First of all, the general structure of the UWB receiver is introduced. According to the
definition [1], the matched filter is the model of the optimized receiver. The matched
filter might be realized by correlating the input signal with the desired received signal.
W
r (t ) = ∫ r (t ) ⋅ sWHPSODWH (t )dt
W −7
(4-1)
where sWHPSODWH (t ) is the known template pulse (which is the same as s (t ) ) and r (t ) is the
input signal at the receiver.
Output
r (t ) r (nT )
r (t ) Integrator Sample and I
()
hold
Reset
sWHPSODWH (t )
Clock
Figure 4-1: Correlation receiver (Matched-filter) for UWB
The matched filter for UWB signaling is illustrated in Figure 4-1, where a correlation
processor using an integrate-and-dump filter is shown. This is also the basis structure of
the correlator that will be used in the later mentioned Rake receiver.
2⋅ T330 B VKLIW
Figure 4-2: BPPM modulation template pulse sWHPSODWH (t ) − s WHPSODWH (t − 2 ⋅ T330 B VKLIW )
28
For the BPPM modulated UWB signal, the template pulse sWHPSODWH (t ) is modified as
shown in Figure 4-2. This template pulse can be studied with the BPPM modulation
scheme that is defined in Chapter 3. The first half of the template pulse will lead to a
negative correlation result with the received pulse “0”, and the second half of the
template pulse will lead to a positive correlation result with the received pulse “1”. Thus,
the UWB signal can be demodulated by this BPPM template pulse. After sampling at the
end of each frame, binary data sequence m ~ (t ) can be produced, which stands for the
demodulated transmitted data in the receiver. m ~ (t ) can be compared with the input
binary serial data m(t ) to study the performance (BER vs. SNR) of the UWB receiver.
4.2 Rake receiver

The detection of signals in a multipath environment leads to a Rake receiver. The typical
Rake receiver consists of a bank of correlators, each with a different delay, so that the
multipath signal can be demodulated. For its characteristic, the Rake receiver is widely
used in spread spectrum systems, which gives resistance to ISI (Caused by multipath
transmission) through path diversity. For instance, conventional CDMA systems use a
Rake receiver to enhance the detection performance.
In UWB systems, the idea of Rake combining can also be utilized to detect signals in the
multipath environment. Since we are aiming at single user environment, only one
correlator is integrated with Rake receiver in the simulation. The implementation of the
Rake structure here is mainly for the purpose of collecting energy from the multipath
signals. The structure of the UWB Rake receiver is shown below in Figure 4-3.
Figure 4-3: The structure of UWB Rake receiver (5 fingers) with one correlator
Compared to the correlation receiver, the UWB Rake receiver traces signals from several
different paths and adds all the energy together to achieve a better pre-correlation signal.
The paths (the received pulses) selected by Rake receiver should be optimal, which
29
means they are the most resolvable paths6. (How to get the most resolvable paths can
refer to the pulse detection in Chapter 5). The minimum time separation between Rake
fingers is set to 250ps as a reasonable physical limitation for receiver hardware [23].
As shown in Figure 4-3, the Rake fingers are implemented by delay lines. If the delay
lines are setup correctly, five most resolvable pulses can be collected at the outputs of
delay lines at the same time. The five delayed signals are shown in Figure 4-4.
τ
Figure 4-4: Snapshot of signals without noise in the UWB Rake receiver. Signal 1 is
the first delayed signal of the original input UWB singal; Signal 2 is the delay
version of singal 1, and the similar with signal 3, 4, 5; Signal 6 is the combination of
those five delay signals (1, 2, 3, 4, 5). (Noise free)
Arrows of “a, b, c, d and e” in Figure 4-4 represent the five best resolvable paths. After
passing through delay lines in the Rake receiver, these five paths are summed at time
point τ . Their energy can be accumulated to create a better UWB pulse, which is arrow
“s” in signal 6.
The performance enhancement by Rake structure is more obvious with the noise (Figure
4-5). Because adding several paths of signal together could average the noise effect.
Therefore, the output SNR of Rake structure is increased and the performance of Rake
receiver is also better than the pure UWB correlation receiver.
6
“The resolvable” means the pulse should have a good shape and reasonable energy
30
τ
Figure 4-5: Snapshot of signals with band limited noise in UWB Rake receiver.
Signal 1 is the first delayed signal of the original input UWB singal; Signal 2 is the
delay of singal 1, and the similar rules with signal 3, 4, 5; Signal 6 is the combination
of those five delay signals (1, 2, 3, 4, 5).
It can be seen that Rake receiver has a good performance in multipath environment, and
can reduce the ISI problem drastically. But it is also obvious that Rake receiver has a
higher requirement on the timing accuracy, which is the correct time of arrival of the
resolvable pulses. Since the UWB pulse is extremely narrow, the timing requirement is
even more critized. Therefore, study on performance effects made by timing error is
necessary for designing the UWB system.
4.3 Processing gain factors

There are two SNR gain factors regarding the proposed Rake receiver structure. One is
the processing gain due to multiple Rake fingers and the other one is the processing gain
due to the matched filter.
In the Rake receiver as shown in Figure 4-3, the combination of 5 resolvable pulses can
be used as the input of matched filter. If these 5 resolvable pulses are of equal amplitude
and they are absolutely not overlapping with any other received pulses, the signal energy
of the summation pulse will be 25 times the single pulse and the noise energy will be 5
times the single pulse. In such a case, the processing gain on SNR can be obtained as:
Increased signal power
Gain 5DNH = 10 log( )
Increased noise power
31
= 10 log(25 / 5) = 7 dB (4-2)
It is obvious that this processing gain will be 0dB if only one Rake finger is used.
The matched filter also has the processing gain on SNR, and this is defined as [6]:
S S
  = 2 ⋅ T ⋅ BW   (4-3)
 N  RXW N
LQ
where T is the correlation signal duration and BW is the measured noise bandwidth.
Further, we can obtain the processing gain of the matched filter as:
Gain 0DWFKHG B ILOWHU = 10 log(2 ⋅ T ⋅ BW ) (4-4)
As a conclusion, the combination of these two gain factors will increase the SNR at the
output of the matched filter compared to the input SNR. This influence should be taken
into account when we further discuss the receiver performance.
4.4 Discussion
In this chapter, the correlation receiver and the Rake receiver for UWB systems are
investigated. Among the whole discussion, some major points need to be highlighted
again.
4.4.1 The number of Rake fingers
Theoretically, more Rake fingers means a better overall performance because more UWB
signal paths can be tracked. But in fact, more delay lines, which are the main part of Rake
finger, are costly and also can lead to more timing error problem. Therefore, we do not
suggest implementing too many Rake fingers in the receiver. The suitable number should
be obtained based on sufficient further channel measurement.
4.4.2 Analog and digital circuit design
Although circuit implementation is not the focus of this thesis, some issues related to
circuits can still be considered. First, the mixer has an important role namely to correlate
the amplified incoming UWB signal with the template signal. Thus, it must have
broadband input characteristics and a good linearity. Designing such a mixer with high
requirement will be a challenge topic. Another problem is how to divide the digital end
and analog end in the UWB receiver. Since nowadays, the speed of ADC is not high
enough to produce sampled UWB data at the minimum required sampling rate, we cannot
make the whole system digital. Therefore, much of the processing has to be done in the
analog domain, e.g. the mixers and integrators, as shown in the following figure.
32
W +7
( jΩ)
HBPF LNA ∫W
I
S-H A/D
To DSP
sWHPSODWH (t ) Clock
Figure 4-6: Analog implementation of UWB receiver
33
Chapter 5 Pulse Detection and Synchronization Scheme
5 Pulse Detection and Synchronization Scheme

