Upload
truongtram
View
219
Download
2
Embed Size (px)
Citation preview
Transmission of OFDM-UWB radio signals in FTTH networksusing chirp-managed lasers
Pedro Luís Benito Canêlhas
Dissertação para obtenção do Grau de Mestre em
Engenharia Electrotécnica e de Computadores
Júri
Presidente: Prof. José Manuel Bioucas DiasOrientador: Prof. Adolfo da Visitação Tregeira CartaxoVogal: Prof. António Luís Jesus Teixeira
Outubro 2010
Acknowledgments
I would like to express my gratitude to my supervisor, Professor Adolfo Cartaxo, whose expertise, de-
manding requirements and patience were relevant to my dissertation. I appreciate his vast technical
knowledge in the field of optical communications, and his assistance in the evaluation of my reports.
I would like also to thank the PhD students at the Group of Research on Optical Fiber Telecommunication
Systems (GROFTS) of Instituto de Telecomunicações, Tiago Alves, Nelson Costa and Filipe Wiener, for
their prompt availabitlity and continued support during the many weeks we worked side by side in the
optical laboratory at Instituto Superior Técnico.
An acknowledgement is addressed to Instituto de Telecomunicações in recognition for the scolarship and
the use of their facilities.
Lastly, and most importantly, I am extremely grateful to my family and friends for the encouragement
and support they provided me through my entire life.
I dedicate this dissertation to my mother and father for their consistent guiding and strong commitment
providing me with all the educational tools to achieve my academic goals, and to my brother, whose
example and advice were also determinant in the choices I had to make in my life.
i
Abstract
Fiber-to-the-home (FTTH) networks have been proposed as a means to extend the application of ultra-
wideband (UWB) radio signals in future wireless personal area networks (WPAN) to the delivery of
high data-rate applications. In this dissertation, it is investigated, using numerical simulation through
MATLAB R©, the transmission of orthogonal frequency-division multiplexing ultra-wideband (OFDM-
UWB) radio signals in FTTH networks using chirp-managed lasers (CML). The main parameters that
influence the performance degradation of the OFDM-UWB signals distribution in FTTH networks are
identified and optimized. The main transmission impairments are discussed and the maximum transmis-
sion distance for a required minimum quality is assessed.
The performance optimization shows that the optical transmitter originates a half-clipped OFDM-UWB
signal. A half-clipped OFDM-UWB signal has reduced distortion although half of its amplitude was
removed. The optical transmitter applies a higher gain to the OFDM-UWB signal balancing an increase
of signal distortion and the raise of the power level of the UWB subbands.
The chirp-managed-based transmission achieves an improvement of 10 km of the maximum transmis-
sion distance compared to the use of a directly-modulated laser (DML). It is concluded that the quality
OFDM-UWB optical signal shows a higher resilience to the fiber chromatic dispersion in the CML-based
transmission.
keywords: Chirp-managed lasers, directly-modulated lasers, fiber-to-the-home networks, optical com-
munications, orthogonal frequency-division multiplexing ultra-wideband radio signals.
ii
Resumo
A utilização de redes de fibra até à casa do cliente (FTTH) tem sido proposta como o meio de distribuição
de sinais rádio de banda ultra-larga (UWB) nas futuras redes pessoais sem fios (WPAN) para a difusão
de aplicações de alto débito. Nesta dissertação, investiga-se a transmissão de sinais rádio de banda
ultra-larga usando multiplexagem ortogonal por divisão na frequência (OFDM-UWB) em redes FTTH
utilizando lasers com chirp controlado (CML). Os principais parâmetros que influenciam a degradação
do desempenho da transmissão de sinais OFDM-UWB em redes FTTH são identificados e optimizados.
Os principais problemas associados à transmissão são analisados e a distância máxima alcançada para
uma qualidade mínima requirida é avaliada.
A optimização do desempenho mostra que o transmissor óptico origina um sinal OFDM-UWB semi-
cortado. Este sinal apresenta distorção reduzida apesar de ter sido removida metade da sua amplitude.
O transmissor aplica um maior ganho ao sinal OFDM-UWB equilibrando uma maior distorção de sinal
com o aumento do nível de potência das sub-bandas UWB.
Obtém-se uma melhoria de 10 km no alcance máximo de transmissão usando CML em vez de lasers
modelados directamente. Conclui-se que o sinal óptico OFDM-UWB mostra maior resistência à disper-
são cromática da fibra quando se usa CML.
Palavras-chave: Comunicações ópticas, laser com chirp controlado, laser modulado directamente, re-
des de fibra até à casa do cliente (FTTH), sinais de rádio de banda ultra-larga usando multiplexagem
ortogonal por divisão na frequência.
iii
Contents
Acknowledgments i
Abstract ii
Resumo iii
Contents iii
List of figures vi
List of tables ix
List of acronyms xii
List of symbols xiv
1. Introduction 1
1.1. Scope of the work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2. Objectives and structure of the dissertation . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3. Original contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2. OFDM-UWB Radio Signals 6
2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2. OFDM-UWB radio signals mathematical formulation . . . . . . . . . . . . . . . . . . . 6
2.2.1. Baseband and bandpass OFDM . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.2. Cyclic prefix introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.3. Discrete Fourier transform implementation of OFDM . . . . . . . . . . . . . . . 8
2.2.4. OFDM transmitter and receiver . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.5. UWB standard for OFDM (OFDM-UWB) . . . . . . . . . . . . . . . . . . . . . 10
2.3. Time and frequency analysis of OFDM-UWB radio signals . . . . . . . . . . . . . . . . 12
2.4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
iv
3. OFDM-UWB FTTH network model 18
3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2. FTTH network architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2.1. OFDM-UWB transmitter, gain and DC current . . . . . . . . . . . . . . . . . . 20
3.2.2. Clipper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2.3. Directly modulated laser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2.4. Optical Spectrum Reshaping filter . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.5. Optical filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2.6. Optical amplifier EDFA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2.7. Optical fiber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.8. Power splitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.9. PIN Photodetector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.10. Receiving filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.11. OFDM-UWB receiver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3. Performance evaluation methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3.1. Error Vector Magnitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3.2. Semi-analytical Gaussian approach . . . . . . . . . . . . . . . . . . . . . . . . 28
3.4. Effects of clipping in an OFDM-UWB radio signal . . . . . . . . . . . . . . . . . . . . 29
3.5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4. OFDM-UWB FTTH Network Optimization 34
4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2. Model of the system to be optimized . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.3. Optimization of the currents in single subband transmission . . . . . . . . . . . . . . . . 36
4.3.1. Back-to-back configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.3.2. Performance with fiber transmission . . . . . . . . . . . . . . . . . . . . . . . . 38
4.4. Optimization of the currents in multi-subband transmission . . . . . . . . . . . . . . . . 41
4.4.1. Subbands used: 1,2 and 1,2,3 . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.4.2. Subbands used: 1,2,3,4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.5. Optical spectrum reshaper filter optimization . . . . . . . . . . . . . . . . . . . . . . . . 50
4.6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5. Transmission performance of the multi-subband OFDM-UWB signal in the optimized FTTH
network 56
5.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
v
5.2. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.2.1. Assessment of the distortion inflicted by the OFDM-UWB FTTH network . . . . 59
5.2.2. Influence of the optical noise generated by the EDFA at the central node . . . . . 60
5.2.3. Influence of optical and electrical noise . . . . . . . . . . . . . . . . . . . . . . 62
5.3. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6. Final conclusion 64
6.1. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.2. Suggestions for future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
A. Directly-modulated laser 67
A.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
A.2. Description of the directly modulated laser . . . . . . . . . . . . . . . . . . . . . . . . . 67
A.3. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
B. Optimization results 73
C. References 82
vi
List of Figures
2.1. OFDM symbol with cyclic prefix. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2. Block diagram of the OFDM transmitter. . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3. Block diagram of the OFDM receiver. . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4. Frequency band allocation for the OFDM-UWB system. . . . . . . . . . . . . . . . . . 10
2.5. Amplitude response of a 6th-order low-pass Bessel filter. . . . . . . . . . . . . . . . . . 12
2.6. OFDM-UWB signal using subband 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.7. OFDM-UWB current signal using subband 1 with a mean of 60 mA and a peak-to-peak
current of 80 mA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.8. PSD of an OFDM-UWB signal using subband 1. . . . . . . . . . . . . . . . . . . . . . 14
2.9. Close-up of the PSD of an OFDM-UWB signal using subband 1. . . . . . . . . . . . . . 14
2.10. Mean, maximum and minimum values of the PAPR for a 2lseq deBruijn bit sequence
mapped into QPSK symbols in an OFDM-UWB signal using subband 1. . . . . . . . . . 15
3.1. Schematic of the FTTH network. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2. Block diagram of the FTTH network. . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.3. Block diagram of the equivalent model of the FTTH network from single wavelength
view point. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.4. Static response of the directly modulated laser. . . . . . . . . . . . . . . . . . . . . . . . 21
3.5. Constellation diagram for QPSK. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.6. OFDM-UWB current signal using subbands 1,2,3,4. . . . . . . . . . . . . . . . . . . 30
3.7. PSD of OFDM-UWB current signal without clipping using subbands 1,2,3,4. . . . . . 30
3.8. EVM of the OFDM-UWB signal using subbands 1,2,3,4 as a function of the percent-
age of the bottom half removal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.9. Example of an OFDM-UWB current signal using subbands 1,2,3,4 after clipping. . . . 31
3.10. PSD of the OFDM-UWB current signal using subbands 1,2,3,4 with 40%, 80% and
100% of bottom half clipped. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.1. Block diagram of the equivalent FTTH network adequate for an optimization. . . . . . . 36
vii
4.2. Contour of the representation of the required OSNR as a function of the current level in
the back-to-back configuration for subband 1. . . . . . . . . . . . . . . . . . . . . . . . 37
4.3. OFDM-UWB current signal of three consecutive OFDM-UWB symbols for subband 1
using optimal current levels for the back-to-back configuration. . . . . . . . . . . . . . . 38
4.4. OSNR and EVM as a function of the SSMF length for the optimal current levels in
single-subbands 1, 2, 3 and 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.5. Psb of single subbands 1, 2, 3 and 4 for optimal current levels as a function of the SSMF
length. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.6. Required OSNR as a function of the SSMF length for the optimal current levels for
subbands 1,2 and 1,2,3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.7. Psb as a function of the SSMF length for multi-subband transmission of 1,2 and 1,2,3. 45
4.8. Required OSNR as a function of the SSMF length for the optimal current levels for
subbands 1,2 and 1,2,3 comparing to the single subband transmission of 2 and 3. 46
4.9. Required OSNR as a function of the SSMF length for the optimal current levels for sub-
bands 1,2 and 1,2,3 comparing the Bessel and rectangular filters use at the OFDM-
UWB transmitter and receiver. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.10. BER contour for an OSNR = 15 dB as a function of the bias and peak-to-peak current in
the back-to-back configuration for subband group 1,2,3,4. . . . . . . . . . . . . . . . 47
4.11. Required OSNR and EVM as a function of the SSMF length for the optimal current
levels in transmission of subbands 1,2,3,4. . . . . . . . . . . . . . . . . . . . . . . . . 49
4.12. Required OSNR as a function of the SSMF length for the optimal current levels in trans-
mission of subband groups 1,2,3,4 with maximum and minimum PAPR. . . . . . . . . 49
4.13. OSNR as a function of the SSMF length for the optimal current levels in multi-subband
groups 1,2, 1,2,3 and 1,2,3,4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.14. Frequency chirp of the DML output when modulated by an OFDM-UWB signal using
subband group 1,2,3,4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.15. OFDM-UWB current signal of three consecutive OFDM-UWB symbols for subband
group 1,2,3,4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.16. Required OSNR for BER = 10−4 as a function of the peak-to-peak current in 80 km fiber
transmission with an optimized OSR. . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.17. Required OSNR for BER = 10−4 as a function of the peak-to-peak current in 100 km
fiber transmission with an optimized OSR. . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.18. PSD of the OFDM-UWB signal in several points of the network. . . . . . . . . . . . . . 54
5.1. BER as a function of the SSMF length of the FTTH network considering N=64. . . . . . 58
viii
5.2. BER as a function of the SSMF length of the FTTH network considering N=32. . . . . . 58
5.3. BER as a function of the SSMF length of the FTTH network considering N=16. . . . . . 59
5.4. EVM value at the OFDM-UWB receiver for the CML and DML-based transmission. . . 60
5.5. Constellation of the received symbols from subbands 1, 2, 3 and 4 of CML and DML-
based 100 km transmission. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
A.1. Frequency characteristics of the assemblage effects on the DML. . . . . . . . . . . . . . 70
A.2. Static response of the MQW laser. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
A.3. Dynamic response of the DML when modulated by small current signals. . . . . . . . . 71
B.1. Contour of the representation of the required OSNR as a function of the current level for
subband 1 considering a fiber transmission from 0 to 100 km. . . . . . . . . . . . . . . . 74
B.2. Contour of the representation of the required OSNR as a function of the current level for
subband 2 considering a fiber transmission from 0 to 100 km. . . . . . . . . . . . . . . . 75
B.3. Contour of the representation of the required OSNR as a function of the current level for
subband 3 considering a fiber transmission from 0 to 100 km. . . . . . . . . . . . . . . . 76
B.4. Contour of the representation of the required OSNR as a function of the current level for
subband 4 considering a fiber transmission from 0 to 100 km. . . . . . . . . . . . . . . . 77
B.5. Contour of the representation of the required OSNR as a function of the current level for
subbands 1 and 2 considering a fiber transmission from 0 to 100 km. . . . . . . . . . . . 78
B.6. Contour of the representation of the required OSNR as a function of the current level for
subbands 1, 2 and 3 considering a fiber transmission from 0 to 100 km. . . . . . . . . . . 79
B.7. Contour of the representation of the required OSNR as a function of the current level for
subbands 1, 2, 3 and considering a fiber transmission from 0 to 80 km. . . . . . . . . . . 80
B.8. BER contour for an OSNR = 15 dB as a function of the current levels for subbands 1, 2,
3 and 4 considering a fiber transmission from 0 to 100 km. . . . . . . . . . . . . . . . . 81
ix
List of Tables
2.1. PAPR mean, maximum and minimum values for different subbands used. . . . . . . . . 16
4.1. Noiseless EVM of the single subband OFDM-UWB signal in several points of the system
in the back-to-back configuration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2. Optimal current levels for single subband transmission and corresponding required OSNR
for BER=10−4 for the back-to-back configuration. . . . . . . . . . . . . . . . . . . . . . 38
4.3. Optimal current levels for the fiber transmission using subband 1 and corresponding
required OSNR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.4. Optimal current levels for the fiber transmission using subband 2 and corresponding
required OSNR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.5. Optimal current levels for the fiber transmission using subband 3 and corresponding
required OSNR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.6. Optimal current levels for the fiber transmission using subband 4 and the corresponding
required OSNR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.7. Noiseless EVM of the OFDM-UWB signal in several points of the system for the multi
subband transmission 1,2 in the back-to-back configuration. . . . . . . . . . . . . . . 43
4.8. Noiseless EVM of the OFDM-UWB signal in several points of the system for the multi
subband transmission 1,2,3 in the back-to-back configuration. . . . . . . . . . . . . . 43
4.9. Optimal current levels for the multi-subband fiber transmission of 1,2 and the corre-
sponding minimum required OSNR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.10. Optimal current levels for the multi-subband fiber transmission of 1,2,3 and the corre-
sponding minimum required OSNR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.11. Noiseless EVM of the OFDM-UWB signal in several points of the system for the multi
subband transmission 1,2,3,4 in the back-to-back configuration. . . . . . . . . . . . . 48
4.12. Optimal current levels for a multi-subband fiber transmission of 1,2,3,4 and the corre-
sponding minimum required OSNR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.13. OSR filter optimization results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
x
5.1. OSNR calculated at the DEMUX input in the CML-based transmission. . . . . . . . . . 61
A.1. Ortel MQW laser parameter values. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
xi
List of acronyms
ASE Amplified Spontaneous Emission
BER Bit Error Ratio
BPSK Binary PSK
CML Chirp-Managed Laser
DAB Digital Audio Broadcasting
DAC Digital-to-Analog Conversion
DC Direct Current
DFT Discrete Fourier Transform
DML Directly-Modulated Laser
DVB Digital Video Broadcasting
DWDM Dense Wavelength Division Multiplexing
ECMA European Computer Manufacturers Association
EDFA Erbium-Doped Fiber Amplifier
ER Extinction Ratio
EVM Error Vector Magnitude
FCC Federal Communications Comission
FFT Fast Fourier Transform
FM Frequency Modulation
FTTH Fiber-To-The-Home
HDTV High-Definition Television
IDFT Inverse Discrete Fourier Transform
IEEE Institute of Electrical and Electronics Engineers
IFFT Inverse Fast Fourier Transform
IM Intensity Modulation
IR-UWB Impulse Radio Ultra-Wideband
xii
ISI Inter-Symbol Interference
LP Low-Pass
LR-PON Long-Reach Passive Optical Network
MC Monte Carlo
M-PSK M-ary Phase Shift Keying
MQW Multiple Quantum Wells
NRZ Non-Return-to-Zero
OFDM Orthogonal Frequency Division Multiplexing
OFDM-UWB Orthogonal Frequency Division Multiplexing Ultra-Wideband
OFDM-UWB RX OFDM-UWB Receiver
OFDM-UWB TX OFDM-UWB Transmitter
OSNR Optical Signal-to-Noise Ratio
OSR Optical Spectrum Reshaper
PAM Pulse Amplitude Modulation
PAPR Peak-to-Average Power Ratio
PON Passive Optical Networks
PSD Power Spectrum Density
PSK Phase Shift Keying
QAM Quadrature Amplitude Modulation
QPSK Quadrature PSK
PIN Positive-Intrinsic-Negative
RF Radio Frequency
RIN Relative Intensity Noise
RMS Root-Mean-Square
SAGA Semi-Analytical Gaussia Approach
SSMF Standard Single Mode Fiber
UWB Ultra-Wideband
VA Variable Attenuator
WPAN Wireless Personal Area Network
xiii
List of symbols
Amux losses from the multiplexer/demultiplexer
Anr nonradiative recombination rate coefficient
Aspl losses from the splitter
Atotal minimum losses from the FTTH network
Ax losses from connectors and splices
B radiative recombination rate coefficient
B -10 dB bandwidth of a signal
B0 reference optical bandwidth of 0.1 nm
C Auger recombination coefficient
I DC current applied to the OFDM-UWB signal in the FTTH network
c light velocity
D dispersion parameter of the SSMF
Ein(t) electrical field at the EDFA input
Eout(t) electrical field at the EDFA output
fc center frequency of a signal
fn,e noise factor
fRF central frequency of a OFDM passband signal
f equivalent baseband frequency
gedfa gain of an EDFA
gedfa, CN gain of the EDFA at the central node
gedfa, LE gain of the EDFA at the local exchange
g gain applied to the OFDM-UWB signal in the FTTH network
gc gain parameter
h constant of Planck
xiv
iPIN(t) output current of the PIN photodetector
kB Boltzmann constant
lseq length of the bit sequence to be mapped in an OFDM signal
L length of the optical fiber
n order of the OSR filter
N number of subscribers in the FTTH network
N0 carrier density at transparency
Nsc number of subcarriers in an OFDM symbol
nsp spontaneous emission noise factor
PCN imposed mean power at the output of the central node
pPIN(t) optical power at the PIN input
POF mean power of the signal at the output of the optical filter that simulates the multiplexer
Popt mean optical power at the VA input
Psb mean power of a subband
Rb bias resistor of the photodetector
Rλ responsivity of the PIN
Sideal,r expected complex value of a demodulated symbol
Smeas,r complex of the demodulated value of the actual received symbol
sk Complex symbol that modulate an OFDM subcarrier
sRF (t) passband OFDM signal
S dispersion slope of the SSMF
s(t) baseband OFDM signal
xv
Tg guard interval of an OFDM symbol
T room temperature
t time signal
Ts duration of an OFDM symbol
Ts sampling period of an OFDM signal
Ttot duration of an OFDM symbol with cyclic extension
Vact volume of the active region
α loss coefficient of the SSMF
αc linewidth enhancement factor
β spontanerous emission factor
β2 group-velocity dispersion coefficient
∆ f separation in frequency of OFDM subcarriers
η differential quantum efficiency
ηi internal quantum efficiency
ε gain compression factor
δ [k− l] Kronecker delta symbol
Γ confinement factor
λ optical wavelength
ν optical frequency
νc central frequency of the OSR filter
νg bandwidth at -3 dB of the OSR filter
ν0 optical carrier frequency
τp photon lifetime
υ optical frequency
ϕk function in time that defines a OFDM subcarrier
xvi
1. Introduction
1.1. Scope of the work
In the near future, indoor communications will be supported by a single digital wireless network. Such
network requires: high data rates, so that several high-definition television (HDTV) channels and new
advanced multimedia-based applications could be provided; low cost, for a consumer massification;
and low power consumption, for applications using battery-powered devices. Due to its inherent high
capacity, ultra-wideband (UWB) signals have been proposed as a solution to satisfy these requirements
[1].