5.1 Problem discovered and existing methods for pulse
detection
The fine time resolution of UWB signals enables the system to be robust against
multipath fading, and therefore theoretically, the received pulses from multipath can be
well resolved. But in the real channel in an indoor environment, most of these received
pulses are found to be overlapping. It is also noticed that the multipath time-spread in
many channels of interest is often 100 to 1000 times the pulse width of the UWB pulse
width.
Because many received UWB pulses are distorted seriously due to fading and
overlapping between each other, they are not easily applicable for the receiver. Therefore,
a mechanism is required to find the “best” one or several resolvable pulses among all
received multipath pulses. The “best” means that the pulse should have a good shape and
reasonable energy: A good shape implies this pulse is not distorted by other pulses and
still has a high correlation with the template signal; a reasonable energy implies for a
given pulse width, the peak amplitude of the pulse should be larger than a certain
threshold level. The best resolvable path is the strongest path in most of the case, but it
might not be the direct path or one of the first arriving paths. Based on these resolvable
pulses (or paths) and the known UWB modulation technique, the receiver can be
synchronized and the transmitted data can be retrieved from the received signal.
Since UWB technology already exists for several decades, some pulse detection
techniques have been proposed by different research groups. The methods like leading
edge detection, template signal match detection, and correlation detection can be found in
Barrett’s paper [4]. For a dense multipath environment, another interesting detection
method named ToA (Time of Arrival) measurement algorithm was introduced by Scholtz
in [15]. However, they either have high computational requirements or are limited by the
operating conditions. Therefore, another simpler pulse detection method for
synchronization is proposed in this thesis and its performance is analyzed.
5.2 Proposed pulse detection method

5.2.1 Template signal match detection
When a single UWB pulse (1st derivative Gaussian pulse) is transmitted, the received
signal is composed of the direct path signal, reflected signals, noise, and interference (we
will assume interference to be absent). So the received signal r (t ) can be represented by
/
r (t ) = ∑
Q
a Q s (t − τ Q ) + n(t )
=
(5-1)
where τ < τ < ….< τ Q . The parameters τ Q and a Q are the time delay and the attenuation
factor of the nth multipath component, among which τ and a stand for the direct path.
34
The waveform s (t ) denotes the received single pulse, which is assumed to be the second
derivative of the Gaussian pulse (refer to Chapter 3). L is the number of multipath signals.
The noise n(t ) is assumed to be AWGN, and interference is assumed to be absent.
In the conventional pulse detection methods, for example template signal match detection
(Matched Filter), the received signal r (t ) is evaluated based on its cross correlation
RU B V (τ ) with the template pulse sWHPSODWH (t ) . sWHPSODWH (t ) is the same as the expected
received signal pulse s (t ) in principle.
+∞
R U B V (τ ) = ∫ r (t )s
−∞
WHPSODWH (t + τ )dt (5-2)
High correlation values occur when s WHPSODWH (t + τ ) is in-phase overlapping with the
resolvable multipath components in r (t ) . Therefore, the timing positions of the
resolvable multipath components can be obtained based on the peak values in RU B V (τ ) .
Since the correlation has to be calculated across the whole range of the received signal in
an exhaustive way, this method requires heavy computation and may not be very feasible
in a low power system.
5.2.2 Received single pulse with a sinusoidal signal
In the following sections, a simpler pulse detection method is proposed by using a normal
sinusoidal signal as the template, which can be implemented easily compared to the
template signal match detection. Let us assume the center frequency of the received
single pulse s (t ) is f F and the frequency of the selected sinusoidal reference signal
ref (t ) is f U . f F and f U are quite close to each other, and the exact value of f U should be
chosen such as to maximize the output of the cross correlation between s (t ) and the
sinusoidal signal ref (t ) :
+∞
f = arg max( ∫ s (t )ref (t + τ )dt )
U
(5-3)
τ
IU
−∞
+∞
= arg max( ∫ s (t ) sin(2πf (t + τ ))dt )U
(5-4)
τ
IU
−∞
Within this section, the relations between s (t ) and ref (t ) are studied both
mathematically and graphically, namely the ideal in-phase situation, the normal
overlapping situation and the approximate in-phase situation. For convenience, s (t ) is
kept stable and ref (t ) can be shifted in time to study the above-mentioned relations.
Their cross correlation function is defined as:
+∞
cross _ corrV B UHI (τ ) = ∫ s(t )ref (t + τ )dt
−∞
(5-5)
35
5.2.2.1 Ideal in-phase and normal overlapping with a sinusoidal signal
The received single pulse s (t ) and the sinusoidal reference signal ref (t ) are called ideal
in-phase when the cross correlation function cross _ corrV B UHI (τ ) has the highest value,
which is shown in Figure 5-1.
We can define the ideal in-phase reference signal ref LGHDO B LQSKDVH (t ) as:
ref LGHDO B LQSKDVH (t ) = ref (t + τ LGHDO B LQSKDVH ) = sin(2πf U (t + τ LGHDO B LQSKDVH )) (5-6)
where τ LGHDO B LQSKDVH = arg max(cross _ corrV B UHI (τ )) . The correlation function of s (t ) and
τ
ref LGHDO B LQSKDVH (t ) can be given by:

W
corrL B LQSKDVH (t ) = ∫ s(t )ref
−∞
LGHDO B LQSKDVH (t )dt (5-7)
Figure 5-1: Ideal in-phase (0 phase shift) and its correlation result
As shown in Figure 5-2, the received single pulse s (t ) and the sinusoidal reference signal
ref (t + τ LGHDO B LQSKDVH + ∆τ ) are called the normal overlapping while a time shift ∆τ is
added to the ideal in-phase sinusoidal signal ref LGHDO B LQSKDVH (t ) to simulate a more common
relation between s (t ) and ref (t ) .
Figure 5-2: Normal overlapping and its correlation result
36
In both of Figure 5-1 and Figure 5-2, the cycled curve is s (t ) , the dotted curve is the
sinusoidal reference signal, the solid curve is the multiplication result and the stared
curve is the correlation result. The left figure indicates the relation between the received
pulse and the sinusoidal reference signal, and the right figure shows their correlation
result.
5.2.2.2 Approximate in-phase with a sinusoidal signal
In reality, the ideal in-phase situation is hard to achieve. Instead, the existence of a small
time shift between the received single pulse and a sinusoidal reference signal is the more
common case. Similar to the normal overlapped reference signal, the approximate in-
phased reference can be defined as:
ref D B LQSKDVH (t ) = ref (t − τ LGHDO B LQSKDVH − τ D ) (5-8)
where τ D is the small time shift between ideal in-phased reference signal and
approximate in-phased reference signal as shown in Figure 5-3.
The correlation function of s (t ) and ref D B LQSKDVH (t ) can be obtained as:

W
corrD B LQSKDVH (t ) = ∫ s(t )ref
−∞
D B LQSKDVH (t )dt (5-9)
Since the absolute value of τ D is very small compared to ∆τ and approaching to zero, the
maximum value of corrD B LQSKDVH (t ) is close to the maximum value of cross _ corrV B UHI (τ ) .
Figure 5-3: Approximate in-phase ( 0.2π phase shift) and its correlation result
5.2.3 In-phase situation detection
The difference between the situations of in-phase (approximate or ideal) and normal
overlapping can be seen easily in the correlation output. A threshold value might be set
after the correlation to detect the in-phase position. If the in-phase situation is detected,
there is a big chance that the in-phased received pulse is resolvable. Therefore, by
judging whether the received pulse is in-phased with the sinusoidal reference signal or
not, we can approximately evaluate the quality of this received pulse. The resolvable
37
multipath components in r (t ) might be detected by evaluating the correlation between

the sinusoidal reference signal and r (t ) :
W
corrU B UHI (t ) = ∫ r (t )ref (t )dt
−∞
W /
= ∫ [∑ a Q s (t − τ Q ) + n(t )]ref (t )dt
− ∞ Q =
W / W
= ∫ ∑ a Q s(t − τ Q )ref (t )dt + ∫ n(t )ref (t )dt
− ∞ Q = −∞
(5-10)
Because n(t ) is assumed to be AWGN with zero mean and we study in a high SNR
W
environment, the value of ∫ n(t )ref (t )dt