The UWB technology has been standardized by the U.S. Federal Communications Comission (FCC) [2],
which defines UWB as any transmission scheme that occupies a fractional bandwidth greater than 0.2
or a signal bandwidth of more than 500 MHz. The fractional bandwidth is defined as B/ fc, where B
represents the -10 dB bandwidth and fc is the center frequency of a signal.
There has been a growing interest in this technique due to its features such as high bit-rate over short-
range broadcasting, low self interference, tolerance to multi-path fading, low probability of interception
and the possibility of coexistence with other deployed wireless technologies [3].
UWB signals are suited for high-data-rate wireless personal area networks (WPAN) which can be defined
as networks that transmit at data rates ranging from 100 to 500 Mbps within a distance of 20 m. Recently,
IEEE 802.15.3 standard task group has established a study group [4] to define a new physical layer
concept for high-data-rate WPAN applications.
Although FCC has regulated the spectrum and transmitter power levels for an UWB signal, there is
currently no standard for an UWB transmission scheme. A possible approach for the implementation
of UWB technology is based on single-carrier system or impulse radio communications. Impulse radio
UWB (IR-UWB) refers to the generation of a series of impulselike waveforms of very short duration.
Each pulse occupies a bandwidth that fits the FCC requirements. The information is modulated directly
1
into the sequence of pulses using, for example, pulse amplitude modulation (PAM) [5]. However, a
receiver of the single-carrier system requires a complex receiver with RF and analog circuits.
A multi-carrier approach has been proposed to overcome the drawback of single-carrier systems. It
consists in dividing the UWB frequency band from 3.1 to 10.6 GHz into several smaller bands, named as
subbands [1]. A part of the information to be transmitted is given to each subband. Within each subband,
different types of data modulation can be adopted. WiMedia Alliance suggested selecting Orthogonal
Frequency Division Multiplexing (OFDM) for high-speed UWB wireless [6], leading to OFDM-UWB
signals. Later, European Computer Manufacturers Association (ECMA) adopted the WiMedia Approach
and ratified ECMA-368 standard [7].
One great advantage of multi-carrier UWB is that it is software configurable, so any frequency bands
can be turned off to meet specific spectral requirements. The OFDM-UWB solution shows flexibility
to provide multiple access, tolerance to multi-path fading and inter-symbol interference. The OFDM
technique has been used in digital video broadcasting (DVB) and digital audio broadcasting (DAB)
which allows achieving a fast and low cost integration.
Due to very low-power, wide bandwidth and high frequency signals characteristic of UWB, the trans-
mission of UWB over coaxial cable is not feasible because of its exceedingly high cost and high rate
of attenuation for a wideband signal. However, optical fiber provides an efficient alternative for the
transmission of UWB signals due to its low cost, wide bandwidth and low loss.
The distribution of OFDM-UWB radio signals over a fiber-to-the-home (FTTH) network has been pro-
posed in [8],[9]. This approach present several advantages: FTTH networks provide enough bandwidth
to distribute a large number of UWB signals; no transmodulation and frequency up-conversion is re-
quired for a fiber-to-wireless conversion at the subscriber premises; and, FTTH networks are transparent
to the UWB implementation employed.
One of the most attractive architectures of FTTH networks is the passive optical network (PON). PON
consists in a network shared among many costumers which increases its cost-effectiveness. Research
attention has been recently focused on new types of optically amplified long-reach (100 km) PONs (LR-
PONs) [10]. These LR-PONs would replace the separate metro and access portions of the network with
a single, integrated, all-optical communication system. The European Union-funded project PIEMAN
is performing physical layer research aimed at achieving a bandwidth per wavelength of up to 10 Gbps
downstream and upstream, where each 10 Gbps is shared by up to 512 costumers [11].
An experimental demonstration of OFDM-UWB distribution in FTTH networks is shown in [8],[12],[13].
2
In these works, significant performance degradation was observed when the optical fiber length in-
creases.
External modulation schemes, using for example Mach Zehnder modulator, have been adopted in optical
transmitters. In addition to the expensive use of external modulators, dispersion compensation modules
are required to overcome the transmission dispersion effects along a standard single-mode fiber (SSMF).
The use of external modulation is very inconvenient in FTTH networks where cost-effectiveness is a key
issue. In [14] the operation of a gigabit-PON (GPON) with 2.488 Gbps downstream and 1.244 Gbps
upstream over 135 km using external modulation techniques has been reported.
Currently, there has been a growing interest in utilizing directly modulated lasers (DML) because of their
potentially low cost, compact size, low power consumption and high optical output power characteristics.
Therefore, DML becomes a solution to overcome the cost issue of the optical transmitters.
The transmission of OFDM-UWB signals over single-mode fiber of 5 km length [15] and multimode fiber
of 400 meters length [16][17] using directly modulated lasers has been experimentally demonstrated and
the performance degradation of OFDM-UWB transmission over fiber has been experimentally investi-
gated and theoretically analyzed in [18]. In this case, the optical transmission was accomplished for a
maximum distance of 40 km while, in this dissertation, a maximum optical link of 100 km is established.
It was shown that direct modulation is limited by laser bandwidth, linewidth, stability, and relative inten-
sity noise (RIN) [19]. In [20], it was shown that RIN significantly degrades the OFDM-UWB over fiber
system, particularly for the higher frequency UWB channels.
The directly-modulated laser has been investigated as optical transmitter in FTTH networks [21]. It
reduces the network complexity, the installation and maintenance cost without compromising the flex-
ibility of the network [22]. However, the DML current modulation gives rise to high frequency chirp:
the intensity modulation is accompanied by frequency modulation [23]. This results in a broad spectrum
that severely limits the transmission distance due to the effect of chromatic dispersion.
Through a tuned optical spectrum reshaper filter (OSR) placed after the DML, bandwidth and chirp
reductions are achieved after the DML modulation. This filter performs a frequency modulation (FM) to
intensity modulation (IM) conversion, leading to a signal more robust to the dispersion of the fiber.
The grouping of the DML and the OSR composes the chirp-managed laser (CML) as presented in [24].
The optical signal obtained at the OSR output is more resilient to inter-symbol interference caused by
the dispersion effects associated with the transmission through the fiber.
The CML technique has been demonstrated in 10 Gbps transmission over 200 km of single-mode fiber
without dispersion compensation [25] and 675 km transmission using a combination of electronic dis-
3
persion compensation and tunable dispersion compensation [26]. Recently, it has been experimentally
demonstrated the 40 Gbps CML transmission over 20 km of single-mode fiber [27]. It was also achieved
a bit error ratio (BER) of 10−6 in the CML transmission of 40 Gbps in the metro network consisting of
a 240 km long link with dispersion compensating fiber.
In this work, the transmission of OFDM-UWB signals through a FTTH network using chirp-managed
lasers is proposed and investigated as a means to overcome the cost issue of the optical transmitter
and extend the transmission distance achieved by using conventional directly modulated lasers. The
CML transmitter is a highly practical solution to meet simultaneously the size, power, capacity and cost
requirements of the optical links in FTTH networks.
1.2. Objectives and structure of the dissertation
The objective of this dissertation is the analysis and performance optimization of the transmission of
OFDM-UWB radio signals in FTTH networks using chirp-managed lasers. The time and frequency
characteristics of OFDM-UWB signals are investigated. The influence of the each component of the
FTTH network on the OFDM-UWB signal quality is assessed. The key system parameters that influence
the transmission performance are identified and optimized through extensive numerical simulations and
the maximum transmission distance achieved for a given minimum quality is assessed.
The dissertation is structured as follows. Chapter 2 presents a detailed description of OFDM-UWB
signals. The mathematical formulation of OFDM-UWB signals is presented and the FCC standard for
UWB is detailed. A time and frequency analysis of OFDM-UWB signals is accomplished and the main
characteristics of the signals are identified.
In chapter 3, the FTTH network is analyzed as a means to transmit OFDM-UWB signals. An equivalent
model of the network is presented and each component is described - a careful description of the DML
operation is performed in Appendix A. The methods of performance evaluation of OFDM-UWB signals
in FTTH networks are presented in order to be used in the following chapters. At the end, chapter 3
analyzes the time and frequency effects of the non-linear operation of clipping performed at the FTTH
network transmitter.
Chapter 4 accomplishes the optimization of the current levels that modulate the optical transmitter in
the DML-based single and multi-subband transmission in the back-to-back configuration and with fiber
transmission. The influence of each component of the network is assessed and the impact of the si-
multaneous transmission of several subbands is analyzed. Based on the results from the current levels
4
optimization, the optical spectrum reshaper filter is introduced in the system and its characteristics are
optimized.
Chapter 5 analyzes the performance of the transmission of OFDM-UWB signals in the FTTH network
based on the outcome of chapter 4. The maximum transmission distance assuring a defined minimum
quality is calculated, and the DML and CML-based transmissions are compared.
Chapter 6 draws the main conclusions of this dissertation and topics for further investigations are sug-
gested.
1.3. Original contributions
In the analysis performed in this work, several original contributions were introduced relative to other
studies in the field. In the following, the list of the most important contributions of this work are pre-
sented:
1. time and frequency analysis of the non-linear operation (clipping) accomplished to OFDM-UWB
signals at the FTTH transmiter and its influence on the error vector magnitude.
2. optimization of the input currents of the DML in the single and multiple subband OFDM-UWB
transmission in the back-to-back configuration and in the configuration with fiber transmission for
lengths up to 100 km.
3. analysis of the distortion experienced by the OFDM-UWB signal in the clipper, in the DML and
by the fiber transmission.
4. analysis of the performance degradation of a single and multiple subband OFDM-UWB DML-
based transmission in the back-to-back configuration and in the configuration with fiber transmis-
sion for lengths up to 100 km with optimized input currents.
5. optimization of the optical spectrum reshaper filter parameters in the multi-subband OFDM-UWB
transmission in the configuration with fiber transmission of 80 and 100 km.
6. analysis of the performance degradation improvement of the multi-subband OFDM-UWB CML-
based transmission compared to the DML-based transmission.
7. analysis of the maximum transmission distance of an OFDM-UWB signal along FTTH network
achieved for a minimum defined quality for the CML-based and the DML-based optical transmit-
ters.
5
2. OFDM-UWB Radio Signals
2.1. Introduction
As mentioned in the Chapter 1, OFDM-UWB radio signals have been proposed to support high data rate
applications in the near future.
In this chapter, OFDM-UWB signals are introduced analytically in section 2.2: the general definition
of baseband and bandpass OFDM is detailed along with the implementation of OFDM in the UWB
standard. A time and frequency analysis of OFDM-UWB signals is presented in section 2.3.
2.2. OFDM-UWB radio signals mathematical formulation
2.2.1. Baseband and bandpass OFDM
OFDM is a form of multicarrier modulation, where a single data stream is transmitted over a number of
lower rate subcarriers [28],[29]. OFDM increases the robustness of the signal against frequency selective
channels and narrowband interference.
Let skk=0,1,... be the complex information symbols to be transmitted, obtained by mapping the binary
input data into a symbol stream using a signal format such as quadrature amplitude modulation (QAM)
or phase-shift keying (PSK). Each information symbol modulates different subcarriers separated by ∆ f
during a limited period of time.
The baseband OFDM signal is formed by a sequence of OFDM symbols, each one with a fixed duration
of Ts. Each OFDM symbol has Nsc subcarriers, corresponding to a group of Nsc information symbols,
and it can be expressed as
s(t) =Nsc−1
∑k=0
ske j2π fkt =Nsc−1
∑k=0
skϕk, for 0≤ t ≤ Ts, (2.1)
where fk = k∆ f , and
6
ϕk =
e j2π fkt , if 0≤ t ≤ Ts
0, otherwise(2.2)
The orthogonality condition, characteristic of OFDM signals, imposes an OFDM symbol duration given
by
Ts ·∆ f = K , K ∈ N (2.3)
It is easily proven that the set of functions ϕk are orthogonal along an OFDM symbol:
1Ts
∫ Ts
0ϕkϕ
∗l dt
=1Ts
∫ Ts
0e j2π( fk− fl)tdt
=1Ts
∫ Ts
0e j2π(k−l)∆ f ·tdt
= δ [k− l] (2.4)
where δ [k− l] is the Kronecker delta symbol defined by
δ [k− l] =
1, if k = l
0, otherwise(2.5)
Using the property of orthogonality, an OFDM symbol can be easily demodulated.
2.2.2. Cyclic prefix introduction
A guard interval with duration Tg is introduced between consecutive OFDM symbols to prevent Inter-
Symbol Interference (ISI) due to the delay spread of the transmission channel. A fraction of the guard
interval is used for transmitting a copy of the final part of the OFDM symbol in order to eliminate
intersubcarrier interference (see figure 2.1). This copy is named cyclic prefix. Thus, the baseband OFDM
signal can be written as
s(t) =
s(t +Ts), if −Tg ≤ t ≤ 0
s(t), if 0≤ t ≤ Ts
(2.6)
With the cyclic extension, the actual OFDM symbol duration is increased to Ttot = Ts +Tg, resulting in
7
the symbol rate of
RS =Nsc
Ttot=
Nsc
Ts +Tg(2.7)
Figure 2.1.: OFDM symbol with cyclic prefix.
In order to obtain OFDM bandpass signals, a carrier with frequency fRF is modulated by the OFDM
baseband signal. This frequency upconversion makes a complex-to-real conversion of s(t) as follows
sRF(t) = ℜ
s(t)e j2π fRF t= ℜs(t) · cos(2π fRFt)−ℑs(t) · sin(2π fRFt) (2.8)
where sRF (t) is the OFDM symbol transmitted with carrier frequency fRF .
2.2.3. Discrete Fourier transform implementation of OFDM
If s(t) is sampled with a sampling period of Tsa = Ts/Nsc it follows that
Sn = s(nTsa) =Nsc−1
∑k=0
ske j2π fknTsNsc =
Nsc−1
∑k=0
ske j 2πknNsc = IDFTsk (2.9)
where IDFT denotes the inverse discrete Fourier transform.
Thus, the discrete values of s(t) are the Nsc-point IDFT of the information symbols sk. The IDFT/DFT
approach simplifies the implementation of an OFDM transmitter and receiver (OFDM-TX and OFDM-
RX, respectively). It avoids an extremely complex architecture, involving many oscillators and filters,
that would be necessary at both OFDM-TX and OFDM-RX in the case of an implementation based on a
large number of subcarriers modulated by each symbol sk.
The s(t) signal is generated by a digital-to-analog conversion (DAC) of the samples coming from the
IDFT calculation of the sequence sk. The efficient implementation of the IDFT/DFT is the fast Fourier
8
transform (IFFT/FFT) which reduces the number of multiplications from N2sc to (Nsc/2)· log2(Nsc).
A zero padding is introduced within the input sequence s0, ...,sNsc−1 before computing the inverse
Fourier transform due to an intolerable aliasing from the DAC. The zero padding is implemented by
introducing additional subcarriers with zero amplitude. The zeroes are inserted in the middle of the
symbol sequence since, after the IFFT calculation, it corresponds to the output edges of the signal band,
minimizing the aliasing resulting from the DAC.
2.2.4. OFDM transmitter and receiver
In figures 2.2 and 2.3, the block diagrams of the OFDM transmitter and receiver are presented, respec-
tively.
The OFDM transmitter illustrated in figure 2.2, shows that the bit sequence generated is mapped into a
symbol stream using a signal format such as M-PSK. Afterwards, the IFFT is calculated to the symbol
stream with the insertion of the zero-padding. After the digital-to-analog conversion, s(t) is obtained and
will modulate a carrier with frequency fRF .
The OFDM receiver, shown in figure 2.3, performs backwardly the same operations as the transmitter
with the insertion of an equalizer after the FFT block. The equalization is used to compensate for the
amplitude and phase distortion induced by the transmission channel and its transfer function is estimated
from the pilot carriers information of the OFDM signal received.
Figure 2.2.: Block diagram of the OFDM transmitter.
9
Figure 2.3.: Block diagram of the OFDM receiver.
In figure 2.2, the low-pass filter placed after the digital-to-analog converter is used to reduce, or even
eliminate, the influence of the aliasing components on the system performance. At the OFDM receiver
(figure 2.3), the low-pass filters, after the beating with RF oscillators, are used to reduce the noise power
added by the transmission system and they present, usually, the same characteristics as the low-pass
filters used in the OFDM transmitter.
2.2.5. UWB standard for OFDM (OFDM-UWB)
FCC allocated a bandwidth of 7.5 GHz (3.1∼10.6 GHz) for the use by UWB signals. In the OFDM-
UWB approach, the spectrum is divided into subbands that are 528 MHz wide, forming a total of 14 (as
shown in figure 2.4), with central frequencies given by [1]
fRF [n] = 2904+528n [MHz] n = 1,2, ...,14 (2.10)
Figure 2.4.: Frequency band allocation for the OFDM-UWB system.
Each OFDM symbol in an UWB subband consists in 128 subcarriers:
• 100 data subcarriers: e.g. with BPSK, QPSK or QAM modulation.
• 12 pilot subcarriers: to provide information for the equalizer.
10
• 6 null subcarriers: to relax filter requirements.
• 10 guard subcarriers: used for different purposes.
The time duration of an OFDM-UWB symbol in one subband is defined as 312.5 ns. A 9.5 ns guard
interval is sufficient for a safe switching between two subbands and a 60.6 ns zero-padded prefix provides
both robustness against multi-path and decreases the power needs.
In this work, a quadrature phase-shift keying (QPSK) is chosen as modulation format to generate the
symbols from the bit sequence that modulate each subcarrier in an OFDM subband and its data rate per
subband is 640 Mbit/s.
A single subband OFDM-UWB signal is generated by an OFDM transmitter as presented in subsection
2.2.4. The transmitter adopts the UWB standard, namely the number of subcarriers and the OFDM-
UWB symbol duration time. The carrier frequency fRF , given by (2.10), is assigned to the transmitter
according to desired subband.
A multiple subband OFDM-UWB signal is obtained by summing the output of single subband OFDM-
UWB transmitters.
The most adequate low-pass filters for both OFDM-UWB transmitter and receiver would be a rectangular
filter with a 264 MHz bandwidth - matching the used frequencies of each subband - because it would
minimize the crosstalk between adjacent subbands and fully eliminate the noise outside the subband of
interest. However, the forementioned filter is not realizable and a 6th-order Bessel filter is adopted with
an optimized bandwidth of 315 MHz as given in [21]. This filter bandwidth reduces the distortion within
each subband caused by the filter not having a totally flat passband. It also reduces the crosstalk between
subbands and out-of-subband noise which comes from the slow transition between the passband and
stopband. The amplitude response of the filter is drawn in figure 2.5. Considering the frequencies of the
UWB subband, the amplitude response of the 6th-order Bessel filter shows a 2.5 dB difference between
the center of the subband and its edges (±264 MHz), which contributes to the distortion of the subband
of the OFDM-UWB signal.
11
Figure 2.5.: Amplitude response of a 6th-order low-pass Bessel filter.
2.3. Time and frequency analysis of OFDM-UWB radio signals
Figure 2.6 shows an example of an OFDM-UWB symbol in time using the first UWB subband ( fRF
= 3432 MHz). The signal is obtained directly from the OFDM-UWB transmitter from mapping a bit
sequence into QPSK symbols that modulate the 128 subcarriers of the UWB subband.
Considering an optical network using intensity modulation and direct detection (as in this work), the
transmitting signal modulates the intensity (power) of the carrier frequency of the optical transmitter, not
its amplitude or phase. This means that the transmitting signal must be positive and real. As observed
in figure 2.6, a signal obtained by an OFDM-UWB transmitter has both positive and negative values.
Therefore, a gain should be applied along with the adding of a DC component in order to fit the OFDM-
UWB transmitter output to the optical transmitter input restrictions.
Figure 2.7 shows an example of an OFDM-UWB current signal when a gain is applied and a DC com-
ponent added so that it has a peak-to-peak current of 80 mA and mean current of 60 mA. This current
signal meets the restrictions of the optical transmitter used in this work - a directly modulated laser.
From the OFDM-UWB current signal shown in figure 2.7, it was estimated that during 87% of the time,
the signal has an amplitude in the region of 50∼70 mA. The estimation suggests that most of the electrical
power of the signal has a low current amplitude when compared with its peaks. This result comes from
the high peak-to-average power ratio (PAPR) typical of OFDM signals which will be discussed later.
12
Figure 2.6.: OFDM-UWB signal using subband 1.
Figure 2.7.: OFDM-UWB current signal using subband 1 with a mean of 60 mA and a peak-to-peakcurrent of 80 mA.
Figure 2.8a shows the power spectrum density (PSD) of an OFDM-UWB signal. The first UWB subband
is assigned to the OFDM-UWB signal which centers the PSD in 3432 MHz with a bandwidth of 528
MHz. It is observed power outside the subband of interest that comes from the aliasing of the digital-to-
analog conversion at the OFDM-UWB transmitter. The aliasing is limited by a 6th order Bessel low-pass
filter with 315 MHz of bandwidth. If not attenuated, the aliasing power could interfere with adjacent
subbands.
The PSD shown in figure 2.8b is calculated by doing the mean over a 10 samples window of the actual
PSD - shown in figure 2.8a. This approach facilitates the analysis of the spectrum due to the smaller
variation of the samples. From this point on, all the PSD representations will be a mean over a 10
samples window of the original PSD.
13
(a) Original PSD (b) 10-samples window mean PSD
Figure 2.8.: PSD of an OFDM-UWB signal using subband 1.
Figure 2.9a shows a close-up to the PSD of the UWB subband. It is observed that there is an approx-
imately 3 dB difference between the center of the UWB subband and its edge. The round top of each
subband comes from the low-pass filters used in the OFDM-UWB transmitter not having a totally flat
passband as already discussed in section 2.2.5. Figure 2.9b shows the PSD of the OFDM-UWB current
signal when rectangular filters with a 264 MHz bandwidth are used at the OFDM-UWB transmitter. This
bandwidth fits thoroughly the UWB subband which is approximately flat due to the ideal characteristics
of the filters.
(a) Bessel filter used at OFDM-TX (b) Rectangular filter used at OFDM-TX
Figure 2.9.: Close-up of the PSD of an OFDM-UWB signal using subband 1.
As mentioned above, the high peak-to-average power ratio (PAPR) is known as one of the major draw-
backs of the OFDM signals [29].
14
The PAPR of the OFDM signal is defined as
PAPR =max
|s(t)|2
mean|s(t)|2
, (2.11)
where s(t) is given by (2.1).
As explained in [29], the theoretical PAPR maximum is 10log10(Nsc) in dB, obtained by setting sk = 1
and t = 0 in (2.1). For a system with Nsc = 128, the theoretical maximum PAPR is 21 dB.
The PAPR theoretical maximum value is excessively high, especially if it modulates a directly modulated
laser: each laser has its own limited region of modulation and an OFDM signal can be severely distorted
if it has a high PAPR 1.
Several values of PAPR were calculated when a deBruijn bit sequence of length 2lseq is mapped into
QPSK symbols that modulate each OFDM subcarrier. For a given deBruijn sequence of length 2lseq ,
the PAPR value was computed for sequences resulting from circular shifting of the original deBruijn bit
sequence 2lseq times. Figure 2.10 shows the PAPR mean, maximum and minimum for different values
of lseq using the first subband. It is observed that, for a longer deBruijn bit sequence, the PAPR mean
value increases. The maximum value found was 16.5 dB which is below the theoretical maximum, but
the graph from figure 2.10 suggests higher PAPR values for longer deBruijn bit sequences.
Figure 2.10.: Mean, maximum and minimum values of the PAPR for a 2lseq deBruijn bit sequence mappedinto QPSK symbols in an OFDM-UWB signal using subband 1.
The OFDM-UWB symbol sequence used in this work comes from the deBruijn sequence of lseq = 11
(signal format of QPSK) with the highest PAPR found. The PAPR of each OFDM-UWB signal depends
on which and how many subbands are used.
1The optical transmitter used in this work is detailed in chapter 3.
15
An OFDM-UWB signal with a high peak is more likely to be distorted by the optical transmitter and,
therefore, it represents, possibly, the worst case scenario for the optical network.
Table 2.1 shows the PAPR mean and maximum values for different subbands used. For a single-subband
transmission, the PAPR value is independent of the subband used. For multiple-subband transmission, it
is observed that the PAPR value increases with the number of subbands assigned.
Table 2.1.: PAPR mean, maximum and minimum values for different subbands used.
Subbands used meanPAPR maxPAPR minPAPR
1 13.90 15.48 12.24
2 13.91 15.49 12.32
3 13.91 15.50 12.29
4 13.90 15.49 12.33
1,2 16.67 18.45 14.97
1,3 16.61 18.32 14.92
1,2,3 18.36 20.16 16.70
1,2,3,4 19.56 21.30 17.94
2.4. Conclusion
In this chapter, OFDM-UWB radio signals were introduced and described. An analitycal approach was
followed for the definition of this type of signals in section 2.2. It was demonstrated that a general OFDM
signal can be easily obtained by calculating the IDFT of the symbol sequence to be transmitted which
substantially reduces the complexity of the OFDM implementation.
It was discussed that the original OFDM-UWB signal is not suited for optical transmission using intensity
modulation and that a modification in the mean and amplitude of the signal should be performed.
The OFDM transmitter and receiver block diagram was presented and each function block explained.
The FCC specifications were detailed to the OFDM-UWB approach, namely the 14 UWB subbands used
frequencies, the 128 subcarriers per subband and the duration of 312.5 ns of an OFDM-UWB symbol.
Time and frequency analysis of OFDM-UWB radio signals were performed in section 2.3. The power
spectrum density of a single subband OFDM-UWB signal showed that the 6th order Bessel low-pass
filters used in the OFDM-TX and OFDM-RX reduce the aliasing coming from the digital-to-analog
16
conversion although it introduces distortion to the subband caused by the filter not having a totally flat
passband. The high PAPR was mentioned as one of the major drawbacks of the OFDM-UWB radio
signals reaching 21.3 dB for an OFDM-UWB signal using subbands 1, 2, 3 and 4.
17
3. OFDM-UWB FTTH network model
3.1. Introduction
The transmission of OFDM-UWB radio signals in a fiber-to-the-home (FTTH) network is a cost-effective
solution to extend the coverage of these signals. It can support a transparent electro-optical signal conver-
sion, avoids using some expensive high-frequency electronic components required during signal trans-
mission and the optoelectronic integration technologies have their size and power consumption of the
optical UWB transmitters and receivers decreased.
For the evaluation of the impact of the transmition along the FTTH network on the OFDM-UWB signal
quality, a model of the network adapted to the OFDM-UWB signal transmission is presented and each
component of the network is described. The methods that measure the quality of an OFDM-UWB signal
are detailed and discussed.
In this Chapter, section 3.2 presents the FTTH network and each of its elements are described. In section
3.3, the methods of performance evaluation of the FTTH system are introduced and discussed, and in
section 3.4, the effects of the non-linear operation performed in the FTTH network (clipping) is anal-
ysed.
3.2. FTTH network architecture
Fiber-to-the-home is defined as a telecommunications network in which an optical path is estabilished
from the switching equipment of the telecom operator (designated as central node) to the home premises
of the subscriber. A simplified block diagram of this network is drawn in figure 3.1.
18
Figure 3.1.: Schematic of the FTTH network.
As mentioned in the Chapter 1, FTTH networks are usually based on existing passive optical networks
(PONs). A PON-based FTTH network consists of an optical line terminal (OLT) at the central node
of the service provider and a number of optical network units (ONUs) near end users. A PON is a
point-to-multipoint network which reduces the amount of fiber used.
In figure 3.2, the block diagram of the LR-PON is shown. As observed, the central node generates and
multiplexes the OFDM-UWB signal making use of dense wavelength division multiplexing (DWDM).
This technology is used for the transport of several optical carrier signals onto a single optical fiber of 80
km from the central node until the local exchange [10].
The erbium-doped fiber amplifier (EDFA) at the central node is used to impose a defined mean optical
power before the fiber transmission. At the local exchange, the EDFA compensates for the losses inflicted
by the transmission and, afterwards, the optical signal is demultiplexed and distributed to N subscribers
by a N-splitter. Depending on the geographic location of each subscriber, the optical path between the
local exchange and the premises of the subscriber may have a length between 0 and 20 km.
Since the analysis of wavelength multiplexing/demultiplexing technology is out of scope for this work,
an equivalent model of the FTTH network is adopted which neglects the crosstalk between different
wavelengths.
The goal of this dissertation is to evaluate the impact of the transmission along the FTTH network on the
OFDM-UWB signal quality. Thus, from this point on, the discussion will be focused on an individual
wavelength channel. The multiplexers/demultiplexers represented in figure 3.2 are replaced by the equiv-
alent model from the single-wavelength view point. The equivalent FTTH network model is represented
in figure 3.3.
19
Figure 3.2.: Block diagram of the FTTH network.
Figure 3.3.: Block diagram of the equivalent model of the FTTH network from single wavelength viewpoint.
3.2.1. OFDM-UWB transmitter, gain and DC current
Figure 3.3 shows, in the central node, the OFDM-UWB transmitter as described in chapter 2. Its output
is approximately symmetric around zero and its amplitude is not suited for the modulation of the directly
modulated laser. Therefore, an adaptation of the OFDM-UWB original signal to the restrictions of the
optical transmitter is necessary.
The gain g in figure 3.3 imposes the swing of the current that modulates the laser and I sets its bias. These
parameters alter the OFDM-UWB transmitter output in order to meet the optical transmitter input current
restrictions. Thus, I and g have a strong influence on the performance of the system and, therefore, should
be optimized.
3.2.2. Clipper
A clipping operation is performed to the current signal before the DML modulation in order to ensure
the variation of the OFDM-UWB current signal inside the laser linear operating region defined as the
range between 18 and 100 mA. The lower limit of the region is 18 mA because a shorter proximity
to the threshold current of the laser could lead to a slower operation of the DML. The upper value of
20
the modulation region sets the transition to a multimode emission which should avoided (the general
operation of the laser is detailed in subsection 3.2.3).