−∞
is negligible and it can be replaced with zero.
W /
corrU B UHI (t ) = ∫∑
Q
a Q s(t − τ Q )ref (t )dt
− ∞ =
/ W
= ∑ a Q ∫ s (t − τ Q )ref (t )dt
Q = −∞
/
= ∑ a Q corrQ (t ) (5-11)
Q =
We conclude that the correlation between r (t ) and the sinusoidal reference signal can be
regarded as the combination of the correlations between the separate multipath
components and the sinusoidal reference signal. The correlation for the ith multipath
component can be obtained by:
L L −
∑ a corr (t ) − ∑ a corr (t )
Q Q Q Q
corr (t ) = Q = Q
= corrU B UHI (τ L + T S ) − corrU B UHI (τ L ) (5-12)
=
a
L
where T S is the received single pulse’s width. If the ith multipath component is in-phased
with the sinusoidal reference signal, its correlation result could indicate such a fact. A
threshold θ can be added to corr (t ) for judging whether there is an in-phased pulse or
L
not. The value of θ is decided by the maximum correlation result of received single pulse
s (t ) and the sinusoidal reference ref D B LQSKDVH (t ) . It is advisable that in the real system, the
value of θ should be an adaptive setting according to the non-stable received pulse
amplitude a . L
38
Figure 5-4: The difference corr (t ) between corrU B UHI (τ L + T S ) and corrU B UHI (τ L )
L
5.2.4 Sinusoidal referenced pulse detection
According to the detection method that is introduced above, a pulse detection structure by
utilizing sinusoidal reference signals is described in this section.
Figure 5-5 Overview structure of the pulse position detection
The pulse detection structure is shown in Figure 5-5, and its corresponding detection
algorithm is given as:
Pulse is detected when

corrU B UHI (t ) − corrU B UHI (t − T S ) > θ
or
corrU B UHI (t ) − corrU B UHI (t − T S ) > θ
In this detection, two sinusoidal signals (In-phase and Quadrature with frequency f U ) are
used as reference signals that have a phase shift of π/2. The delay line is set to T S . The
threshold value θ is set in the “Hit Crossing” block. For high SNR, this structure can
produce acceptable results, which point out the timing position of the most resolvable
pulse(s). This is discussed in detail in section 6.2.1.
39
When we increase the number of reference sinusoidal signals from 2 to 3 or higher, the
efficiency can be improved. Because in a fixed period of time, the chance of approximate
in-phase or even ideal in-phase between the resolvable UWB pulse and reference
sinusoidal signals is higher when we increase the number of reference signals. Besides
increasing the number of reference signals to enhance the performance, more frames in
the UWB signal can be tracked to average the timing error or to find all the resolvable
pulses. However, the detection time will increase too.
5.2.5 Comparison between two detection methods
The difference between the performance of “template signal match detection” and the
performance of the proposed simpler pulse detection method can be studied based on the
cross correlation results in these two methods, which are shown in the lower figure in
Figure 5-6.
The cycled curve is the cross correlation cross _ corrV B UHI (τ ) , and the situations of
approximate in-phase and normal overlapping are indicated with τ D and ∆τ
respectively. The solid curve is the cross correlation RU B V (τ ) between the received single
pulse s (t ) and the template signal sWHPSODWH (t ) or the autocorrelation of the template signal
sWHPSODWH (t ) because s (t ) and sWHPSODWH (t ) have the same shape.
Since RU B V (τ ) has a narrower positive part compared to cross _ corrV B UHI (τ ) , we can
expect a more accurate detection result from RU B V (τ ) than cross _ corrV B UHI (τ ) . But since
the convergence of cross _ corrV B UHI (τ ) is simpler than RU B V (τ ) , we can expect a
detection method from cross _ corrV B UHI (τ ) with a simpler structure and a lower
computational demand. Or we can combine their advantages together: Using sinusoidal
referenced pulse detection for a coarse searching and using template signal match
detection for the next fine searching.
40
τD
∆τ
Figure 5-6: Cross correlation between UWB pulse and reference sinusoidal signal
41
Chapter 6 Evaluation by Simulation
6 Evaluation by Simulation
6.1 UWB system simulation setup
The goal of the system simulation is to derive specifications for receiver and transmitter
building blocks and study the system performance. This work is done based on existing
knowledge of the UWB system, related modulation technology and the properties of the
new timing generator made by Philips Research.
6.1.1 System simulation overview
The UWB system simulation is implemented in Matlab and designed in a flexible manner,
which enables quick modifications. The following diagram illustrates the general
structure of UWB system simulation.
Figure 6-1: UWB system simulation overview
6.1.1.1 UWB pulse generator
The UWB timing generator is one of the major components in the UWB communication
system, particularly in hardware implementation. It is obvious that a more accurate
timing generator can lead to a better overall performance. But in the simulation, the UWB
pulse generator that includes the timing generation can be easily and precisely built in
software. It supports generating different UWB pulses (Gaussian and Raised Cosine
Pulses) in different modulation schemes (PPM and Bipolar). More selections of pulse
shape and modulation schemes can be included in the later work.
6.1.1.2 Multipath channel simulation
This module is built to simulate the multipath channel environment of UWB

communication, where one input pulse can lead to multiple delayed pulses at the output.
Two vectors can be modified: One is the delay time, which represents different lengths of
transmission paths. Another is the normalized attenuation factor that is used to simulate
the reflections and energy loss of different paths.
For example in a 20 multipath simulation, both of the two vectors have 20 values that are
given as:
42
delay _ vector = [0 0.9 1.4 2.2 3.1 3.3 3.6 4 4.5 4.8 5.3 6.6 8 9.5 11 14 16 20 23 28]⋅ 10 −
attenuation = [0.8 - 0.3 0.2 0.1 0.2 0.3 0.2 1 0.2 - 0.2 - 0.43 - 0.36 0.7 - 0.25 0.6 0.5 0.3 - 0.2 0.4 - 0.3]
These two vectors represent a realization of UWB channel model with 20 multipath. One
may notice that five major paths contain most part of the energy among these 20 channels.
[ ][ ][ ] [ ]
They are [0,0.8], 4 ⋅ 10 − ,1 , 8 ⋅ 10 − ,0.7 , 11 ⋅ 10 − ,0.6 and 14 ⋅ 10 − ,0.5 . Therefore,
combining those five paths’ energy together could be a helpful solution to improve the
performance of UWB receiver.
6.1.1.3 AWGN simulation
During a simulation, a data sequence (UWB signal) is sent through the AWGN building
block, where the noise signal is added to the propagating UWB signal. This noise signal
can be created as pure additive white Gaussian noise, or more realistic band-limited noise
to simulate the noise after BPF (Band Pass Filter). In this simulation, particularly on the
Rake receiver performance measurements, band-limited noise is used as the referenced
noise signal, and it is generated by applying a BPF on a white noise as shown in Figure
6-2 (a).
Figure 6-2: Band limited noise generator (a) and its PSD (b)
6.1.1.4 UWB receiver and demodulation
The UWB receiver module mainly contains two important components, the UWB pulse
detector and the Rake receiver. The UWB pulse detector detects the time of arrival of
resolvable received pulses. When the channel is known, UWB Rake receiver can retrieve
43
the UWB baseband signal from the received signal. At the demodulation block, the UWB
baseband signal is sampled and fed into a threshold device (a comparator). The threshold
device produces the binary serial-data waveform m ~ (t ) , which can be compared with
input binary serial data m(t ) to evaluate the UWB system.
6.1.2 Variable sample rate simulation
In a UWB signal, the pulse only comes in a very short period, which is typically less than
several hundred Pico-seconds. The other part of the signal is silence, which contains only
data of zeros. Therefore, doing the system simulation or the spectrum analysis with
uniformly sampled UWB signal is not wise, due to the simulation time is mainly
consumed by dealing with zeros. On the other hand, if we decrease the sample rate to
accelerate the simulation, the resolution of UWB pulse will drop drastically.
To prevent this tradeoff problem in the fixed sample rate simulation, a new method by
means of variable sample rate simulation is introduced. Its main advantage is that an
enormous amount of unnecessary computation can be avoided. In this method, as shown
in Figure 6-3, we only sample within the time windows when UWB pulses occur and
jump to the next time window when the sampling of last time window is finished.
To the time window

Time window in the next frame
nT (n+1)TI t
I
nT
nT 6-3: Variable sample rate UWB simulation
Figure
6.1.3 Factors related to the performance of simulation
Several relevant factors can be identified which influence the simulation performance, in
particular, the simulation speed.
• Time window length. The time window introduced with the variable sampling rate
indicates the length of the time spread of the multipath channels that can be
observed by the simulation. Physically, the spread of the multipath reflections is
100 or 1000 times the UWB pulse width and it can even affect the next
transmitted data symbol and lead to ISI problems. For having an acceptable
simulation speed, we fix the observation window length to about 4 nano-seconds.
This parameter can be changed to do a more realistic simulation. But one should
be aware that the longer the window length, the longer the simulation time, and
they grow proportionally.
• Oversampling rate. Within the time window, the oversampling is implemented to
sample UWB pulses for increasing the resolution. The higher the oversampling
44
rate, the more sample points need to be calculated and the lower the simulation
speed is.
• Number of Rake fingers. As addressed in chapter 4, delay lines are key
components in a Rake receiver. In the variable sample rate simulation, the
function of a delay line is implemented by a data buffer. More delay lines requires
more buffers and more shuffle operations, and consequently, requires the longer
simulation time.
6.2 Evaluation of pulse detection by simulation