The non-linear operation of clipping introduces some distortion to the signal, but it has the advantage of
reducing the PAPR of the signal. A deeper analysis of the clipping effect is presented in section 3.4.
3.2.3. Directly modulated laser
The DML used in the FTTH network is a Multiple Quantum Wells (MQW) and it is operating at the
wavelength of 1550.2 nm. Its behaviour characteristics are well described by the rate equations that gov-
ern the interaction of photons and electrons inside the active region of the laser [31]. The rate equations
are carefully detailed in Appendix A.
The static response of the laser is shown in figure 3.4 and derived in Appendix A. It relates the output
power when a constant current is modulating the laser. It is observed that, for a current below 14.34
mA, the output power is approximately 0 because there is only spontaneous emission of photons. Above
this threshold, there is stimulated emission of photons and, therefore, the output power shows a linear
increase with the input current. This linear region is limited until 100 mA, which is the maximum input
current value before the laser presents multi-mode emission. Thus, for OFDM-UWB signal transmission
through this DML, the input current levels should remain inside the forementioned region.
Figure 3.4.: Static response of the directly modulated laser.
The DML model also considers the effects originated from the assemblage of the laser. These effects are
represented by a first order low-pass filter with a bandwidth of 6 GHz.
The dynamic response of the laser to a small current signal modulation is also shown and derived in
Appendix A. It is observed that the bandwidth of the DML depends on the bias input current: for an
21
increase of the bias current from 20 mA to 60 mA, the bandwidth at -3 dB increases approximately from
10 GHz to 25 GHz. Thus, the overall bandwidth of the DML is the result of the effects of its assemblage
and the input bias current. Due to the bandwidth limitation of the DML, the transmission of OFDM-
UWB signals is accomplished only for the first four UWB subbands ( fRF = 3.43 GHz; 3.96 GHz; 4.49
GHz; 5.02 GHz).
A simultaneous modulation in intensity (IM) and frequency (FM) is observed in the dynamic response of
the DML. This simultaneous modulation becomes a drawback when a IM signal is sent through the fiber:
the signal becomes wider in frequency due to FM and more sensitive to the dispersion of the fiber.
The instanteous frequency shift is related to carrier-induced index changes caused by different levels of
input currents. The frequency chirp is also influenced by small oscillations of the carriers number as
shown in Appendix A.
By placing a tuned optical filter at the laser output, FM to IM conversion is accomplished, increasing the
resilience to the dispersion of the fiber and reducing the inter-symbol interference caused by the optical
transmission. As demonstrated in [24], the CML reaches higher performances in the transmission of a
10 Gbps non-return-to-zero (NRZ) signal through a direct-detection optical fiber communication system,
when compared to the use of the DML.
3.2.4. Optical Spectrum Reshaping filter
As mentioned above, the optical spectrum reshaper filter plays an essential part in the chirp control.
Many possiblities for the filter type have been considered [27]:
• Butterworth: flat response in the pass band; overshoot and ringing in the step response.
• Chebyshev: better rate of attenuation in the stop-bands than Butterworth; more ringing in step
response than Butterworth.
• Bessel: minimal overshoot and ringing in the step response; slowest rate of attenuation beyond the
pass band.
• Super-Gaussian: transfer function without dispersion; easily implemented.
The super-Gaussian filter is selected to be the OSR filter because it is easily implemented and is also the
type of filter suggested in [27]. The transfer function of the filter is given by
T (ν) = exp
(− ln
(√2)∣∣∣∣ν−νc
νg/2
∣∣∣∣2n)
(3.1)
22
where ν is the optical frequency, νc is the central frequency of the OSR, νg is the bandwidth at - 3dB of
the filter and n the filter order. The bandwidth and central frequency of the OSR are optimized choosing
the first-order Gaussian filter (n = 1) for an optimal chirp control.
3.2.5. Optical filter
The optical filter placed after the CML is the equivalent model of the multiplexer from the single-
wavelength point of view. This filter has the same transfer function as the one of the demultiplexer
at the local exchange.
The optical filter at the local exchange narrows the noise spectrum of the amplified spontaneous emission
(ASE) generated by the EDFA reaching the OFDM-UWB receiver, but it also distorts the optical signal
if not wide enough. The optimal bandwidth of the filter constitutes a balance between the two mentioned
factors. Based on the outcome of [21], a 2nd order super-Gaussian filter with an optimized bandwidth of
35 GHz is considered.
An insertion loss of 2.5 dB, as given in [32], is also taken into account in the equivalent model of both
multiplexer and demultiplexer.
3.2.6. Optical amplifier EDFA
For the compensation of the power losses induced by the OSR and the optical filter that simulates the
multiplexer, an erbium-doped fiber amplifier (EDFA) sets the mean power of the optical signal at the
output of the central node to a defined value of, typically 0, 3 or 5 dBm. Thus, independently of the input
current levels of the optical transmitter, the transmission of OFDM-UWB signals have the same mean
optical power at the central node output.
The gain of this optical amplifier is given by
gedfa, CN =POF
PCN(3.2)
where POF is the mean power of the signal at the output of the optical filter that simulates the multiplexer
and PCN the imposed mean power at the output of the central node.
At the local exchange, the role of the EDFA is to compensate for the losses inflicted by the fiber trans-
mission from the central node to the local exchange, splices, connectors, multiplexer, demultiplexer and
splitter present in the FTTH network.
23
Considering the cost of an EDFA with a gain higher than 35 dB and the loss inflicted by the FTTH
network (minimum 30 dB as demonstrated later), it is considered a 30 dB gain in the optical amplifier at
the local exchange (gedfa, LE = 1000). Thus, the EDFA will not compensate totally the losses inflicted by
the network from the central node to the facilities of the subscriber.
An ideal model of the EDFA was considered: constant gain along frequency and no amplifier saturation.
The electrical field at the EDFA output is then given by
Eout(t) = Ein(t)√
gedfa (3.3)
where Ein(t) is the electrical field at the EDFA input, and gedfa is the power gain.
Along with the amplification of the optical signal, there is introduction of noise by the EDFA from
amplified spontaneous emission. The noise field is assumed zero mean additive white Gaussian noise
with a power density per polarization mode defined by the optical gain of the EDFA (gedfa) as
SASE = hν0nsp (gedfa−1) (3.4)
where h is the constant of Planck, nsp = 1.26 is the spontaneous emission noise factor and ν0 is the
optical carrier frequency.
3.2.7. Optical fiber
A standard single mode fiber (SSMF) was considered for the FTTH network.
The equivalent baseband transfer function of the SSMF is given by [33]
H f (∆ω) = exp(− jβ (∆ω)L) · exp(−α
L2
), (3.5)
where α is the loss coefficient of the fiber with a value of 0.21 dB/km, L is the fiber length and β (∆ω) is
approximated by a Taylor series as
β (∆ω)≈ β0 +β1 (∆ω)+12
β2 (∆ω)2 +16
β3 (∆ω)3 , (3.6)
in which ∆ω is given by
∆ω = ω−ω0 = 2π f , (3.7)
24
where f is the equivalent baseband frequency.
β2 is the group-velocity dispersion coefficient and is given by [33]
β2 =−λ 2D2πc
, (3.8)
where λ is the optical wavelength (1550.2 nm), c is the light velocity and D is the dispersion parameter
with a value of 17 ps/(km-nm).
β3 is given by [33],
β3 =−S−(4πc/λ 3
)β2
(2πc/λ 2)2 , (3.9)
where S is the dispersion slope with a value of 0.09 ps/(km-nm2).
3.2.8. Power splitter
The power splitter at the local exchange output divides equally the power by the N subscribers using the
same wavelength. The power loss inflicted by the splitter is given in dB by
Aspl = 10log10(N)+0.5log2(N) (3.10)
The first term represents the losses from dividing the input power by N and the second term, the losses
from insertion: 0.5 dB from each level of division.
Assuming, in this work, that N = 16, 32 or 64, the losses inflicted by the splitter (Aspl) vary from 14 to
19 dB.
Defining Amux as the losses from the multiplexer and demultiplexer, and Ax the losses from connectors
and splices with an assumed value of 3 dB, the minimum total loss inflicted by the FTTH network
from the central node until the local exchange is given by Atotal = α · 80+Ax +Amux +minAspl =
16.8+3+5+14 = 38.8 dB.
3.2.9. PIN Photodetector
The optical-to-electrical conversion is performed by a photodetector PIN due to low cost, compared to
other optical receivers type (e.g. avalanche photodiode). The low cost is essential for the implementation
of a FTTH network.
25
The output current of a PIN photodetector iPIN(t) is given by
iPIN(t) = Rλ pPIN(t) (3.11)
where pi(t) is the optical power at the PIN input and Rλ is responsivity of the PIN with a value of 1
A/W.
3.2.10. Receiving filter
The receiving filter before the OFDM-UWB receiver is a 5th-order Bessel low-pass filter with a band-
width of 10 GHz. This bandwidth ensures the filtering until the 4th UWB subband (5 GHz) without
adding distortion. This filter performs a signal conversion from current to voltage and limits the noise
generated by the optical amplification.
3.2.11. OFDM-UWB receiver
As mentioned in chapter 2, the OFDM-UWB receiver demodulates the electrical signal and, according
to the information from the pilots subcarriers, an equalization is performed. This process adjusts the
strength of certain frequencies within the received signal and it improves the performance of the system.
The equalizer uses the information from the 12 pilot subcarriers of the UWB subband - 6 subcarriers
with frequencies above fRF and 6 subcarriers below - and estimates the transfer function of the channel
by using a polynomial of degree 2.
The electrical-to-optical conversion and amplification accomplished at the premises of the subscribers
contribute to raise the level of noise. The electrical noise is generated by the resistive and active elements
present in the receiver. The power spectrum density (one-sided) of the electrical generated noise is given
by
Sc =4kBT
Rbfn,e (3.12)
where kB is the Boltzmann constant, T is the room temperature with an assumed value of 290 Kelvin, Rb
is the bias resistor of the photodetector (50 Ω) and fn,e is the noise factor that represents the influence of
the active elements of the receiver.
Typical values of the electrical power spectrum density are in the range from 1·10−24 A2/Hz to 1·10−22
A2/Hz [34]. In this case, a PSD of 2.5·10−23 A2/Hz was considered, leading to the fn,e value of 11.1
dB.
26
3.3. Performance evaluation methods
The two major causes of performance limitation of the OFDM-UWB optical system are noise and dis-
tortion. It leads to incorrect demodulation at the receiver and, consequently, to bit errors. It is indeter-
minable which has a heavier weight on the system performance because none has a dominant role and
their influence changes with the system parameterization.
Although noise is present in every component of the real system it is considered in the numerical sim-
ulations, for simplicity, that is generated optically at the EDFA and electrically at the OFDM-UWB
receiver.
The presence of non-ideal components, non-linear operations performed (e.g. clipping) and the fiber
transmission change the shape of the signal introducing distortion to the signal. In this work, the er-
ror vector magnitude (EVM) is adopted as a measure of the distortion inflicted by the system and its
definition is detailed below.
For a complete performance evaluation of the optical transmission system, which accounts for the noise
and distortion, it is used the semi-analytical Gaussian approach [35]. This method gives the bit error
ratio, which is the most common performance assessment method of a digital transmission system. This
method is also detailed in the subsection below.
3.3.1. Error Vector Magnitude
The EVM is a common figure of merit that evaluates the quality of a digital transmission. Its value ex-
presses the difference between the value of the actual received symbol (Smeas,r) and the expected complex
voltage/current value of the demodulated symbol (Sideal,r), demonstrated in figure 3.5.
Generally, the EVM takes into consideration the noise introduced by the system. In the numerical simu-
lations performed, each symbol is obtained after the equalization at the OFDM receiver and its complex
value is not affected by noise.
27
Figure 3.5.: Constellation diagram for QPSK.
The definition of EVM is the root-mean-square (RMS) value of the difference between a collection of
measured symbols and ideal symbols. The EVM can be expressed mathematically as
EVMRMS =
√√√√ 1N ∑
Nr=1 |Sideal,r−Smeas,r|2
1N ∑
Nr=1 |Sideal,r|2
(3.13)
The EVM is usually adopted as a general performance of the system. Nevertheless, since it does not take
into account the noise generated by the system, it is just a measure of the distortion inflicted on the signal
at the OFDM receiver input.
3.3.2. Semi-analytical Gaussian approach
The semi-analytical Gaussian approach (SAGA) allows evaluating the BER of each OFDM subcarrier in
direct detection optical communication systems through numerical simulation.
There are three approaches that can be followed in order to calculate the BER in a digital communication
system:
• Monte Carlo (MC) simulation: the noise is introduced in the system and propagated with the
signal. The BER is obtained by direct error counting. Although this method gives an accurate
BER estimation, it requires unacceptable computation time for low BER levels.
• semi-analytical method: the signal is propagated through the optical system. The BER and noise
are characterized analytically by using the statistical properties of the noise.
• analytical methods: signal and noise are both characterized analytically. This method neglects the
distortion inflicted by the optical system but provides a fast estimation of the BER.
28
The semi-analytical approach constitutes a balance between the MC simulation method and the analytical
method.
The SAGA method takes into account the distortion induced by the components of the system (e.g.
filter shapes) and considers both optical and electrical noise introduced at the EDFA and the receiver,
respectively.
The SAGA closed-form analytical expressions for the mean and variance of signal-noise and noise-noise
beat terms of each subcarrier at the equalizer output are detailed in [35]. The noise field is assumed zero
mean additive white Gaussian noise with a power density with contributions from the circuit generated
noise and the noise originated at the EDFA.
This method is a powerful tool for the optimization of the network under analysis and its estimates have
shown excellent agreements with the MC simulation.
3.4. Effects of clipping in an OFDM-UWB radio signal
The clipping operation accomplished before the electro-optical conversion fits the OFDM-UWB current
signal into the linear operating region of the laser. Although it causes an increase of distortion, a benefit
that comes from removing the peaks of the signal is the reduction of the PAPR.
In this section, the distortion inflicted by the clipping is assessed in terms of the EVM. It is considered
an multi-subband OFDM-UWB signal sequence transmitting simultaneously in the first 4 subbands with
a PAPR of 21 dB. The current at the clipper input is defined by a mean of 60 mA and a peak-to-peak
current of 80 mA so that the electrical signal fits the linear operating region of the laser (18∼100 mA).
Figure 3.6 presents the mentioned OFDM-UWB current signal. As observed, most of the electrical
power of the signal has a low current amplitude when compared with its peaks due to the high PAPR
(mentioned in chapter 1).
29
Figure 3.6.: OFDM-UWB current signal using subbands 1,2,3,4.
The power spectrum density of the mentioned OFDM-UWB current signal is shown in figure 3.7. It is
observed that the PSD is composed by the DC component which carries the highest power (50 dB above
the UWB subbands) and the UWB subbands.
Figure 3.7.: PSD of OFDM-UWB current signal without clipping using subbands 1,2,3,4.
The effect on the EVM of each subband in removing a percentage of the bottom half of the current is
represented in figure 3.8. It is observed that the subbands at the borders of the subband group (subbands
1 and 4) present a lower EVM than the subbands in the middle (subbands 2 and 4) due to an interference
coming from only one adjacent subbands instead of two.
As expected, figure 3.8 shows that the absence of clipping results in the lowest value of EVM of each
subband and as the bottom of the signal begins to be removed, the corresponding EVM increases. For a
clipping percentage near 100%, the EVM of each subband shows an abrupt decrease, showing a similar
value to the non-clipped signal. Figure 3.9 shows two examples of the OFDM-UWB current signal after
the clipper.
30
Figure 3.8.: EVM of the OFDM-UWB signal using subbands 1,2,3,4 as a function of the percentage ofthe bottom half removal.
(a) 50% of bottom half clipped (b) 80% of bottom half clipped
Figure 3.9.: Example of an OFDM-UWB current signal using subbands 1,2,3,4 after clipping.
Figures 3.10a, 3.10b and 3.10c show that the percentage increase of the removal of the bottom half of the
signal causes the growing of power outside the subband group. This power raising comes from intermod-
ulation products at frequencies multiple of the sum or difference between the original used frequencies
(multiples of an UWB subband bandwidth). The appearance of unwanted frequencies is a consequence
of the non-linear operation performed and, as observed for 100% removal, the possible introduction of
higher-frequency subbands (namely for subbands 10 and 11) would be affected by clipping noise gener-
ated by the first four subbands.
The result of a low EVM for a half-clipped OFDM signal was also obtained previously in [36]. For an
OFDM signal with only the odd index subcarriers modulated, half-clipping reduces the amplitude of the
subcarriers by exactly one half and all intermodulation products fall on the even subcarriers, reducing
the distortion of the signal although half of the signal amplitude was removed. Figure 3.8 shows that the
31
effect of low distortion for half-clipping an OFDM signal is replicated when modulating both odd and
even indexed subcarriers.
(a) 40% of bottom half clipped (b) 80% of bottom half clipped
(c) 100% of bottom half clipped
Figure 3.10.: PSD of the OFDM-UWB current signal using subbands 1,2,3,4 with 40%, 80% and 100%of bottom half clipped.
In conclusion, a half-clipped OFDM signal is good solution for the high PAPR problem and it shows
reduced distortion. This result is interesting for fiber transmission where there are restrictions in the
amplitude of the input current.