6.2.1 Pulse detection simulation
Several pulse detection simulations are done within this section based on the proposed
sinusoidal signal referenced pulse detection method. The results are shown by a number
of snapshots (Figure 6-4 – Figure 6-8). In each graph, the upper signal is the received
UWB signal r (t ) with noise, and the middle window shows the pulse detection result. To

clarify the detection performance, the pure received UWB signal ∑ a s(t − τ
Q =
Q Q ) without
noise is placed in the bottom as the reference.
Figure 6-4: Pulse detection test 1. Three resolvable pulses are detected
45
Figure 6-5: Pulse detection test 2. Two resolvable pulses are detected
Figure 6-6: Pulse detection test 3. Three resolvable pulses are detected
46
47
We do the pulse detection simulation within the continuous 20 frames in a 3dB SNR
environment. From the snapshots, we found some of the resolvable pulses are detected
twice, e.g. in Figure 6-4 and Figure 6-5. The reason is that the noise makes the correlation
difference crossing the threshold value twice for the same resolvable pulse. If this
happens, we can simply take the timing position of the first crossing as the detected pulse
position. The results are also described in Table 6-1, which includes the number of
detected resolvable pulses in each frame as well as the timing errors between the detected
resolvable pulse timing position and real resolvable pulse timing position. The mean
value and variance value of the timing errors is calculated based on the simulation results:
mean(err ) = 1.9205 × 10 s

var(err ) = 3.4365 × 10
The timing errors can influence the UWB receiver performance very much. We will see
part of the timing errors cannot be tolerated compared to the received pulse width, which
is about 1.7 × 10 − s . If we want to achieve the better performance, we could either track
on multiple frames to average the timing errors or combine the proposed detection
method with the conventional template signal match detection: Using sinusoidal
referenced pulse detection for a coarse searching and using template signal match
detection for the next fine searching.
Frame No. The number of detected The timing errors (×10 − s )
resolvable pulses
1 2 0, -7
2 3 2, 2, 1
3 2 -0.5, -2.8
4 1 0.5
5 2 12, 2
6 3 18, 2, -2
7 1 -0.2
8 2 18, 3
9 3 -0.5, -3, -6
10 3 1.5, 2, -1.5
11 2 4, -5
12 3 17, 12, 3
13 2 1, 2
14 1 -0.5
15 0
16 3 -1, -4, -2.5
17 4 3, -1, 13, 4
18 2 1.5, -4.5
19 2 1.5, -4.5
20 3 2, 1, 2
Table 6-1: The simulation result from 20 frames
48
The detection is for detecting the resolvable paths, and based on locking on these paths,
the receiver works to retrieve the transmitted data from the received UWB signal. If the
demodulated data contains too many errors, we can know the locking was lost. Therefore,
a new round of pulse detection will be done to relock on the resolvable paths.
6.2.2 Error analysis
Pulse detection errors can be classified into two categories: One is the failure alarm that
occurs when a false detection in the noise-only portion of the signal is regarded as a
resolvable path signal. The other is the missed resolvable-path error, which occurs when
the actual resolvable path signals are missed by the detector7.
6.2.2.1 Failure alarm
This kind of errors can be reduced by utilizing a training sequence. Because the failure
alarm due to noise occurs randomly and it does not contain the training sequence
information. For example, let us assume one burst of transmitted UWB signal contains
the training sequence (the header) and modulated pulses (the body).
UWB Header UWB body

Figure 6-9: An UWB burst

Tr _ header (t ) = ∑ gm(t + τ − (n − 1) ⋅ T ) I
(6-1)
Q =
The transmitted UWB header contains five (this number can be modified) continuous
periodic UWB transmitted pulses, namely gm1 , gm 2 , gm3 , gm 4 and gm5 . The
interval time between two continuous pulses is T . After being sent through a multipath
I
channel with 20 different paths, every transmitted pulse will result 20 received pulses
with different delay time. As shown in the pulse detection simulation, there are 3
resolvable paths (a, b and c) among those 20 paths. Therefore, every transmitted pulse
gmi corresponds to 3 resolvable received pulse, namely siD , siE and siF . The three five-

pulse sequences ∑ siD ,
L =
∑ siE and
L =
∑ si
L =
F should also be periodic with the period of T .
I
Therefore, the true resolvable received pulses in the training sequence should be
correlated with a periodic template pulse train. Otherwise, the detected pulse is a failure
alarm due to noise. The whole algorithm is shown in the following flow chart. When we
detect one resolvable pulse, we store its time of arrival; and then this algorithm is started
to judge whether it is a true UWB pulse or not.
7
In such a case, the actual resolvable pulse is not in-phased with any sinusoidal reference signals. Hence,
its existence is not known by the detector.
49
N = Number of training
pulses - 2 (e.g. 5-2 = 3)
Generate a template with

a delay time of T I
N
Correlated?
It is a failure alarm
Y
Generate a template with
a delay time of T I
N = N-1
N
N = 0?
Y
It is a real resolvable pulse
Figure 6-10: Flow chart of failure alarm judgment by utilizing training sequence
6.2.2.2 Missed resolvable path error
Some resolvable paths might be missed in the detection because they are not
(approximate or) in-phase with all reference sinusoidal signals. If too many of them are
missed, the collected power from the detected resolvable paths is not enough for the
UWB receiver to demodulate the transmitted data efficiently.
The limited number of sinusoidal references is the main reason of the missing of
resolvable paths. A large bank of parallel reference signals is expensive to generate, and
only a fixed number of references can be included in practice. Hence, longer pulse
50
detection time or longer training sequence needs to be paid to find out the resolvable
paths and to avoid the missing of the important ones.
6.3 Evaluation of receiver architecture by simulation

Rake receiver is proposed as the UWB receiver. Its performance is studied in this section
by the theoretical analysis and simulations.
6.3.1 Theoretical analysis
For binary signaling corrupted by white Gaussian noise, matched-filter reception, and by
using the optimum threshold setting, the bit error rate can be given by:
EG
PH = Q ( ) (6-2)
2N
+∞
1 1 z
∫e
− λ
where Q ( z ) = dλ = erfc ( ) , and E G is the difference signal energy at
2π ]
2 2
the receiver input (the output of the matched filter), and
7
E G = ∫ [s (t ) − s(t − 2 ⋅ T330 B VKLIW )] dt 8. N is the noise power. For the BPPM modulation,

E E = 0.5 E G , thus we can obtain the theoretical performance of the BPPM UWB receiver:
EG EE 1 EE
PH = Q ( ) = Q( ) = erfc( ) (6-3)
2N N 2 2N
The receiver performance shown in (6-3) is based on the SNR value at the output of the
Matched filter (the input of comparator). Because the received SNR value is measured at
the input of the Rake structure in this simulation, the processing gain between these two
points needs be estimated.
As introduced in Chapter 4, the processing gain contains two parts: the gain led by the
Rake structure and the gain led by the matched filter. Firstly in the simulation, the 5
selected resolvable pulses have normalized amplitudes of {0.8, 1, 0.7, 0.6, and 0.5}. If
they are absolutely not overlapping with any other received pulses, the amplitude of the
summation pulse will be 3.6. Therefore, the processing gain due to 5 Rake fingers on
SNR will be 10 log(3.6 / 5) = 4.1363dB maximally based on (4-2). Secondly in the
matched filter, the correlation duration is 0.165ns and the noise bandwidth is 11.4 GHz,
therefore the matched filter processing gain is 5.8dB based on (4-4). Overall, we have the
processing gain about 10dB for a 5 Rake receiver between the received SNR and the
output SNR of the Matched filter. Having this theoretical processing gain9 and the error
8
Please refer to the template signal for BPPM UWB system that is shown in the Chapter 4.
9
The processing gain resulted by the duty cycle factor is not taken into account. This can be referenced
with the power analysis in Chapter 3. If the duty cycle is considered here, e.g. duty cycle = 0.01, the related
processing gain is another 20dB.
51
function shown in (6-3), the theoretical performance of Rake receiver with the Matched
filter can be obtained (the black curves in Figure 6-11).
6.3.2 Effects of different Rake finger numbers
The delay line is one of the most important components in the Rake receiver. By utilizing
those delay lines, the signal power from multipath can be collected inside of the receiver.
The number of delay lines used in the receiver is an important factor regarding to its
performance. Therefore, a simulation is done to study the effects of different Rake finger
numbers. The detailed simulation method is shown in Appendix with an example.
BER vs. SNR of Rake receiver
-0.5
-1
-1.5
-2
BER (10 power)
-2.5
-3
5 Rake
-3.5 A
3 Rake
2 Rake
1 Rake
-4
Theoretical BER (1 Rake)
Theoretical BER (5 Rake)
-4.5
-10 -8 -6 -4 -2 0 2 4 6 8
SNR (dB)
Figure 6-11: BER vs. SNR simulation of different number of Rake fingers
It is not difficult to see that increasing the number of Rake fingers can improve the
performance of Rake receiver as shown in Figure 6-11. Because more Rake fingers
means more paths of the UWB signal can be tracked, and more energy can be collected
from those paths. But while increasing the number of Rake fingers, the quality (energy
and pulse shape) of new including resolvable path is not comparable with the first several
resolvable paths. Therefore the performance enhancement will become less noticeable for
large number of Rake fingers.
It is also noticed that there is a small difference between the simulation result and the
theoretical result. The reason is for the theoretical result, the processing gain value of the
Matched filter and the Rake structure is taken as the maximum. But in fact in the
52
simulation, they cannot work ideally. Therefore, the actual processing gain is smaller
than the theoretical result.
6.3.3 Effects of timing errors
The function of the correlator is to correlate the input UWB signal r (t ) with the template
pulse sWHPSODWH (t ) . Ideally, sWHPSODWH (t ) should be overlapped with the resolvable pulse very
well. But in real physical circuits, timing mismatch cannot be avoided. Hence, the
correlator timing error always exists and it may reduce the receiver performance seriously.
BER vs. SNR of the Rake receiver with timing error
0
-0.5
-1
-1.5
BER (10 power)
-2
-2.5
-3
-3.5
1 Rake without timing error

-4 Timing error = 1e-11 with 1 Rake
Timing error = 1e-11 with 5 Rake
-4.5
-8 -6 -4 -2 0 2 4 6 8
SNR (dB)
Figure 6-12: SNR vs. BER simulation with timing errors
Some simulation results regarding timing errors are shown in Figure 6-12. It is noticed
that the UWB system can have a reasonable performance while the timing error is smaller
than 10ps. One can see that a 5 finger Rake receiver with the timing error of 10ps has a
comparable performance with the single Rake receiver without timing error. But once
timing error exceeds 20ps, the overall performance is damaged drastically. The reason of
the performance reduction is because that timing error has a large influence on the
correlation result. This effect can be explained based on the autocorrelation function of
the template pulse sWHPSODWH (t ) .
53
Figure 6-13: Autocorrelation function of the template pulse sWHPSODWH (t )
As shown in Figure 6-13, the effects of timing error is studied based on the
autocorrelation function of the template pulse sWHPSODWH (t ) . Point P stands for the desired
overlapping between the received resolvable pulse and template pulse, which means there
is no timing error and highest correlation result can be achieved. It is clear that if the
timing error increases, the correlation result will be smaller and the performance is
getting lower. When the timing error exceeds T/2 (approximately 25ps in the simulation),
the result of correlation will become negative. Such a negative result indicates the signal
energy in the resolvable pulse cannot be transferred to the output of the correlator.
Therefore, the correct transmitted data can not be demodulated from this output signal.
We suggest using 10 Pico-seconds as the maximum tolerated value of timing error for the
UWB system with the pulse width of about 0.165ns.
54
55
Chapter 7 Interference From 5 GHz WLAN System
7 Interference From 5 GHz WLAN System

Initially, the main concern about UWB was whether or not they would interfere with
existing RF systems that provide essential military, aviation, fire, police, and rescue
services. For such a reason, the FCC spent about two years evaluating the proposed UWB
specifications and concluded that there should be no major interference from the UWB
systems. This conclusion is made mainly because of the extremely low emission power
limitation on the UWB system.
But on the other hand, low powered UWB equipments themselves are facing significant
interference problem from other wireless systems. Among them, 802.11a WLAN system
is the main concern. Because it has a high emission power and its operating frequency
band (5.150 – 5.825 GHz) is inside of the FCC approved operating band for UWB
systems. Therefore, a performance simulation is done in this section to study the problem
of interference from 802.11a WLAN to UWB system.
7.1 Review of 802.11a WLAN signal

The IEEE 802.11a standard specifies an OFDM physical layer (PHY) that splits an
information signal across separate subcarriers. According to this, a 20MHz 802.11a
signal is built by 64QAM for simulating the interference from WLAN system to UWB
(Figure 7-1). The ratio between the peak voltage and its RMS voltage is observed as
3.2227, which is important for generating the 5GHz WLAN signal with a specific power
level.
Figure 7-1: One 802.11a WLAN signal and its spectrum (5.3 GHz, 64QAM)
7.2 Analysis of UWB system performance with 802.11a

interference
In Figure 7-2, the simulation result of the performance of the UWB receiver with WLAN
interference is shown. Compared to the simulation with bandlimited noise (Figure 4-5),
For the UWB system, its ability against the WLAN interference is actually better than the
ability against thermal noise. This is because that WLAN signal is narrow-band
56
Chapter 7 Interference From 5 GHz WLAN System
interference signal to UWB system, and it only distorts a small frequency portion of
UWB signal. But because normally the WLAN interference signal is much stronger than
thermal noise signal from the energy point of view, this interference problem is more
serious than thermal noise effects for the UWB system.
BER vs. SIR of Rake receiver with 5.3GHz 802.11a interference
-0.5
1 Rake
5 Rake
-1
-1.5
BER (10 power)
-2
-2.5
-3
-3.5
-4
-8 -6 -4 -2 0 2 4 6
SIR (dB)
Figure 7-2: SNR vs. BER simulation with WLAN interference (noise free)
7.3 Solution to the interference problem

As shown in the simulation result and stated above, WLAN interference signal is the
killer to UWB system, and the solution needs to be found out to allow UWB system
operating with a nearby 5GHz WLAN interference source. One possible solution is
applying a bandstop filter with a stop band from 5.1 – 5.9 GHz on the received UWB
signal. Thus, if there is a 5 GHz WLAN interference signal in the received UWB signal,
it can be suppressed by the bandstop filter. But there are two major shortcomings for this
solution: The first one is that a good implementation of this bandstop filter is hard due to
high frequency requirements; The second one is that this bandstop filter can also distort
the UWB signal and lower down the overall performance.
Multi-band solution is another method to solve this interference problem. In this proposal,
7.5 GHz of unlicensed UWB spectrum is divided into several non-overlapping frequency
bands, and each band is between 500 and 700 MHz wide. Data can be transmitted
through all the sub-bands or several of them. In the case of coexistence with IEEE
802.11a, one or two corresponding (the interfered) UWB sub-bands can be removed from
the transmission to guarantee there is no interference of 802.11a system in UWB signal.
57
Chapter 8 Conclusions and Recommendations
8 Conclusions and Recommendations