The optimization performed in chapter 3 combine the current characteristics of the OFDM-UWB signal
via the gain g, the DC input current I and the clipping operation. In addition to fitting the current signal to
the DML linear operating region, the clipper reduces the PAPR minimizing the distortion of the signal.
32
3.5. Conclusion
In this chapter, the FTTH architecture was described as an optical fiber network that links the switching
equipment of the operator to the facilities of the subscriber. An equivalent single-wavelength model of
the FTTH was detailed and each element characterized.
The directly modulated laser general behavior was described and its main drawbacks were identified:
high frequency chirp and limited input region of operation. The high frequency chirp effect can be
minimized by the filtering the DML output by a optical spectrum reshaping filter. This filter performs
a frequency to intensity conversion, increasing the tolerance of the optical signal to the fiber dispersion
effect.
Two performance methods were introduced: error vector magnitude and semi-analytical gaussian ap-
proach. The first one refers to a measure of the distortion inflicted by the FTTH network to the trans-
mitting signal. The SAGA method, gives the bit error ratio as the main performance measure. This
method considers the distortion inflicted by the system and the optical and electrical noise added by the
amplifiers.
The effect of clipping an OFDM-UWB was analysed. It was concluded that a half-clipped OFDM-UWB
shows low distortion and it represents a solution for the limited region of modulation of the DML and
the high peak-to-average power ratio typical of OFDM signals. However, it was demonstrated that clip-
ping originates out-of-subband noise which can significantly affect the introduction of higher frequency
subbands.
33
4. OFDM-UWB FTTH Network Optimization
4.1. Introduction
The FTTH network for OFDM-UWB signal distribution was presented in chapter 2, being the directly
modulated laser announced as the optical transmitter. Although the adoption of a DML reduces the
complexity of the network, cost of implementation and power consumption, it has a limited region of
linear operation (18-100 mA) and it exhibits an instantaneous frequency shift caused by different levels
of input current.
The limited region of linear operation is a drawback for the distribution of OFDM-UWB signals due to
the high PAPR typical of these signals as discussed in section 2.3. The frequency shifting broadens the
frequency spectrum making the transmitting signal more sensitive to the fiber chromatic dispersion.
The clipping operation accomplished before the electrical-to-optical conversion, imposes the levels of
current inside the linear operating region of the laser. The gain g and the added DC current I determine
which part of the amplitude of the original signal is clipped and, therefore, represent key parameters of
the system. In this chapter, gain g and I are optimized in order to obtain the levels of current that, along
with the clipping operation, lead to the best performance of the system.
As mentioned in chapter 2, the use of an OSR after the DML minimizes the chirp produced by the
electrical-to-optical conversion. The chirp-managed approach performs a frequency to intensity conver-
sion, making the optical signal more resilient to the fiber dispersion. For an optimal chirp control, in this
chapter, the central frequency and bandwidth of the OSR filter are optimized.
This chapter is structured as follows: in section 4.2 is presented a simplified, yet equivalent, model of
the FTTH network that is more adequate for the optimization. Sections 4.3 and 4.4, the results of the
optimization of single subband and multi-subband transmission are presented and discussed, respectively.
Section 4.5 shows the performance improvement of an optimal chirp control through the use of an OSR
filter.
34
4.2. Model of the system to be optimized
Figure 4.1 shows the block diagram of the equivalent FTTH network adequate for an optimization. The
corresponding OFDM-UWB transmitter and receiver block diagrams are detailed in chapter 1.
As mentioned before, the gain g and the DC current I in figure 4.1 impose the swing and bias current that
along with the clipping operation determine the levels of current that modulate the DML. The non-linear
operation of clipping will introduce some distortion to the OFDM-UWB signal with the advantage of
reducing the PAPR. Since these parameters have a strong influence on the performance of the system,
they are simultaneously optimized.
The OSR filter, which accomplishes the chirp control, is a 1st order Gaussian filter (n = 1 in expression
(3.1)). Its parameters are also optimized, namely bandwidth (νg) and central frequency (νc).
As in FTTH network model, a standard single mode fiber is used but it is only characterized by its
dispersive effects: D = 17 ps/nm/km and S = 0.09 ps/(km· nm2).
The photodetector and the receiving filter have the same characteristics as presented in chapter 2.
At the optical fiber end, a variable attenuator (VA) and an erbium-doped fiber amplifier (EDFA) compose
the noise loader. The noise loader is used to adjust the noise power at the PIN photodetector input so
that a defined optical signal-to-noise ratio (OSNR) is obtained. The value of the amplified spontaneous
emission density power noise per polarization and per component (phase and quadrature) is given by
SASE =Popt
4 ·osnr ·B0(4.1)
where Popt is the mean optical power at the VA input and B0 is the reference optical bandwidth of 0.1
nm. The SAGA method, detailed in section 3.3, gives the BER in terms of a desired OSNR.
In the optimization process, the metric used to determine the performance of the system is the required
OSNR value for a BER value of 10−4. This metric is well-known and it is used in similar optimizations.
In the following sections, wherever the required OSNR is mentioned, it means the required OSNR value
for BER = 10−4.
35
Figure 4.1.: Block diagram of the equivalent FTTH network adequate for an optimization.
4.3. Optimization of the currents in single subband transmission
In this section, the values of g and I are set in order to minimize the required OSNR for a BER of 10−4 at
the PIN input, calculated using the SAGA method when only a single subband is used. The optimization
was performed simultaneously for both gain g and I.
As mentioned before, I is the DC current added to the OFDM-UWB signal that defines the bias point
of the current signal in the linear operating region of the DML. The gain g determines the peak-to-peak
current of the OFDM-UWB current signal around the bias current before the clipping operation.
The OSR was not introduced yet in this analysis. The improvements of using the OSR are analyzed in
section 4.5.
4.3.1. Back-to-back configuration
Figure 4.2 represents the contour of the required OSNR value as a function of the bias and peak-to-peak
current for the system in a back-to-back configuration (no fiber transmission) using subband 1. A bias
current of 18 mA and peak-to-peak current of 350 mA was the result of the optimization with a required
OSNR value of 5.08 dB.
A bias current of 18 mA originates a half-clipped OFDM-UWB current signal. Table 4.1 presents the
noiseless EVM of the OFDM-UWB signal in several points of the network1. It shows that the difference
between the EVM of the OFDM-UWB current signal at the clipper input (-41 dB) and after half-clipping
(-40.8 dB) is negligible. Thus, a similar distortion was obtained although half of the amplitude of the
current signal was removed. This result was expected after the discussion presented in section 3.4.
An optimal peak-to-peak current of 350 mA for a back-to-back transmission exceeds the upper limit of
the linear operating region of the DML (100 mA). An increase in g and the corresponding higher clipping
1Each EVM value in Table 4.1 was calculated by demodulating the electrical OFDM-UWB signal obtained at the desiredpoint of the network.
36
of the peaks of the current signal, severely distorts the signal at the clipper output: the EVM increases
from -41 dB to -21.3 dB after the clipping operation, as shown in table 4.1. Nevertheless, a higher value
in g raises the power level of the UWB subbands increasing the resilience to the noise added by the
EDFA. The optimal g sets the balance between increasing the distortion introduced by the clipper and
raising the power level of the UWB subband.
The optimized DML input current for the back-to-back configuration using subband 1 is represented in
figure 4.3.
Figure 4.2.: Contour of the representation of the required OSNR as a function of the current level in theback-to-back configuration for subband 1.
As demonstrated in Appendix B, the transmission of single subbands 2, 3 and 4 in the back-to-back
configuration show similar optimal current levels: a bias current that leads to a half-clipped signal and
a gain g that imposes a peak-to-peak of current before the clipping operation of 350 mA. Table 4.2
shows the optimal current levels for single subband transmission of the first four UWB subbands with
the corresponding minimum required OSNR for BER = 10−4.
It is observed that subbands with higher frequencies have worse performance in terms of the required
OSNR value. The required OSNR decrease results from the larger distortion inflicted by the DML on
the OFDM-UWB current signal using higher frequency subbands. As in [20], a higher degradation was
achieved for higher frequency UWB subbands due to the RIN generated by the laser.
Table 4.1 shows that the difference in dB between the EVM of the OFDM-UWB signal at the DML
input and the EVM calculated at the receiver is increasing with the frequency of the subbands. Thus, it
is demonstrated that an OFDM-UWB current signal using higher frequency subbands experience more
distortion by the DML leading to worse performance.
37
Table 4.1.: Noiseless EVM of the single subband OFDM-UWB signal in several points of the system inthe back-to-back configuration.
Single subband used 1 2 3 4
EVM [dB] @ clipper input -41 -41 -41 -41
EVM [dB] @ clipper input (but) half-clipped -40.8 -40.5 -30.2 -30.3
EVM [dB] @ DML input -21.32 -21.38 -19.41 -19.39
EVM [dB] @ OFDM-RX -18.25 -17.82 -15.57 -14.43
EVM difference between of DML input and OFDM-RX [dB] 3.07 3.57 3.84 4.94
Figure 4.3.: OFDM-UWB current signal of three consecutive OFDM-UWB symbols for subband 1 usingoptimal current levels for the back-to-back configuration.
Table 4.2.: Optimal current levels for single subband transmission and corresponding required OSNR forBER=10−4 for the back-to-back configuration.
Single subband used 1 2 3 4
Optimal bias current 18 18 20 20
Optimal peak-to-peak current 350 350 350 350
Required OSNR [dB] for BER = 10−4 5.08 5.35 5.92 6.31
4.3.2. Performance with fiber transmission
The bias and peak-to-peak current were optimized for an OFDM-UWB signal using the first four single
subbands in the configuration with fiber transmission from 0 to 100 km.
38
As shown in Appendix B, similar results were obtained in the optimal current levels comparing with
the back-to-back configuration. Tables 4.3, 4.4, 4.5 and 4.6 present the optimal bias and peak-to-peak
current for single subbands 1, 2, 3 and 4, respectively. Figure 4.4a exhibits the performance degradation
in terms of the required OSNR for BER=10−4 for the configuration with fiber transmission. It is observed
that the transmission with a higher fiber length increases the required OSNR value leading to worse
performance. This results from the fiber dispersion effects that are stronger for fiber transmission with a
higher length.
Figure 4.4b shows the noiseless EVM of the OFDM-UWB RX signal input as a function of the SSMF
length. It can be noticed a slight decrease of the EVM value for the fiber transmission until 60 km. This
means that the fiber compensates for the distortion inflicted to the signal although, in the end, it causes a
higher performance degradation.
Table 4.3.: Optimal current levels for the fiber transmission using subband 1 and corresponding requiredOSNR.
SSMF length [km] 0 20 40 60 80 100
Optimal bias current [mA] 18 18 18 18 18 19
Optimal peak-to-peak current [mA] 350 350 350 350 350 350
Required OSNR [dB] for BER=10−4 5.08 5.19 5.42 5.73 6.14 6.66
Table 4.4.: Optimal current levels for the fiber transmission using subband 2 and corresponding requiredOSNR.
SSMF length [km] 0 20 40 60 80 100
Optimal bias current [mA] 18 18 18 18 18 18
Optimal peak-to-peak current [mA] 350 350 350 350 350 350
Required OSNR [dB] for BER=10−4 5.35 5.55 5.94 6.50 7.26 8.28
39
Table 4.5.: Optimal current levels for the fiber transmission using subband 3 and corresponding requiredOSNR.
SSMF length [km] 0 20 40 60 80 100
Optimal bias current [mA] 20 20 20 20 22 22
Optimal peak-to-peak current [mA] 350 350 350 350 350 350
Required OSNR [dB] for BER=10−4 5.92 6.22 6.82 7.69 8.92 10.80
Table 4.6.: Optimal current levels for the fiber transmission using subband 4 and the corresponding re-quired OSNR.
SSMF length [km] 0 20 40 60 80 100
Optimal bias current [mA] 22 22 22 22 19 19
Optimal peak-to-peak current [mA] 350 350 350 350 450 450
Required OSNR [dB] for BER=10−4 6.31 6.76 7.63 8.94 11.12 14.93
(a) OSNR (b) EVM
Figure 4.4.: OSNR and EVM as a function of the SSMF length for the optimal current levels in single-subbands 1, 2, 3 and 4.
For this work, the mean power of a subband (Psb) is defined as the mean power of the signal obtained by
filtering the OFDM-UWB RX input by an ideal band-pass filter centered at the frequency center of the
subband with a 528 MHz bandwidth.
Figure 4.5 presents the Psb of each single subband as a function of the SSMF length. It shows that the Psb
decreases with the length of the fiber transmission.
40
Although the losses induced by the fiber are not introduced in the optical fiber numerical model, its
dispersion effects influence the power of the optical signal through chromatic dispersion. The decrease
of Psb when the level of added noise by the EDFA is the same, makes the required OSNR increase in
order to maintain the BER = 10−4. This leads to the performance degradation demonstrated in figure
4.4a.
It can be noticed from figure 4.4a that the fiber transmission degrades the performance more severely
for higher-frequency subbands. Figure 4.5 shows that the Psb of subband 1 decreases 2 dB between the
back-to-back configuration and the configuration with 100 km fiber transmission, while Psb of subband 4
decreases 8 dB. A higher decrease in Psb leads to worse performance in terms of the required OSNR.
The results here obtained contradict the outcome of [21], in which higher frequency subbands reached
a higher performance due to the gain induced by the fiber intensity transfer function being higher for
subband 4 and lower for subband 1. It seems that the different behaviour results from the transmission
of half-clipped OFDM-UWB signals instead of non-clipped OFDM-UWB signals
Figure 4.5.: Psb of single subbands 1, 2, 3 and 4 for optimal current levels as a function of the SSMFlength.
4.4. Optimization of the currents in multi-subband transmission
In this section, the optimization is performed in a similar way as in the single-subband case, changing the
number of subbands used. The use of multiple subbands raises the interference level between subbands
that leads to a higher performance degradation.
The role of the low-pass (LP) filters in the OFDM-UWB transmitter and receiver is essential in minimiz-
ing the crosstalk between subbands and it is assessed below.
41
Since a required OSNR value for BER=10−4 is computed for each subband, the required OSNR that
indicates the performance of the multi-subband transmission, corresponds to the one of the subband with
the highest required OSNR.
4.4.1. Subbands used: 1,2 and 1,2,3
Back-to-back configuration
As demonstrated in Appendix B, a bias current of 18 mA and a peak-to-peak current of 400 mA constitute
the optimal current levels for the transmission of both subband groups, 1,2 and 1,2,3, in the back-
to-back configuration. For subband group 1,2 a required OSNR of 7.36 dB was obtained, and for the
subband group 1,2,3, 9.13 dB.
The introduction of additional subbands increase the interference between subbands which raises the
level of distortion leading to worse performance compared to the single subband transmission.
Tables 4.7 and 4.8 present the EVM value in dB in several points of the system for both groups of
subbands considered in the back-to-back configuration.
Considering the subband group 1,2, and comparing with the equivalent EVM of the single subbands
(detailed in table 4.1), it can be noticed that the EVM value at the OFDM-UWB receiver increases
approximately 1.5 dB for both subbands 1 and 2. A higher noiseless EVM value at the receiver implies
that the level of distortion is higher and therefore a worse performance is achieved in terms of required
OSNR value for BER = 10−4.
Due to the crosstalk between subbands, the required OSNR for the transmission of the single subband 2
increases from 5.35 dB to 7.36 dB when transmitted simultaneously with subband 1.
Considering subband group 1,2,3, and comparing with the single subband case, it is also observed an
increase of the EVM at the OFDM-UWB RX of 1.5 dB to subbands 1 and 3, and a 2 dB increase to
subband 2. A larger EVM increase to the middle subband results from the higher interference experi-
enced from adjacent subbands at both sides of the frequency spectrum. However, it is subband 3 the one
that imposes the lowest performance due to the more intense distortion inflicted by the laser, which can
be noticed from table 4.8. As in the single subband case, higher frequency subbands experience more
distortion when modulating a DML.
The higher distortion inflicted by the DML along with the crosstalk of the adjancent subbands make the
required OSNR of single subband 3 increase from 5.92 dB to 9.13 dB when transmitted simultaneously
with subband 1 and 2.
42
Table 4.7.: Noiseless EVM of the OFDM-UWB signal in several points of the system for the multi subbandtransmission 1,2 in the back-to-back configuration.
Subband 1 2
EVM [dB] @ clipper input -41 -41
EVM [dB] @ DML input -19.4 -19.7
EVM [dB] @ OFDM-RX -16.7 -16.1
Difference between EVM [dB] of DML input and EVM [dB] of OFDM-RX 2.7 3.6
Table 4.8.: Noiseless EVM of the OFDM-UWB signal in several points of the system for the multi subbandtransmission 1,2,3 in the back-to-back configuration.
Subband 1 2 3
EVM [dB] @ clipper input -39.5 -41 -39.8
EVM [dB] @ DML input -19.3 -19.0 -19.9
EVM [dB] @ OFDM-RX input -16.9 -15.6 -14.3
Difference between EVM [dB] of DML input and EVM [dB] of OFDM-RX 2.4 3.4 5.6
Performance with fiber transmission
Tables 4.9 and 4.10 show the optimal current levels for the fiber transmission of both subband groups
1,2 and 1,2,3 at different SSMF lengths.
The optimal peak-to-peak currents in both cases have a higher value, compared to the single subband
case, due to a higher PAPR: 15.5 dB for single subband 1, 2 or 3, 18 dB for 1,2, and 20 dB for 1,2,3
(shown in table 2.1).
Considering the subband group 1,2,3, a fiber transmission with a length higher than 60 km makes
the optimal peak-to-peak current value increase from 400 to 450 mA. This higher value in g adds more
distortion to the signal, but it also increases the power corresponding to each subband which compensates
for the stronger dispersion effects caused by a longer fiber transmission. These effects are more intense
because, in this case of the multi-subband transmission of 1,2,3, the overall bandwidth of the signal is
wider.
43
Table 4.9.: Optimal current levels for the multi-subband fiber transmission of 1,2 and the correspondingminimum required OSNR.
SSMF length [km] 0 20 40 60 80 100
Optimal bias current [mA] 18 18 18 18 18 18
Optimal peak-to-peak current [mA] 400 400 400 400 400 400
Required OSNR for BER=10−4 7.36 7.54 7.90 8.41 9.11 10.05
Table 4.10.: Optimal current levels for the multi-subband fiber transmission of 1,2,3 and the correspond-ing minimum required OSNR.
SSMF length [km] 0 20 40 60 80 100
Optimal bias current [mA] 18 18 18 18 18 18
Optimal peak-to-peak current [mA] 400 400 400 450 450 450
Required OSNR for BER=10−4 9.13 9.44 10.03 10.97 12.26 14.25
Figure 4.6 shows the required OSNR for BER = 10−4 as a function of the SSMF length with the opti-
mized current levels for multi-subband groups 1,2 and 1,2,3. In both cases, the subband with the
highest frequency in each group imposes the highest OSNR: for subband group 1,2, subband 2 has
the lowest performance, and for 1,2,3, subband 3 imposes the highest OSNR. An inferior performance
in higher-frequency subbands, results from the stronger distortion inflicted by the DML to the OFDM-
UWB current, which was demonstrated in the back-to-back configuration, and the higher decrease of the
Psb for higher frequencies, as concluded in section 4.3.
Figure 4.7 presents the Psb of each subband in the subband groups 1,2 and 1,2,3. As expected, it
shows a larger decrease for higher-frequency subbands which leads to a worse performance.
44
(a) subband group 1,2 (b) subband group 1,2,3
Figure 4.6.: Required OSNR as a function of the SSMF length for the optimal current levels for subbands1,2 and 1,2,3.
(a) subband group 1,2 (b) subband group 1,2,3
Figure 4.7.: Psb as a function of the SSMF length for multi-subband transmission of 1,2 and 1,2,3.
Comparing with the single subband transmission - figures 4.8a and 4.8b - the introduction of another
subband increases by approximately 2 dB the required OSNR value when changing from single subband
2 to subband group 1,2, and approximately 3 dB from single subband 3 to subband group 1,2,3.
The higher performance degradation, compared to the single subband transmission case, results from the
higher level of crosstalk between the used subbands, as observed for the back-to-back configuration. The
lower Psb also contributes substancially to a performance decrease.
45
(a) subbands 1,2 (b) subbands 1,2,3
Figure 4.8.: Required OSNR as a function of the SSMF length for the optimal current levels for subbands1,2 and 1,2,3 comparing to the single subband transmission of 2 and 3.
As mentioned before, the crosstalk between subbands is a major drawback and, as seen in figure 4.9, the
use of ideal rectangular low-pass filters in the OFDM-UWB transmitter and receiver improves the system
performance. The use of rectangular LP filter in the OFDM-UWB transmitter and receiver reduces by
approximately 0.5 dB the required OSNR value when comparing with the adoption of 6th order Bessel
LP filters.
(a) subbands 1,2 (b) subbands 1,2,3
Figure 4.9.: Required OSNR as a function of the SSMF length for the optimal current levels for subbands1,2 and 1,2,3 comparing the Bessel and rectangular filters use at the OFDM-UWB transmitter andreceiver.
46
4.4.2. Subbands used: 1,2,3,4
Back-to-back configuration
The contour of the BER for OSNR = 15 dB 2 as a function of the bias and peak-to-peak current is
represented in figure 4.10. In terms of the required OSNR, a bias current of 18 mA and a peak-to-peak
current of 400 mA constitute the optimal current levels for the simultaneous transmission of subbands 1,
2, 3 and 4.
Figure 4.10.: BER contour for an OSNR = 15 dB as a function of the bias and peak-to-peak current in theback-to-back configuration for subband group 1,2,3,4.
The simultaneous transmission of 4 UWB subbands show a lower performance in terms of required
OSNR (12.39 dB) comparing with the transmission of 1,2,3 (9.13 dB) due to more interference be-
tween subbands and the use of higher-frequency subbands which, as demonstrated in the previous sub-
sections, experiences a stronger distortion and have the corresponding Psb lower.
Table 4.11 presents the EVM value of the OFDM-UWB signal at several points of the system. Comparing
with the single subband transmission case (Table 4.1) the EVM at the OFDM-UWB receiver increases
around 1.5 dB in each subband. It can be noticed, also for the multi-subband transmission of 1,2,3,4,
that the distortion inflicted by the DML is substantially stronger for higher frequency subbands.
2The optimization was performed minimizing the required OSNR for BER = 10−4 but, for the current values considered, thecontour of the required OSNR did not fit the limits of the figure. Nevertheless, the contour of the BER shows a similarbehaviour.
47
Table 4.11.: Noiseless EVM of the OFDM-UWB signal in several points of the system for the multisubband transmission 1,2,3,4 in the back-to-back configuration.
Subband 1 2 3 4
EVM [dB] @ clipper input -39.5 -39.5 -39.8 -39.8
EVM [dB] @ DML input -22.5 19.8 -19.2 -22.1
EVM [dB] @ OFDM-RX -18.7 -16.2 -13.9 -12.9
EVM difference between of DML input and OFDM-RX [dB] 3.8 3.6 5.3 9.2
Performance with fiber transmission
Figure 4.11 shows the required OSNR for BER=10−4 for subbands 1, 2, 3 and 4 when simultaneously
transmitted with optimal current levels as presented in table 4.12.
Subbands 3 and 4 impose the lowest performance: subband 3 experiences a higher level of interference
from subbands at the both sides of the frequency spectrum, but the higher frequency subbands have the
Psb decrease more and are submitted to a higher distortion inflicted by the DML.
For a configuration with no fiber transmission, subband 4 sets the highest required OSNR. For a transmis-
sion with an increasing fiber length, the crosstalk between subbands increase due to the chromatic dis-
persion effects, making subband 3 the one that imposes the lowest performance of the subband group.
Figure 4.11b presents the EVM value of each subband of the OFDM-UWB signal at the receiver. It can
be noticed that the EVM of subband 2 and 3 substantially increase for the 100 km fiber transmission. The
level of interference raises significantly due to the crosstalk from both sides of the frequency spectrum
which is amplified by the fiber dispersion effects, making subband 3 impose the lowest performance.
Table 4.12.: Optimal current levels for a multi-subband fiber transmission of 1,2,3,4 and the correspond-ing minimum required OSNR.
SSMF length [km] 0 20 40 60 80 100
Optimal bias current [mA] 18 18 19 19 19 19
Optimal peak-to-peak current [mA] 400 400 450 450 450 600
Required OSNR for BER=10−4 12.39 12.83 13.59 14.74 16.77 20.09
48
(a) OSNR (b) EVM
Figure 4.11.: Required OSNR and EVM as a function of the SSMF length for the optimal current levelsin transmission of subbands 1,2,3,4.
Figure 4.12 compares the performance of the subband group 1,2,3,4 transmission when the bit se-
quence that originates the QPSK symbols leads to an OFDM-UWB signal with the highest PAPR (as in
any other case analyzed before) and the lowest PAPR (17.9 dB). The OFDM-UWB signal corresponding
to the maximum PAPR shows a 0.5 dB higher OSNR, leading to a worse performance. Since there is a 3
dB PAPR difference, the distortion caused by the system to the maximum PAPR OFDM-UWB signal is
higher.
Figure 4.12.: Required OSNR as a function of the SSMF length for the optimal current levels in transmis-sion of subband groups 1,2,3,4 with maximum and minimum PAPR.
Figure 4.13 shows the performance degradation of multi-subband groups 1,2, 1,2,3 and 1,2,3,4.
As expected, due to higher interference from crosstalk and the use of higher frequency subbands, the
required OSNR for BER = 10−4 increases with the number of subbands used by the multi-subband
OFDM-UWB signal. The performance degradation of the transmission of subband group 1,2,3,4 is
49
approximately value is 5 dB higher than the transmission of 1,2,3, and 10 dB higher than the transmis-
sion of 1,2.
Figure 4.13.: OSNR as a function of the SSMF length for the optimal current levels in multi-subbandgroups 1,2, 1,2,3 and 1,2,3,4.
The performance degradation can be minimized with the use of an OSR at the DML output which reduces
the effect of the fiber dispersion to the optical signal. In section 4.5, the OSR filter is presented and its
performance analyzed.
4.5. Optical spectrum reshaper filter optimization
The electrical-to-optical conversion by the DML of a current signal occurs simultaneously in intensity
and frequency. The output spectrum of the laser becomes wider due to frequency chirp and the optical
signal is more sensible to the fiber dispersion, limiting the performance of the system. Figure 4.14
shows an example of the frequency shift after the optical conversion of an OFDM-UWB current signal
transmitting the first four subbands. The OFDM-UWB current signal that modulates the DML is shown
in figure 4.15.
50
Figure 4.14.: Frequency chirp of the DML output when modulated by an OFDM-UWB signal using sub-band group 1,2,3,4.
Figure 4.15.: OFDM-UWB current signal of three consecutive OFDM-UWB symbols for subband group1,2,3,4.
As mentioned in chapter 3, with a tuned optical spectrum reshaper filter (OSR) placed after the DML,
bandwidth and chirp reduction are achieved after the DML modulation. The optical signal obtained at
the OSR output is more resilient to inter-symbol interference caused by the dispersion effects associated
with the transmission through the fiber.
Since the optimization of the OSR parameters is an extensive process, the optimization was performed
for the transmission of multi-subband OFDM-UWB signal using subbands 1,2,3,4 in a configuration
with fiber transmission of length 80 and 100 km. The optimization process accomplished for the levels of
current around the optimal values obtained in subsection 4.4.2 and are simultaneous for both bandwidth
νg and central frequency νc.
Figures 4.16 and 4.17 compare the non-managed chirp approach with the introduction of the OSR filter
at the DML output, showing the OSNR for BER = 10−4 values as a function of the peak-to-peak current
51
with I = 18 mA and I = 19 mA.
In terms of the required OSNR for BER = 10−4, there is a performance improvement with the use of the
OSR: a 3 dB decrease in the required OSNR for BER = 10−4 value for the fiber transmission of 80 km
and a 5.8 dB decrease for fiber transmission of 100 km - the optimal OSR parameters for 80 and 100 km
of fiber transmission are presented in table 4.13 3.
(a) I = 18 mA (b) I = 19 mA
Figure 4.16.: Required OSNR for BER = 10−4 as a function of the peak-to-peak current in 80 km fibertransmission with an optimized OSR.
(a) I = 18 mA (b) I = 19 mA
Figure 4.17.: Required OSNR for BER = 10−4 as a function of the peak-to-peak current in 100 km fibertransmission with an optimized OSR.
3The results obtained for the OSR optimization shown in figure 4.17 suggest a lower required OSNR for a decreasing peak-to-peak current, however, the required OSNR obtained was unexpectedly high and did not fit the limit of the figure.
52
Table 4.13.: OSR filter optimization results.
SSMF length [km] 80 100
Optimal required OSNR for BER = 10−4 [dB] 13.70 14.30
Optimal peak-to-peak current [mA] 550 500
Optimal mean current [mA] 18 18
Optimal OSR bandwidth [GHz] 5 4.6
Optimal OSR central frequency [GHz] 5 4
Figures 4.18a, 4.18b, 4.18c and 4.18d show the PSD of the OFDM-UWB signal at different points of the
system represented in figure 4.1.
Figure 4.18a shows the PSD at the DML input. It is observed power outside the subband group caused
by intermodulation products coming from the clipping operation as already discussed in section 3.4. In
figure 4.18b, it is presented the PSD of the OFDM-UWB signal after the DML modulation. It can be
noticed that the DML introduces distortion to OFDM-UWB signal: each subband shows different power
levels and there is approximately 15 dB increase of the out-of-subband power in the close proximity of
the UWB subband group borders.
The use of the OSR filter decreases the power outside the subband group as shown in figure 4.18c. It
reduces by approximately 10 dB the DC component. Thus, the mean power of the OFDM-UWB signal
decreases making the power associated with the UWB subband group higher relatively to the mean.
Thus, the interference from power outside the subband group is lower leading to a better performance of
the system.
At the OFDM-UWB receiver, the PSD of the signal shows an approximately 30 dB loss when compar-
ing to the signal at the fiber input as seen in figure 4.18d. This power decrease comes from the fiber
transmission and the optical-to-electrical conversion performed by the PIN.
53
(a) DML input (b) DML output
(c) OSR output (d) OFDM-UWB RX
Figure 4.18.: PSD of the OFDM-UWB signal in several points of the network.
In conclusion, the introduction of the OSR filter leads to an improvement that reaches 6 dB in terms
of the required OSNR when compared to the transmission using only the DML for the 100 km fiber
transmission. The chirp-managed method reduces the chirp generated by the DML modulation and
raises the subband group power level relative to the unwanted power outside the subband group.
4.6. Conclusion
In this chapter, the transmission of OFDM-UWB signals in an optical fiber was optimized: the bias and
peak-to-peak currents that minimize the required OSNR for BER = 10−4 were calculated.
The optimal bias current for both single and multi-subband transmission, originated a half-clipped OFDM-
UWB signal which shows reduced distortion although half of the amplitude of the signal was removed.
54
The optimal peak-to-peak current sets the balance between increasing the distortion introduced by the
clipper and raising the power level of the UWB subband. Its value increases with the number of subbands
used due to the higher PAPR shown in OFDM-UWB signals using more subbands.
It was concluded that the performance degradation is severely damaged by: the distortion inflicted by the
laser, which is more intense for higher frequency subbands; the larger decrease of Psb also for higher fre-
quency subbands which is enhanced by the fiber transmission; and the higher interference from crosstalk
between subbands in multi-subband transmission which is raised by the chromatic dispersion of the
fiber.
A required OSNR of 20 dB for BER = 10−4 was obtained when transmitting subbands 1, 2, 3 and
4 simultaneously through the configuration with 100 km of fiber. This value is 5 dB higher than the
transmission of 1,2,3, and 10 dB higher than the transmission of 1,2
The chirp-managed laser technology was introduced and an improvement of 5.8 dB in the required OSNR
for BER = 10−4 was achieved in the simultaneous transmission of subbands 1, 2, 3 and 4, through the
configuration with 100 km of fiber. The method reduces the chirp generated by the DML modulation
and raises the subband group power level relative to the unwanted power outside the subband group.
55
5. Transmission performance of the multi-subband
OFDM-UWB signal in the optimized FTTH
network
5.1. Introduction
The analysis of transmission performance of OFDM-UWB signals (detailed in chapter 2) along a FTTH
network (presented in chapter 3) is accomplished in this chapter.
The chirp-managed-based approach for the optical transmitter is compared to the use of DML. The
performance improvement is quantified from the maximum transmission distance for a required level of
BER.
The outcome of chapter 4, namely the optimal current levels that modulate the DML and the optimal
bandwidth and central frequency that characterize the OSR filter, are adopted to the transmission of the
multi-subband OFDM-UWB signal using subbands 1, 2, 3 and 4.
The analysis is performed for the power levels of 0, 3 and 5 dBm, at the central node output (PCN), and a
number of subscribers (N) of 16, 32 and 64.
5.2. Results and discussion
Figures 5.1, 5.2 and 5.3 show the BER as a function of the SSMF length of the FTTH network transmit-
ting an OFDM-UWB signal using subband group 1,2,3,4 considering N = 64, 32 and 16, respectively.
As expected, the power level of the received signal substantially influences the performance of the sys-
tem. Figures 5.1, 5.2 and 5.3 show that the BER increases with the number of subscribers due to a higher
loss inflicted by the N-splitter. For higher PCN , the BER of the system decreases: considering N = 16 in
the CML-based transmission of 80 km, the BER decreases approximately 7 orders of magnitude when
PCN increases from 0 to 3 dBm, and 10 orders of magnitude when PCN increases from 3 to 5 dBm. This
56
result is expected because an OFDM-UWB signal experiences the same level of distortion independently
of its mean power and, as demonstrated in section 5.2.3, the level of introduced noise does not increase
significantly with the mean power of the OFDM-UWB signal.
As concluded in Chapter 4, the transmission using the chirp-managed laser improves the performance
of the system compared to the use of the DML. A 6 dB improvement in the required OSNR for BER =
10−4 was obtained for the transmission of an OFDM-UWB signal using subband group 1,2,3,4 with
a fiber length of 100 km. This technology reduces the frequency shift generated by the DML and raises
the subband power level relative to the unwanted power outside the subband group.