8.1 Conclusions
In this paper, the Gaussian Pulse and Raised Cosine Pulse were studied for generating the
UWB signal. The spectral properties of the pulses were compared to FCC & ETSI UWB
spectral masks. Due to its easier implementation, the Gaussian Monocycle is chosen for
the simulation. The pulse should be slightly modified (by applying mask filter) to satisfy
the FCC specification. BPPM with pseudo random time hopping is selected as the UWB
modulation scheme within this project because of its good spectral properties and its
multipath performance.
The UWB receiver structure with multiple Rake fingers was simulated. Its performance
was evaluated and compared. We conclude that UWB system can operate perfectly in a
low SNR environment due to the properties of UWB signal and matched filter reception
combined with Rake structure. But as a tradeoff, we find that the UWB system requires
higher timing accuracy to provide a high-performance wireless link.
In UWB receiver design, pulse detection is regarded as an important part to ensure a good
overall performance. Therefore, this topic is highlighted and a new pulse detection
method is introduced and simulated. Compared to other detection methods [13], the
proposed one is easier to be realized and might lead to a faster synchronization
mechanism.
Another important result of this thesis is obtained by the evaluation of 802.11a WLAN
interference on the UWB system. It is concluded that single-band UWB system is not
robust enough against this WLAN interference. This is because of the large difference
between UWB signal power and 802.11a WLAN signal power and their overlapping
frequency bands. Therefore, additional concerns have to be taken to solve this critical
problem. This one seems to be the most serious problem that is being faced by the single-
band UWB system, and it is also the reason that many UWB companies shifted from the
single-band solution to the multi-band solution, recently. Unfortunately, due to the
critical time and lack of references, more detailed discussion on the multi-band UWB
solution is not carried out within this thesis project.
8.2 Recommendations and further research topics

When we were studying the transmitted and received UWB pulse shape, the shape
distortion problem was not taken into account. At the moment, an ideal pulse shape (2nd
Gaussian derivative pulse) is used as the template pulse and implemented in the system
simulation. After further sufficient measurements on channel and transmitting and
receiving antennas, the more realistic received pulse shape can be obtained.
The single-band UWB receiver structure is relatively simpler than many other radio
systems. This is one of the main advantages of UWB system. But for gaining this benefit,
a higher timing accuracy is required. Therefore, having a good pulse detection or
58
Chapter 8 Conclusions and Recommendations
synchronization method is essential for the whole implementation. Although one possible
method is introduced in this thesis, it is still a long way to go to achieve a good solution.
For instance, the results of channel modeling can be used to optimize the synchronization
scheme. Preferably, the aiming solution is in the analog domain. Because the pure digital
solutions require the high speed and high powered ADC, which is very costly.
In UWB receivers, signal power from multipath components are collected by the Rake
structure, so that the total signal energy is increased to assure a better performance. It is
obvious that the number of Rake fingers is very related to the performance: The more
Rake fingers, the more signal power can be gathered and the higher the signal to noise
ratio can be obtained. Therefore, further research is necessary on selecting the suitable
number of Rake fingers, which could be done based on more channel measurements and
cost consideration.
UWB circuits, components like filters and mixers need to operate at much higher
bandwidth compared to the other commercial wireless communication systems.
Therefore, the design requirements of large working bandwidth, good linearity and
acceptable operating power level are necessary. Besides these electronic components,
UWB antenna design and implementation is another challenging problem that should be
solved by further research. Because those conventional antennas are designed to radiate
only over a relatively narrow range of frequencies, and if an impulse (Wide band
frequency signal) is fed to such an antenna, it tends to ring, severely distorting the pulse
and spreading it out in time.
Although the Multi-band solution is being more and more emphasized and lots of UWB
companies support the Multi-band-based 802.15.3 proposal now, the single-band solution
that is introduced and discussed in this thesis still has its own advantage and importance.
Moreover the idea of dual-band solution (Proposed by Xtremespectrum Inc.), one of the
multi-band solutions, is very related to the single-band system. Therefore, research
achievements and development results in the single-band system might be shifted and
utilized in the dual-band system.
59
Appendixes
9 Appendixes
9.1 Appendix A
Since we have:
*0 I = − M ⋅ π ⋅ π ⋅ I ⋅ τ ⋅ H −π τ $

I
:H FDQJHWWKHHQHUJ\VSHFWUXPRIWKLVVLQJOH*DXVVLDQSXOVH
*0 I = π ⋅ I ⋅ τ ⋅ H −π τ $

I
7KH FHQWHUIUHTXHQF\ I F VKRXOGVDWLVI\WKHHTXDWLRQ

G *0 I F GI = ⋅ π ⋅ I F ⋅ τ ⋅ H − π − ⋅ π ⋅ I F ⋅ τ H − π =

I Fτ
I Fτ
⋅ I F ⋅τ = ⋅ I F ⋅τ ⋅ π

I Fτ π = $
7KHUHIRUHZHSURYHWKDW I F LV WKHUHFLSURFDORI τ
7KH QH[W VWHS LV WR ILQG WKH G% EDQGZLGWK RI WKH VLQJOH SXOVH¶V HQHUJ\ VSHFWUXP 7KH
G% IUHTXHQF\ I G% VKRXOGFRPSO\WR

*0 I G% = *0 I F $

I G% ⋅ τ ⋅ π ⋅ H −π I G%τ = ⋅ I F ⋅ τ ⋅ π ⋅ H − π I τ

F
OQ I G% − π I G%τ = OQ I F − π I Fτ

/HWXVDVVXPH I G% = PI F WKHQ

OQ + OQ P = P − π I Fτ
%HFDXVHDVSURYHGDERYHWKDW I Fτ π =
OQ + OQ P = P − $
%DVHGRQ)LJXUHZHREWDLQWKHYDOXHRI P DVRUWKXVWKHG%EDQGZLGWK
FDQ EHFDOFXODWHGDV

%:G% = − I F = I F $
60
$SSHQGL[HV
Figure 9-1: The calculation of equation A-5

9.2 Appendix B

7KH PD[LPXPWUDQVPLWWHG8:%VLJQDOSRZHUDFFRUGLQJWR)&&VSHFLILFDWLRQLV
37[ = −G%P 0+] ⋅ *+] = PLOOLZDWWV = −G%: %

/HW¶VDVVXPHWKHGLVWDQFHEHWZHHQWKHUHFHLYLQJDQWHQQDDQGWUDQVPLWWLQJDQWHQQDLV
PHWHUVDFFRUGLQJWRWKH±/26PRGHOWKDWLVSURSRVHGE\3KLOLSV5HVHDUFK5HGKLOO
35[ = H − G ⋅ 37[

%

:H FDQKDYHWKHUHFHLYHG8:%VLJQDOSRZHU
35[ B G% = ORJH − − = − − = − G%: %

%DVHGRQWKHDVVXPSWLRQWKDWFLUFXLW¶VQRLVHLVIURPWKHUPDOVRXUFHVWKHQRLVHSRZHULV
HVWLPDWHGDVIROORZLQJ
31RLVH B G% = ORJ 31RLVH + )/1$ B G%
= ORJ N ⋅ 7 ⋅ % + )/1$ B G%
= ORJ N ⋅ ⋅ ⋅ +
= G%: %
ZKHUH N LV%ROW]PDQQ¶VFRQVWDQW N = × - . 7 LVWKHDEVROXWHWHPSHUDWXUHRI
−
FLUFXLWV IRU H[DPSOH .HOYLQ % LV EDQGZLGWK RI UHFHLYHG 8:% SXOVH10 )G% B /1$ LV
WKH QRLVHILJXUHRI/1$DQGG%LVXVHGDVDQDSSUR[LPDWHYDOXH

7KXV 615FDQEHFDOFXODWHGEDVHGRQWKHUHFHLYHG8:%VLJQDOSRZHUDQGWKHUPDOQRLVH
SRZHU
615 = 35[ B G% − 31RLVH B G% = − − − = G% %
10
7KH YDOXHRI*+]LVREVHUYHGE\ WKHVLPXODWLRQ7KHG%EDQGZLGWKRIWKHUHFHLYHG8:%SXOVH
VHFRQG *DXVVLDQ GHULYDWLYH SXOVH LV IURP *+]²*+] FRUUHVSRQGLQJ WR WKH EDQGZLGWK RI WKH
WUDQVPLWWHG8:%SXOVHILUVW*DXVVLDQGHULYDWLYHSXOVHIURP*+]²*+]

$SSHQGL[HV
%HFDXVH8:%SXOVHVRQO\RFFXURFFDVLRQDOO\LQWLPHDQGLIWKHSRZHURI8:%VLJQDOLV
IL[HG WKHSHDNYDOXHRI 8:%SXOVHFDQEHVWLOO TXLWHKLJKLQWKHVLWXDWLRQRIORZGXW\
F\FOH IDFWRU11 7R DYRLG WKH LQWHUIHUHQFH WR RWKHU V\VWHPV )&& DOVR SURSRVHG WKH SHDN
OLPLW RI 8:% SXOVH 7KH DEVROXWH SHDN HPLVVLRQ OHYHO LV QRW SHUPLWWHG WR H[FHHG WKH
DYHUDJHOLPLWE\PRUHWKDQG%²G%12 7KHUHIRUHDFWXDOO\G%SURFHVVLQJJDLQFDQ
EH REWDLQHG LQ 8:% VLJQDO SRZHU E\ IROORZLQJ WKH VWULFWHVW OLPLW :H GHILQH
615LQVWDQWDQHRXV DV WKHVLJQDOWRQRLVHUDWLRLQWKHWLPHVORWZKHQ8:%SXOVHRFFXUV
615 = 615 + = G% %