Considering the FTTH configuration with 64 subscribers, the chirp-managed-based transmission does not
ensure the minimum quality of BER = 10−3 for PCN = 0 dBm, but this minimum quality is achieved for
the maximum distance of 81 and 90 km for the fiber transmission with PCN of 3 and 5 dBm, respectively.
As observed in figure 5.1, the DML-based transmission only assures a BER not exceeding 10−3 for
the maximum fiber length of 83 km with PCN = 5 dBm. Thus, the CML-based transmission improves
the maximum reach of the network that ensures a minimum quality of BER = 10−3, compared to the
DML-based transmission, by approximately 1 and 7 km, respectively for PCN of 3 and 5 dBm.
For N = 32, as considered in figure 5.2, the transmission using the DML only assures the minimum
quality of BER = 10−4 for the maximum distance of 83 km for PCN of 3 dBm and 89 km for PCN of 5
dBm. The transmission using the CML ensuring the minimum quality of BER = 10−4 is accomplished
for a maximum distance of 90 km for PCN of 3 dBm, and 100 km for PCN of 5 dBm. The CML-based
transmission with PCN of 0 dBm does not ensure the minimum quality of BER = 10−4. An improvement
of 7 and 11 km is achieved for the cases with a PCN of 3 and 5 dBm.
As it can be noticed in figure 5.3, for N = 16 the transmission using the DML only assures the minimum
quality of BER = 10−9 for a maximum transmission length of 80 km with PCN of 3 dBm, and 86 km with
PCN of 5 dBm. The CML based transmission ensures the minimum required quality for the transmission
length of 87 and 96 km with PCN of 3 and 5 dBm, respectively. For both DML and CML-based trans-
mission the minimum quality not exceeding BER = 10−9 is not achieved with PCN of 0 dBm. Thus, an
improvement of 7 and 10 km is achieved for PCN of 3 and 5 dBm.
57
Figure 5.1.: BER as a function of the SSMF length of the FTTH network considering N=64.
Figure 5.2.: BER as a function of the SSMF length of the FTTH network considering N=32.
58
Figure 5.3.: BER as a function of the SSMF length of the FTTH network considering N=16.
5.2.1. Assessment of the distortion inflicted by the OFDM-UWB FTTH network
Although the same power level is assigned to both CML and DML-based approaches, the chirp-controlled
optical OFDM-UWB signal is more resilient to the fiber transmission. Figure 5.4 shows the noiseless
EVM of the OFDM-UWB signal at the receiver as a function of the SSMF length. It can be observed that
the chromatic dispersion associated with the fiber transmission has a stronger effect on the DML-based
OFDM-UWB signal: there is a higher increase of the EVM value compared to the EVM of the CML-
based transmission. The subbands in the middle of the subband group have an EVM increase mainly
due to a larger crosstalk coming from subbands at both sides of the spectrum. The distortion coming
from crosstalk is enhanced by the higher sensitivity of the non-chirp-controlled OFDM-UWB signal to
the fiber dispersion.
Figures 5.5a and 5.5b show the difference between the received constellation of a chirp-managed and a
non-chirp-managed OFDM-UWB signal. As expected, the demodulated symbols of the chirp-managed
OFDM-UWB signal are less spread than the demodulated symbols of the OFDM-UWB signal without
controlled chirp.
As in Chapter 4, subband 3 and 4 impose the lowest performance: subband 3 experiences a higher level
of interference from subbands at the both sides of the frequency spectrum, but higher frequency subbands
have their Psb decrease more due to the fiber dispersion and are submitted to a higher distortion by the
DML (as Chapter 4). For the DML-based transmission, the subband with worst performance varies
between 3 and 4. The higher level of crosstalk caused by a longer fiber transmission makes the subband
with worst performance change from 4 to 3.
59
In the CML-based transmission, subband 4 imposes the highest distortion. As observed in figure 5.4a,
the fiber dispersion effect does not influence the EVM value as much as in the transmission using the
DML. A chirp-controlled OFDM-UWB signal exhibits a higher resilience to the chromatic dispersion
of the fiber. Chapter 4 demonstrated that higher frequency subbands experience more distortion by the
DML and, in this case, it determines the subband with worst performance.
(a) CML-based transmission (b) DML-based transmission
Figure 5.4.: EVM value at the OFDM-UWB receiver for the CML and DML-based transmission.
(a) CML-based transmission (b) DML-based transmission
Figure 5.5.: Constellation of the received symbols from subbands 1, 2, 3 and 4 of CML and DML-based100 km transmission.
5.2.2. Influence of the optical noise generated by the EDFA at the central node
As explained in chapter 3, the role of the EDFA prior to the fiber transmission is to impose a defined
mean optical power in order to compensate for the losses inflicted by the OSR. Along with the optical
60
amplification, the EDFA increases the level of noise power.
In this subsection, the influence of the optical noise generated by the post-amplifier at the central node is
assessed in the OSNR value at the local exchange.
The noise originated by the EDFA depends on its gain as given in (3.4). The gain of the EDFA at the
central node is influenced by the mean optical power imposed at the central node output (PCN). A higher
level of mean optical power imposed at the central node output increases the gain of the first EDFA
raising the ASE noise power at its output.
Table 5.1 shows the OSNR calculated at the splitter input neglecting the noise generated at the central
node, and the OSNR considering both the noise generated from the EDFA at the local exchange and the
noise coming from the EDFA at the central node 1. As expected, the influence of the noise generated
at the central node increases with PCN. However, the OSNR difference does not exceed 0.1 dB for
OSNR values near 30 dB, which means that the noise generated by the first EDFA does not influence the
performance of the CML-based system.
Table 5.1.: OSNR calculated at the DEMUX input in the CML-based transmission.
SSMF length after the splitter [km] 80 85 90 95 100
OSNR neglecting the noise from the 1st EDFA
for PCN = 0 dBm [dB]21.67 21.67 21.67 21.67 21.67
OSNR accounting for the noise from the 1st
EDFA for PCN = 0 dBm [dB]21.67 21.67 21.67 21.67 21.67
OSNR neglecting the noise from the 1st EDFA
for PCN = 3 dBm [dB]24.67 24.67 24.67 24.67 24.67
OSNR accounting for the noise from the 1st
EDFA for PCN = 3 dBm [dB]24.63 24.64 24.64 24.64 24.65
OSNR neglecting the noise from the 1st EDFA
for PCN = 5 dBm [dB]26.67 26.67 26.67 26.67 26.67
OSNR accounting for the noise from the 1st
EDFA for PCN = 5 dBm [dB]26.59 26.59 26.60 26.60 26.61
Since for the DML-based transmission there is no loss from the chirp control, the EDFA gain is lower
than in the CML-based transmission. Therefore, it can be concluded that the ASE noise of the EDFA at
1The OSNR value shown in Table 5.1 varies with the fiber length because the OSR considered for each fiber length hasdifferent characteristics.
61
the central node does not influence the OSNR at the receiver and, consequently, the overall performance
of the system.
5.2.3. Influence of optical and electrical noise
As explained in Chapter 3, the noise in the FTTH network is originated by the EDFAs at the central node
and local exchange, and by the electrical circuitry at the premises of the user.
It is considered that the optical noise is dominant, when the variance (that affects each OFDM-UWB
subcarrier) of the optical noise is one order of magnitude higher than the variance of electrical noise.
The electrical noise is dominant, if the variance of electrical noise is one order of magnitude higher than
the variance of optical noise.
As presented in Chapter 3, the power spectral density of the electrical noise is constant and its value is
given by expression (3.12). The power spectral density of the noise generated by the EDFA is given by
(3.4), and it is influenced by the number N of subscribers and the length of the second section of fiber.
An increasing number of subscribers or a longer fiber transmission raises the level of loss inflicted by the
FTTH, reducing the PSD of the optical noise incident on the PIN. A lower value of the power spectrum
density of the optical noise reduces its influence on the performance of the system.
Considering N = 16, the variances of electrical and optical noise have the same order of magnitude
in most of the configurations. Due to a higher loss from a longer fiber transmission noise generated
electrically prevails over the optical noise.
For N = 32 and 64, the electrical noise is dominant for all configurations. The PSD of noise generated by
the EDFA is significantly reduced by the 32 and 64-splitter, making the noise generated by the electrical
circuitry have a dominant role in determining the performance of the system.
5.3. Conclusion
The performance of transmission of OFDM-UWB signals along the FTTH network was analyzed in this
chapter. The chirp-managed-based approach for the optical transmitter was compared to the use of the
DML, reaching an improvement of 10 km of the maximum transmission distance for a minimum quality
of BER = 10−9 in the configuration with 16 subscribers. It is concluded that the chirp-controlled OFDM-
UWB optical signal is more resilient to the fiber dispersion, showing a noiseless EVM that is not much
influenced by fiber transmission compared to the EVM of the DML-based transmission.
62
The impact of the noise originated from the EDFA at the central node was assessed. It was concluded
that noise originated by ASE of the first EDFA is neglectable since the maximum difference between
the OSNR of the OFDM-UWB optical signal at the splitter input taking into account the noise from the
central node and neglecting it, does not exceed 0.1 dB for the power levels at the central node output of
0, 3 and 5 dBm.
The weight on the FTTH performance of optical generated noise and electrical generated noise was
evaluated. It was concluded that, for most cases, the electrical noise has a dominant role. The high loss
inflicted by the FTTH network along with the increase of the number of subscribers and the longer second
section of fiber significantly reduces the noise generated by the EDFA, making the electrical circuit noise
have a dominant role in determining the performance of the system.
63
6. Final conclusion
6.1. Conclusion
The objective of this dissertation was the analysis and performance optimization of the transmission of
OFDM-UWB radio signals in FTTH networks using the chirp-managed laser.
In chapter 2, OFDM-UWB radio signals were introduced and described. An analitycal approach was fol-
lowed for the definition of this type of signals and it was presented the IFFT/FFT-based implementation
of the OFDM-UWB transmitter and receiver which significantly reduces its complexity. It was discussed
that the original OFDM-UWB signals are not suited for the optical transmission using intensity mod-
ulation and that the mean and amplitude of the signal should be adequately modified. The time and
frequency analysis of OFDM-UWB radio signals demonstrated that the 6th order Bessel low-pass filters
used in the OFDM-TX and OFDM-RX reduce the aliasing coming from the digital-to-analog conversion
although introduces distortion to the subband caused by the filter not having a totally flat passband. The
high PAPR was mentioned as one of the major drawbacks of the OFDM-UWB radio signals. Higher
PAPR values were obtained for OFDM-UWB signals originated by longer bit sequences and using more
subbands. A PAPR of 21.3 dB was reached for an OFDM-UWB signal using simultaneously subbands
1, 2, 3 and 4.
In chapter 3, the FTTH architecture was presented and each of its components was described: the general
behavior of the directly modulated laser was detailed and its main drawbacks were identified: high
frequency chirp and limited input region of operation. The high frequency chirp effect can be reduced
by filtering the DML output by an optical spectrum reshaping filter. This filter performs a frequency to
intensity conversion, increasing the tolerance of the optical signal to the fiber dispersion effect. The main
characteristics of the optical spectrum reshaper were discussed being chosen a 1storder Gaussian filter to
control the chirp generated by the DML modulation.
There were introduced two performance methods: noiseless error vector magnitude and semi-analytical
Gaussian approach (SAGA). The first one refers to a measure of the distortion inflicted by the FTTH net-
64
work on the transmitting signal. The SAGA method gives the bit error ratio and considers the distortion
inflicted by the system and the optical and electrical noise added by the amplifiers.
The effect of the non-linear operation of clipping in the OFDM-UWB signal was analysed and it was
concluded that a half-clipped OFDM-UWB signal shows a reduced distortion and it represents a solution
for the limited region of modulation of the DML and the high peak-to-average power ratio typical of
OFDM signals. However, it was demonstrated that clipping originates out-of-subband noise which can
significantly affect the introduction of higher frequency subbands.
In chapter 4, the transmission of OFDM-UWB signals in a DML-based system was optimized: the bias
and peak-to-peak currents that minimize the required OSNR for BER = 10−4 were calculated. The
optimization was performed for both single and multi-subband transmission of OFDM-UWB signals
in the configuration with fiber transmission from 0 to 100 km. For all cases, the optimal bias current
originated a half-clipped OFDM-UWB signal. The half-clipped OFDM-UWB signal shows reduced
distortion although half its amplitude was removed. The optimal peak-to-peak current sets the balance
between increasing the distortion introduced by the clipper and raising the power level of the UWB
subband. The optimal current amplitude increases with the number of subbands used due to the higher
PAPR of OFDM-UWB signals using more subbands. The performance degradation was analyzed and
it was demonstrated that it is imposed by the distortion inflicted by the laser, which is more intense for
higher frequency subbands. The larger decrease of Psb for higher frequency subbands significantly limits
the performance of the system and it was shown that this effect is augmented by the fiber transmission.
The higher interference from crosstalk between subbands in multi-subband transmission substantially
degrades the performance of the system and it is also augmented by the chromatic dispersion of the fiber.
A required OSNR of 20 dB for BER = 10−4 was obtained when transmitting subbands 1, 2, 3 and 4
simultaneously through a configuration with 100 km of fiber transmission. This value is 5 dB higher
than the transmission of subbands 1, 2 and 3, and 10 dB higher than the transmission of subbands 1 and
2.
The chirp-managed laser was optimized and an improvement of 6 dB in the required OSNR for BER =
10−4 was achieved when subbands 1, 2, 3 and 4 are simultaneously transmitted along 100 km of fiber.
The method reduces the chirp generated by the DML modulation and raises the subband group power
level relative to the unwanted power outside the subband group, making the optical OFDM-UWB signal
more resilient to the chromatic dispersion of the fiber.
The transmission performance of OFDM-UWB signals along the FTTH network using chirp-managed
lasers was investigated in chapter 5. The chirp-managed-based approach for the optical transmitter was
compared to the use of the DML, reaching an improvement of 10 km of the maximum transmission
65
distance for a minimum quality of BER = 10−9. It was concluded that the chirp-controlled OFDM-UWB
optical signal is more tolerant to the fiber dispersion, showing a noiseless EVM that is approximately
constant with the fiber transmission compared to the EVM of the DML-based transmission.
The impact of the noise originated from the EDFA at the central node was assessed. It was concluded
that noise originated by ASE of the first EDFA can be neglected since the maximum difference between
the OSNR of the OFDM-UWB optical signal at the splitter input taking into account the noise from the
central node and neglecting it, does not exceed 0.1 dB.
The weight on the FTTH performance of optical generated noise and electrical generated noise was
evaluated. It was concluded that, for most cases, the electrical noise has a dominant role. The high
loss inflicted by the FTTH network along with the increase of the number of subscribers and the longer
second section of fiber significantly reduces the noise generated by the EDFA, making the electrical noise
have a dominant role in determining the performance of the system.
6.2. Suggestions for future work
Following the conclusions drawn above, some topics of work are suggested as interesting to investigate
in order to complement or continue the work accomplished in this dissertation:
• optimization of the equalizer and analysis of its impact on the performance of the CML-based
transmission of OFDM-UWB transmission in FTTH networks;
• experimental demonstration of the CML-based transmission of OFDM-UWB transmission in FTTH
networks;
• theoretical analysis of the half-clipped OFDM-UWB signals and its transmission in a DML-based
optical system;
• analysis of the transmission along CML-based FTTH networks of OFDM-UWB signals using
other modulation formats (e.g. 64-QAM);
• analysis of the transmission in CML-based FTTH networks of IR-UWB signals.
66
A. Directly-modulated laser
A.1. Introduction
The description and analysis of the directly-modulated laser used in this work is performed in this chapter.
The analytical model of laser is detailed, being derived and analyzed its static and dynamic response.
A.2. Description of the directly modulated laser
The DML used in this dissertation is a Multiple Quantum Wells from Ortel Corporation and it is operating
at the wavelength of 1550.2 nm. Its behaviour characteristics are well described by the rate equations
that govern the interaction of photons and electrons inside the laser’s active region [31].
The rate equations take the form of (A.1), (A.2) and (A.3), where N is the carrier density, S is the photon
density and ∆ν is the instantaneous frequency shift, also designated as frequency chirp.
dNdt
=ηiIA
qVact−R(N)−g(N,S)S (A.1)
dSdt
= Γg(N,S)S− Sτp
+RS(N) (A.2)
∆ν =1
2π
dφ
dt=
αc
2Γg0(N−Nbias) (A.3)
In (A.1), ηi represents the internal quantum efficiency, IA the current applied to the active region, q is the
electric charge of an electron and Vact is the volume of the active layer. R(N) is a function that defines
the total recombination rate,
R(N) = AnrN +BN2 +CN3, (A.4)
67
where Anr is the nonradiative recombination rate coefficient, B, the radiative recombination rate coeffi-
cient, and C is the Auger recombination coefficient. g(N,S) is designated as optical gain and it is given
by,
g(N,S) =g(N)
1+ εS, (A.5)
in which ε is the nonlinear gain compression factor and g(N) is expressed as,
g(N) = gc ln(
R(N)
R(N0)
)= gc ln
(AnrN +BN2 +CN3
AnrN0 +BN20 +CN3
0
), (A.6)
where gc is the gain parameter and N0 is carrier density at transparency.