LQVWDQWDQH RXV
9.3 Appendix C

$V ZH NQRZQ IRU D VLQXVRLGDO VLJQDO 9UPV = 9 SHDN ZKLFK LV DQ LPSRUWDQW SURSHUW\
WKDW QHHGV WR EH FRQFHUQHG ZKHQ HVWLPDWLQJ WKH VLQXVRLGDO VLJQDO SRZHU 6LPLODUO\ WKH
UDWLREHWZHHQ 9UPV DQG 9 SHDN IRUDUHFHLYHG*DXVVLDQSXOVHndGHULYDWLYH*DXVVLDQSXOVH
KDV WREHIRXQG7KHQEDVHGRQWKLVUDWLRYDOXHZHFDQJHQHUDWHWKHULJKW*DXVVLDQSXOVH
ZLWKDFHUWDLQSRZHUOHYHOZKLFKLVYHU\LPSRUWDQWIRUWKHODWHUSHUIRUPDQFHDQDO\VLV

$FFRUGLQJWRWKHHTXDODYHUDJHHQHUJ\ZHKDYH
9UPV 9 SHDN
=< 9 SHDN V W ⋅ , SHDN V W >=< 9 SHDN V W ⋅ V W > &

5 5
7KHQ
9 SHDN
9UPV = 9 SHDN < V W >= V W GW = 9 SHDN ⋅ &

7
7 SXOVH ∫
SXOVH

ZKHUH ZDVPHDVXUHGE\WKHVLPXODWLRQ)LQDOO\ 9 SHDN FDQEHFDOFXODWHGDV

9 SHDN = ⋅ 9UPV = ⋅ 9UPV &

7KHUHIRUH ZH FDQ FRQFOXGH WKDW WKH SHDN YDOXH RI WKH 8:% UHFHLYHG SXOVH VHFRQG
*DXVVLDQGHULYDWLYHSXOVHLVDURXQGWLPHVLWV506YROWDJHYDOXH&RPSDUHGWRWKH
UDWLRYDOXHRI IRUDVLQXVRLGDOVLJQDOLWLVREYLRXVWKDWWKHHQHUJ\RI8:%SXOVHLV
PRUH FRQFHQWUDWHGLQWKHWLPHGRPDLQ

9.4 Appendix D

/HW XV WDNH WKH VLPXODWLRQ RQ SRLQW $ )LJXUH DV DQ H[DPSOH 7KH nd GHULYDWLYH
*DXVVLDQSXOVHZLWKSHDNYROWDJHYDOXHRIPLOOLYROWVLVFUHDWHGDVWKHVLPXODWHG
UHFHLYHG8:%SXOVH7KXVWKHVLJQDOSRZHUFDQEHREWDLQHGDV
11
'XW\F\FOHIDFWRULVWKHUDWLREHWZHHQ8:%SXOVHZLGWKDQGDYHUDJHSXOVHUHSHWLWLRQSHULRG
12
7KHYDOXHRIG%LVGHILQHGLQWKH)&&UHSRUW&)5EG%LVWKHFRQVLGHULQJYDOXHIRU
IXUWKHUVWLSXODWLRQLQ)&&

$SSHQGL[HV
9UPV 9 SHDN T
3V = '

5 5
=
ZKHUH T LVWKHUDWLRIDFWRUEHWZHHQWKHSXOVH¶VSHDNYROWDJHDQGLWV506YROWDJHZLWKLQ
WKH SXOVHZLGWKLWVYDOXHLVREWDLQHGDVEDVHGRQWKHFDOFXODWLRQLQ$SSHQGL[&
7KHUHIRUHLIZHWDNH 5 DVRKPVZHREWDLQWKHVLJQDOSRZHUDV
⋅ −
3V = = ⋅ − :DWWV = G%: '

$V VKRZQLQ$SSHQGL[%WKHLQEDQGWKHUPDOQRLVHSRZHUFDQEHFDOFXODWHGDV
3Q = ⋅ ORJ N ⋅ ⋅ ⋅ + = G%: '
7KHUHIRUHZHFDQREWDLQWKHUHFHLYHG615LQWKLVH[DPSOHVLPXODWLRQDWSRLQW$
615 = 3V − 3Q = G% '
7KH %(5 DFKLHYHG E\ WKH VLPXODWLRQ LV × − 7KHQ EDVHG RQ WKH %(5 DQG WKH
FDOFXODWHG 615 WKH SHUIRUPDQFH UHVXOW LV VKRZQ LQ )LJXUH DV SRLQW $ 7KH RWKHU
VLPXODWLRQUHVXOWVDUHREWDLQHGZLWKWKHVDPHDSSURDFK

5HIHUHQFHV
References

>@ %RE +DIQHU DQG ,DQ .HHQH *DUWQHU ,QF ³7KH 1HWZRUNLQJ +\SH &\FOH´
KWWSZZZJDUWQHUFRPBHYHQWVDXGLRFRQIHUHQFHVDWWDFKPHQWV70SSW th
0D\
>@ 3HQ&/L3KLOLSV6HPLFRQGXFWRUV³&KDOOHQJHVIRU$96WUHDPLQJ2YHU
:LUHOHVV /$1´ KWWSZZZH
LQVLWHQHWHFQPDJLQGH[DVS"OD\RXW DUWLFOH DUWLFOHLG &$ WH[W SKLOLSV
th0D\
>@ 5HYLVLRQ RI 3DUW RI WKH &RPPLVVLRQ¶V 5XOHV UHJDUGLQJ 8OWUD:LGHEDQG
7UDQVPLVVLRQ V\VWHPV KWWSZZZIFFJRY%XUHDXV(QJLQHHULQJB7HFKQRORJ\
2UGHUVIFFSGI
>@ 7HUHQFH : %DUUHWW ³+LVWRU\ RI 8OWUD :LGH %DQG 8:% 5DGDU
&RPPXQLFDWLRQV 3LRQHHUV DQG ,QQRYDWRUV´
KWWSZZZQWLDGRFJRYRVPKRPHXZEWHVWSODQEDUUHWBKLVWRU\BSLHUVZILJVSGI
>@ /HRQ : &RXFK ,, ³'LJLWDO DQG $QDORJ &RPPXQLFDWLRQ 6\VWHPV´ WK (GLWLRQ
3UHQWLFHKDOO,QWHUQDWLRQDO,QF,6%1
>@ 0 *KDYDPL / % 0LFKDHO 5.RKQR ³+HUPLWH )XQFWLRQ %DVHG 2UWKRJRQDO
3XOVHVIRU8OWUD:LGHEDQG&RPPXQLFDWLRQ´,(((VLJQDOSURFHVVLQJOHWWHUV
>@ ,((( VWDQGDUG DQG LWV VWXG\ JURXS
KWWSJURXSHULHHHRUJJURXSVSXE6*DKWPO
>@ 5LFN 5REHUWV .DL 6LZLDN -DVRQ (OOLV ³,((( 36*D )LYH &ULWHULD´
KWWSLHHHRUJSXE6HSU3B6*DB&ULWHULDGRF
>@ ;WUHPHVSHFWUXP :KLWH 3DSHU ³8OWUD:LGHEDQG :LUHOHVV :LWKRXW &RPSURPLVH
± $ 5HQDLVVDQFH ,Q 7KH 0DNLQJ´ 0D\
KWWSZZZ[WUHPHVSHFWUXPFRP3')8:%:KLWH3DSHUSGI
>@ 0DWWKHZ / :HOERUQ ³6\VWHP &RQVLGHUDWLRQ IRU 8OWUD:LGHEDQG :LUHOHVV
1HWZRUNV´;WUHPH6SHFWUXP,QF
>@ 0RQR3KDVH DQG %L3KDVH 8OWUD:LGHEDQG :+,7( 3$3(5 ;WUHPH6SHFWUXP
,QF KWWSZZZ[WUHPHVSHFWUXPFRP3')%LSKDVHBYVB0RQRSKDVHSGI $SULO