In (A.2), τp represents the photon lifetime and RS(N) the rate of spontaneous emission which is related
to the carrier density N by,
RS(N) = ΓβBN2, (A.7)
where β is the fraction of spontaneous emission noise coupled into the lasing mode.
Finally, in (A.3), αc is the linewidth enhancement factor, Nbias is the carrier density for the polarization
current Ibias and g0 is the differential gain coefficient.
Table A.1 shows the values of the laser intrinsic parameters.
68
Table A.1.: Ortel MQW laser parameter values.
Parameters Value
Carrier Density at Transparency, N0 1.7× 1024 m−3
Volume of the Active Region, Vact 29.2× 10−18 m3
Confinement Factor, Γ 0.13
Internal Quantum Efficiency, ηi 1
Differential Quantum Efficiency, η 0.139
Gain Parameter, gc 10.6× 1012 s−1
Gain Compression Factor, ε 5× 10−23 m3
Photon Lifetime, τp 0.98 ps
Linewidth Enhancement Factor, αc 2.8
Spontanerous Emission Factor, β 2.5× 10−5
Nonradiative Recombination Rate Coefficient, Anr 5× 107 s−1
Radiative Recombination Rate Coefficient, B 4.6× 10−16 m3s−1
Auger Recombination Coefficient, C 1.08× 10−41 m6 s−1
The optical power at the laser output is related with the photon density through the expression
P(t) =Vactηhυ
ΓτpηiS(t), (A.8)
where h is Planck’s constant, η is the differential quantum efficiency and υ is the optical frequency.
The DML model also considers the effects originated from the assemblage of the laser. These effects are
simulated by a first order low-pass filter with a time constant of τ=26 ps (as deduced in [37]):
H( jω) =1
1+ jωτ(A.9)
The bandwidth at -3 dB of filter is 6.1 GHz and its frequency characteristics are shown in figure A.1.
69
(a) Magnitude Response (b) Phase Response (c) Delay Response
Figure A.1.: Frequency characteristics of the assemblage effects on the DML.
The static response of the DML is shown in figure A.2. It shows the output power of the DML when a
constant current is modulating the laser.
The analytical results were obtained by solving the rate equations in steady-state mode (d/dt = 0) using
the method of Newton (the analytical solution is derived in [37]). The simulated results come from
solving the rate equations for a constant input current and getting the output power after the transient
period passed.
It is observed that for a current below 14.34 mA the output power is approximately 0 because there is
only spontaneous emission of photons. Above this threshold, there is stimulated emission of photons
and, therefore, the output power increases and it shows a linear dependence with the input current. This
linear region is limited until 100 mA, which is the maximum input current value before the laser reaches
a multi-mode emission. Thus, the current levels should only vary inside the forementioned region.
Figure A.2.: Static response of the MQW laser.
The dynamic response of the laser to a small current signal modulation is shown in figure A.3. When
this input current, defined by
70
I(t) = Ibias + i(ω)e jωt and i(ω)<< Ibias, (A.10)
modulates the laser, the output shows
• Intensity Modulation (figure A.3a): defined as the relation between the optical output power and
the input current (responses derived by [37]),
HIM =pout(ω)
i(ω). (A.11)
• Frequency Modulation (figure A.3b): defined as the relation between the frequency shift of the
output signal, or frequency chirp, and the input current (responses derived by [37]),
HFM =∆ν(ω)
i(ω). (A.12)
Figure A.3a shows that the IM response has a larger bandwidth for a higher bias current. From a Ibias =
20 mA to Ibias = 60 mA, the bandwidth at -3 dB increases approximately from 10 GHz to 25 GHz. The
FM response (figure A.3b) exhibits higher bandwidths for the same bias current, for example, for Ibias =
60 mA, the bandwidth at -3 dB is 30 GHz higher than the corresponding IM, reaching 55 GHz.
This parallel modulation becomes a drawback when a IM signal is sent through the fiber: a signal with
a wide spectrum due to FM is more sensitive to the dispersion of the fiber, limiting the transmission
distance without a compensator.
(a) IM response of the DML (b) FM response of the DML
Figure A.3.: Dynamic response of the DML when modulated by small current signals.
71
A.3. Conclusion
In this appendix, the directly-modulated laser rate equations were presented and detailed. The static
response of the laser was derived and it shows a linear region of operation between 14.34 mA and 100
mA. The dynamic response of the laser exhibits a simultaneous intensity and frequency modulation
which broadens the optical signal spectrum making it more vulnerable to the fiber dispersion effect.
72
B. Optimization results
In this appendix, it is shown the optimization results accomplished for single UWB subbands 1, 2, 3 and
4, and subband groups 1,2, 1,2,3 and 1,2,3,4 with fiber transmission of 0, 20, 40, 60, 80 and 100
km. The results presented follow the description performed in Chapter 4.
The optimization process is demonstrated in terms of the contour of the required OSNR for BER = 10−4
as a function of the mean and peak-to-peak current. For the subband group 1,2,3,4 it is presented the
contour of the BER for an OSNR OF 15 dB as a function of the current levels.
Since the optimization process is an extensive process, for most of the cases the range in the bias and
peak-to-peak current is limited to the bottom of the linear operating region of the laser, leading to half-
clipped OFDM-UWB signals.
73
(a) 0 km (b) 20 km
(c) 40 km (d) 60 km
(e) 80 km (f) 100 km
Figure B.1.: Contour of the representation of the required OSNR as a function of the current level forsubband 1 considering a fiber transmission from 0 to 100 km.
74
(a) 0 km (b) 20 km
(c) 40 km (d) 60 km
(e) 80 km (f) 100 km
Figure B.2.: Contour of the representation of the required OSNR as a function of the current level forsubband 2 considering a fiber transmission from 0 to 100 km.
75
(a) 0 km (b) 20 km
(c) 40 km (d) 60 km
(e) 80 km (f) 100 km
Figure B.3.: Contour of the representation of the required OSNR as a function of the current level forsubband 3 considering a fiber transmission from 0 to 100 km.
76
(a) 0 km (b) 20 km
(c) 40 km (d) 60 km
(e) 80 km (f) 100 km
Figure B.4.: Contour of the representation of the required OSNR as a function of the current level forsubband 4 considering a fiber transmission from 0 to 100 km.
77
(a) 0 km (b) 20 km
(c) 40 km (d) 60 km
(e) 80 km (f) 100 km
Figure B.5.: Contour of the representation of the required OSNR as a function of the current level forsubbands 1 and 2 considering a fiber transmission from 0 to 100 km.
78
(a) 0 km (b) 20 km
(c) 40 km (d) 60 km
(e) 80 km (f) 100 km
Figure B.6.: Contour of the representation of the required OSNR as a function of the current level forsubbands 1, 2 and 3 considering a fiber transmission from 0 to 100 km.
79
(a) 0 km (b) 20 km
(c) 40 km (d) 60 km
(e) 80 km
Figure B.7.: Contour of the representation of the required OSNR as a function of the current level forsubbands 1, 2, 3 and considering a fiber transmission from 0 to 80 km.
80
(a) 0 km (b) 20 km
(c) 40 km (d) 60 km
(e) 80 km (f) 100 km
Figure B.8.: BER contour for an OSNR = 15 dB as a function of the current levels for subbands 1, 2, 3and 4 considering a fiber transmission from 0 to 100 km.
81
C. References
[1] M. G. Benedetto, G. Giancola, Understanding Ultra Wide Band Radio Fundamentals; Prentice Hall;
New Jersey, USA; 2004.
[2] Title 47, Section 15 of the Code of Federal Regulations SubPart F: Ultra-wideband, Federal Com-
munications Commission, 2002.
[3] R. Kohno, State of Arts in Ultra Wideband (UWB) Wireless Technology and Global Harmonization,
in Proc. 34th Eur. Microwave Conf., The Netherlands, pp. 1093-1099, 2004.
[4] IEEE 802.15 WPAN High Rate Alternative PHY Task Group 3a (TG3a).
www.ieee802.org/15/pub/TG3a.html.
[5] T. Valente, A. V. T. Cartaxo, Distribution of IR-UWB Signals With PAM Modulation in Long-Reach
FTTH Networks, Journal of Optical Communications Networks, Vol. 2, No. 11, pp. 892-900, Novem-
ber 2010.
[6] G. Heidari, WiMedia UWB: Technology of Choice for Wireless USB and Bluetooth, 1st ed. Chich-
ester, U.K.: Wiley, ch. 1, 2008.
[7] ECMA-368, High rate ultra wideband PHY and MAC Standard 2nd ed. Geneva, ECMA Interna-
cional, 2007.
[8] R. Llorente, T. Alves, M. Morant, M. Beltran, J. Perez, A. Cartaxo, and J. Marti, Optical Distribu-
tion of OFDM and Impulse-Radio UWB in FTTH Networks; Optical Fiber Communication/National
Fiber Optic Engineers Conference; Pages: 1 - 3; 2008.
[9] R. Llorente, T. Alves, M. Morant, M. Beltran, J. Perez, A. Cartaxo, and J. Marti, Ultra-Wideband
Radio Signals Distribution in FTTH Networks; IEEE Photon. Technol. Lett., vol. 20, no. 11, pp.
945-947; June 2008.
[10] P.D. Townsend, G. Talli, C. W. Chow, E. M. MacHale, C. Anthony, R. Davey, T. De Ridder, X. Z.
Qiu, P. Ossieur, H. G. Krimmel, D. W. Smith, I. Lealman, A. Poustie, S. Randel, and H. Rohde, Long
82
Reach Passive Optical Networks, Laser and Electro-Optics Society, The 20th Annual Meeting of the
IEEE, pp. 868-869, 2007.
[11] Photonic Integrated Extended Metro and Access Network, available at http://www.istpieman.org
[12] T. Alves, M. Morant, R. Llorente, A. Cartaxo, J. Marti, Experimental Demonstration of 1.56 Gbit/s
OFDM-UWB Distribution over 100 km of Standard-Fiber in FTTH Networks, 2008 Opto-Electronics
and Communications Conference (OECC) and the Australian Conference on Optical Fibre Technol-
ogy (ACOFT), Sydney, Australia, article TuB-3, 2008.
[13] M. Morant, T. Alves, R. Llorente, A. Cartaxo, J. Marti, Experimental Comparison of Transmis-
sion Performance of Multi-Channel OFDM-UWB Signals on FTTH Networks, IEEE/OSA Journal
of Lightwave Technology, vol. 27, no 10, pp. 1408-1414, 2009.
[14] R. P. Davey, P. Healey, I. Hope, P. Watkinson, and D. B. Payne, DWDM reach extension of a GPON
to 135 km, Journal of Lightwave Technology, vol. 24, no 1, pp. 29-31, 2006.
[15] M. Yee, V. Pham, Y. Guo, L. Ong, and B. Luo, Performance Evaluation of MB-OFDM Ultra-
Wideband Signals over Single Mode Fiber, in Proc. Int. Conf. Ultra Wideband (ICUWB), pp. 674-
677, 2007.
[16] Y. Guo, V. Pham, M. Yee, L. Ong, and B. Luo, Performance Study of MB-OFDM Ultra-Wideband
Signals over Multimode Fiber, in Proc. Int. Conf. Ultra Wideband (ICUWB), pp. 429-431, 2007.
[17] Y. Ben-Ezra, M. Ran, E. Borohovich, A. Leibobovich, M. Thakur, R. Llorente, and S. Walker,
WiMedia-Defined, Ultra-Wideband Radio Transmission over Optical Fiber, in Proc. Opt. Fiber Com-
mun. Conf. Exposition (OFC) and Natl Fiber Opt. Eng. Conf. (NFOEC), pp. 1-3, 2008.
[18] M. N. Sakib, B. Hraimel, X. Zhang, M. Mohamed, W. Jiang, K. Wu, and D. Shen, Impact of Op-
tical Transmission on Multiband OFDM Ultra-Wideband Wireless System With Fiber Distribution,
Journal of Lightwave Tech., vol. 27, no. 18, pp. 4112-4123, 2009.
[19] A. Pizzinat, B. Charbonnier, and M. Moignard, Analysis of Laser Induced Distortions in Ultra-
Wideband MB-OFDM over Fiber, in IEEE Lasers Electro-Optics Soc. (LEOS) Annu. Meeting, pp.
339-340, 2006.
[20] M. N. Sakib, B. Hraimel, X. Zhang, K. Wu, T. Liu, T. Xu, and Q. Nie, Impact of Laser Relative
Intensity Noise on a Multiband OFDM Ultrawideband Wireless Signal over Fiber System, in J. Opt.
Commun. Netw., vol. 2, no. 10, pp. 841-847, 2010.
83
[21] J. Morgado, A. V. T. Cartaxo, OFDM-UWB Signal Distribution over Long-Reach PON Using Di-
rectly Modulated Lasers, Transparent Optical Networks (ICTON) - 12th International Conference,
article Th.A2.5, pp. 1 - 4, 2010.
[22] R. Giddings, Experimental Demonstration of a Record High 11.25 Gbps Real-Time Optical OFDM
Transceiver Supporting 25 km SMF End-to-End Transmission in Simple IMDD Systems, Optics Ex-
press, vol. 18, no. 6, pp. 5541-5555, 2010.
[23] I. P. Kaminow, and T. Li, Optical Fiber Telecommunications Components IV-A, Academic Press,
chapter 12, pp. 613-625, 2002.
[24] Matsui, Y.; Mahgerefteh, Y.; Cheng, X.; Liao, C.; Fan, Z. F.; McCallion, K.; Tayebati, P. Chirp-
Managed Directly Modulated Laser (CML), IEEE Photon. Technol. Lett., vol. 18, pp. 385-387, 2006.
[25] D. Mahgereften, Y. Matsui, C. Liao, B. Johnson, D. Walker, X. Zheng, Z. F. Fan, K. McCallion,
and P. Tayebati, Error-Free 250 km Transmission in Standard Fiber using Compact 10 Gbit/s Chirp-
Managed Directly Modulated Lasers (CML) at 1550 nm, Electron. Lett., vol. 41, pp. 543-544, 2005.
[26] S. Chandrasekhar, C. R. Doerr, L. L. Buhl, D. Mahgereften, Y. Matsui, B. Johnson, C. Liao, X.
Zheng, K. McCallion, Z. Fan, and P. Tayebati, Flexible Transport at 10-Gb/s from 0 to 675 km (11500
ps/nm) using Chirp-Managed, no DCF, and a Dynamically Adjustable Dispersion Compensating
Receiver, Optical Fiber Communication Conference, vol. 6, 2005.
[27] Yu, J.; Jia, Z.; Huang, M-F.; Harris, M.; Ji, P. N.; Wang, T.; Chang, G-K. Applications of 40-Gb/s
Chirp-Managed Laser in Access and Metro Networks, Journal of Lightwave Technology, vol. 27, no.
3, pp. 253-265, 2009.
[28] Batra, A; Balakrishnan, J.; Dabak, A.; Multi-Band OFDM: a new approach for UWB; Circuits
and Systems, ISCAS ’04, Proceedings of the 2004 International Symposium; Pages: V-365 - V-368;
Vol.5; 2004.
[29] Shieh, W.; Djordjevic, I.; OFDM for Optical Communications; Academic Press; USA; 2010.
[30] Kaminow, I.; Li, T. Optical Fiber Telecommunications IV A Components, Academic Press, San
Diego, Chapter 12, 2002.
[31] Agrawal, G. P. Fiber-Optic Communication Systems, 2nd edition, Wiley-Interscience, USA, Chapter
3, 1997.
[32] JDSU 100 GHz narrowband (Gaussian) arrayed waveguide grating (AWG), Available at:
http://www.jdsu.com/product-literature/awg100n_ds_cc_ae.pdf (accessed: 23 October 2010), 2009.
84
[33] Agrawal, G. P. Fiber-Optic Communication Systems, 2nd edition, Wiley-Interscience, USA, Chapter
2, 1997.
[34] Agrawal, G. P. Fiber-Optic Communication Systems, 2nd edition, Wiley-Interscience, USA, Chapter
4, pp. 166-167, 1997.
[35] T. M. F. Alves, and A. V. T. Cartaxo, Semi-Analytical Approach for Performance Evaluation of
Direct-Detection OFDM Optical Communication Systems, Optics Express, Vol. 17, No. 21, pp.
18714-18729, 2009.
[36] J. Armstrong, B. J. C. Schmidt, D, Kalra, H. A. Suraweera, and A. J. Lowery, Performance of
Asymmetrically Clipped Optical OFDM in AWGN for an Intensity Modulated Direct Detection Sys-
tem, Global Telecommunications Conference (GLOBECOM) - IEEE, SPC07-4, pp. 1 -5, 2006.
[37] J. L. Rebola, C. A. Lourenco, and J. M. Oliveira, Optimização e Simulação de um Sistema de Co-
municação Óptica TDM Multinível com Igualação Discreta, Final year’s project at Instituto Superior
Técnico, Universidade Técnica de Lisboa, September, Chapter 4, 1999.
85