>@ 7LPH'RPDLQ&RUSRUDWLRQ³3XOV217HFKQRORJ\2YHUYLHZ´KWWSZZZWLPH
GRPDLQFRP)LOHVGRZQORDGVWHFKSDSHUV3XOV212YHUYLHZBSGI$SULO
>@ *HQHUDO$WRPLFV,QF*HQHUDO$WRPLFV3+<SURSRVDOWRD0D\
KWWSZHEJDWFRPSKRWRQLFVXZELPDJHVU3B7*D*HQHUDO
$WRPLFV&)33UHVHQWDWLRQ3')
>@ 0RH=:LQ5REHUW$6FKROW]³8OWUD:LGH%DQGZLGWK7LPH+RSSLQJ6SUHDG
6SHFWUXP ,PSXOVH 5DGLR IRU :LUHOHVV 0XOWLSOH$FFHVV &RPPXQLFDWLRQV´ ,(((
75$16$&7,21621&20081,&$7,21692/12$35,/
>@ -</HH5$6FKROW]³5DQJLQJLQD'HQVH0XOWLSDWK(QYLURQPHQW8VLQJDQ
8:% 5DGLR /LQN´ ,((( MRXUQDO RQ VHOHFWHG DUHDV LQ FRPPXQLFDWLRQV YRO
1R 'HFHPEHU
>@ 0=:LQDQG5$6FKROW]³(QHUJ\FDSWXUHYHUVXVFRUUHODWRUUHVRXUFHVLQXOWUD
ZLGH EDQGZLGWK LQGRRU ZLUHOHVV FRPPXQLFDWLRQV FKDQQHOV´ LQ 3URF 0LOFRP
YRO1RYSS

5HIHUHQFHV
>@ '&DVVLROL0RH=:LQ DQG$QGUHDV)0ROLVFK³7KH8OWUD:LGH%DQGZLGWK

,QGRRU&KDQQHO)URP6WDWLVWLFDO0RGHOWR6LPXODWLRQV´,(((MRXUQDORQVHOHFWHG
DUHDVLQFRPPXQLFDWLRQVYRO1R$XJXVWSS±
>@ 0RH = :LQ *HRUJH &KULVLNRV 1HOVRQ 5 6ROOHQEHUJHU ³3HUIRUPDQFH RI 5DNH
5HFHSWLRQ LQ 'HQVH 0XOWLSDWK &KDQQHOV ,PSOLFDWLRQV RI 6SUHDGLQJ %DQGZLGWK
DQG 6HOHFWLRQ'LYHUVLW\2UGHU´
>@ 0RH = :LQ DQG 5 $ 6FKROW] ³&KDUDFWHUL]DWLRQ RI 8OWUD:LGH %DQGZLGWK
:LUHOHVV ,QGRRU &KDQQHO $ &RPPXQLFDWLRQ7KHRUHWLF 9LHZ´ ,((( -RXUQDO RQ
VHOHFWHGDUHDVLQFRPPXQLFDWLRQVYRO1R'HFHPEHUSS
>@ :LOOLDP *DUGHU ³6WDWLVWLFDO 6SHFWUDO $QDO\VLV D QRQSUREDELOLVWLF WKHRU\´
3UHQWLFH+DOOSS
>@ $QQH&DVZHOO.DULQYDQ'DP³6XUYH\RI,QWHUIHUHQFH&DQFHOODWLRQ7HFKQLTXHV
IRU&'0$7HUPLQDOV´3KLOLSV5HVHDUFK5HGKLOO
>@ * ' )RUQH\ -5 ³0D[LPXP/LNHOLKRRG 6HTXHQFH (VWLPDWLRQ RI 'LJLWDO
6HTXHQFHV LQ WKH 3UHVHQFH RI ,QWHUV\PERO ,QWHUIHUHQFH´ ,((( WUDQVDFWLRQV RQ
LQIRUPDWLRQWKHRU\YRO,71R0D\SS
>@ 4LQJKXD /L /HVOLH $ 5XVFK ,QWHU ,QF ³+\EULG 5DNH0XOWLXVHU 5HFHLYHUV IRU
8:%´ KWWSZZZLQWHOFRPWHFKQRORJ\XOWUDZLGHEDQGGRZQORDGV
+\EULG5$.(0XOWLXVHU5HFHLYHUVIRU8:%SGI

Thesis 2

Uploaded by

Copyright:

Available Formats

Thesis 2

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesis 2

Uploaded by

Copyright:

Available Formats

Modeling and Performance Evaluation of

a BPPM UWB System

Delft University of Technology

Philips Research Laboratories Eindhoven

4.4.2 Analog and digital circuit design .............................................................. 32

ref LGHDO B LQSKDVH (t ) – The ideal in-phase reference signal

I am grateful to people who have contributed towards shaping this thesis.

1.1.1 3G and WLAN systems

In a 3G system, e.g. UMTS or CDMA-2000, user data rates can be provided at up to 2

Figure 1-1: WLAN data throughput vs. range based on measurements[2]

1.1.2 For supporting higher transmission speed – UWB

Display standards VGA SVGA XVGA SXVGA

1.2 Problem Statement

• Synchronization method. Synchronization is one of the crucial tasks of a UWB

1.3 Organization of this thesis

2.2 Features of UWB system

Compared to other narrow band communication systems, which operate in the

Conventional radio signal,

2.2.2 Robust to multipath interference

2.3 Related work and recent progress

2.3.2 FCC and ETSI regulation on UWB products

Figure 2-2: FCC and ETSI UWB specification

2.3.3 IEEE 802.15.3 standard

3 UWB Signal Analysis

3.1 Single pulse representation

3.1.1 Gaussian monocycle

The center frequency f F should satisfy:

Figure 3-1: Gaussian monocycle pulse and its spectrum

3.1.2 Raised Cosine pulse

Its corresponding time domain waveform is calculated as:

rc (t ) and its energy spectrum RC ( f ) are shown in Figure 3-2.

Figure 3-2: Raised Cosine pulse and its energy spectrum

3.1.3 Pulse shape selection

3.2 Pulse train and pseudo random pulse location

p SHULRGLF (t ) = ∑ gm(t − ( j − 1)T ) I

Figure 3-4: Periodic pulse train and its spectrum

Figure 3-5: Random positioned pulse train and its spectrum

The dataset of { ∆T } can be designed as a Pseudorandom Noise (PN) codes. In a

3.3 Modulation techniques

Figure 3-6: PAM modulation [9]

Figure 3-7: OOK modulation

Figure 3-8: PPM modulation

3.3.4 Antipodal Modulation (BPSK)

Figure 3-9: BPSK modulation

3.3.5 Other modulation schemes

Bipolar Modulation is similar to OOK, but a binary “1” is represented by alternately

3.4 Selected modulation scheme

Binary-PPM modulation uses long sequences of monocycles for communications, and

T330 B VKLIW T330 B VKLIW

Figure 3-10: Random positioned pulse train with BPPM modulation

3.5 Power analysis

P7[ = −41.25dBm / MHz ⋅ 7500 MHz = −2.5dBm = 0.56 milliwatts (3-13)

For non-LOS: P5[ = e − G ⋅ P7[

where d is the distance between Tx and Rx in meters.

UWB signal bandwidth 7.5 GHz (3.1 GHz – 10.6 GHz)

3.6 Transmission environment analysis and its effects on UWB

too small value when the transmitting pulse train is built.

3.6.2 Effects caused by related moving between Tx and Rx

A normal indoor office can be simulated as a high-density multipath environment, and

var(err ) = 3.4365 × 10

7KH FHQWHUIUHTXHQF\ I F VKRXOGVDWLVI\WKHHTXDWLRQ

OQ I G% − π I G%τ = OQ I F − π I Fτ

/HWXVDVVXPH I G% = PI F WKHQ

ZKHUH ZDVPHDVXUHGE\WKHVLPXODWLRQ)LQDOO\ 9 SHDN FDQEHFDOFXODWHGDV

>@ '&DVVLROL0RH=:LQ DQG$QGUHDV)0ROLVFK³7KH8OWUD:LGH%DQGZLGWK