Research Collection

Doctoral Thesis

Reflective Semiconductor Optical Amplifiers (RSOAs) as

Colorless Sources in Access Networks

Author(s):
Gebrewold, Simon A.

Publication Date:
2016

Permanent Link:
https://doi.org/10.3929/ethz-a-010781780

Rights / License:
In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more
information please consult the Terms of use.

ETH Library
 DISS.-No. ETH 23541

REFLECTIVE SEMICONDUCTOR OPTICAL AMPLIFIERS

(RSOAS)
AS COLORLESS SOURCES IN ACCESS NETWORKS
A thesis submitted to attain the degree of

Doctor of Sciences of ETH Zurich

(Dr. sc. ETH Zurich)

presented by
SIMON AREGA GEBREWOLD
M.Sc., Karlsruhe Institute of Technology
born on 01.06.1987
citizen of Ethiopia

accepted on the recommendation of

Prof. Dr. Juerg Leuthold, examiner
Prof. Dr. Mario Martinelli, co-examiner

2016
To my parents, your love, kindness and prayers are always with me.

Be not wise in thine own eyes: fear the Lord, and depart from evil.
King Solomon (Proverbs 3:7)
Content
Abstract .......................................................................................................... vii
Zusammenfassung ........................................................................................... xi
Challenge and Achievements of This Work ..................................................... xv
1 Introduction ........................................................................................... 1
1.1 Telecommunication Network....................................................................................... 2
1.2 Optical Access Network ............................................................................................... 4
Fiber to the x (FTTx) .......................................................................................... 4
FTTx Technologies ............................................................................................. 6
Passive Optical Networks (PONs) ...................................................................... 7
PON Generations ............................................................................................ 11
State-of-the-Art: Colorless Upstream Transmitters for WDM-PON ................ 18
1.3 Summary .................................................................................................................... 24
2 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs) ...
............................................................................................................. 27
2.1 Background of SOAs and RSOAs ................................................................................. 28
2.2 Application of SOAs and RSOAs in Optical Access Networks ...................................... 29
Boosters, Inline and Pre-amplifiers ................................................................. 29
Directly Modulated Transmitters .................................................................... 30
Suppression of Intensity Noise ........................................................................ 31
2.3 Operation Principle of SOAs and RSOAs..................................................................... 31
Light Absorption and Emission in SOAs and RSOAs ......................................... 31
pn Junction Structures .................................................................................... 35
Lateral Confinement and Polarization Sensitivity ............................................ 40
Additional Design Challenges .......................................................................... 42
Materials ......................................................................................................... 43
Details on Material Gain and Light Propagation in SOAs and RSOAs ............... 44

3 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

............................................................................................................. 49
3.1 Reflective-SOA Fiber Cavity Laser as Directly Modulated WDM-PON Colorless
Transmitter ........................................................................................................................... 50
Introduction .................................................................................................... 50
WDM-PON Architecture .................................................................................. 51
RSOA-Fiber Laser Modeling ............................................................................. 52
Simulation Results ........................................................................................... 56
Conclusion ....................................................................................................... 64
3.2 Alternative Amplified RSOA-FCL................................................................................. 65
Introduction .................................................................................................... 66
WDM-PON System Architectures .................................................................... 66
Simulation and Experimental Results .............................................................. 67
Conclusion ....................................................................................................... 69

iii
Content

4 Comparison of RSOA-FCL with Alternative RSOA Based Colorless

Transmitters .................................................................................................. 71
4.1 Introduction ............................................................................................................... 72
4.2 Mobile Fronthaul WDM-PON with RSOA-FCL as a Colorless Source .......................... 74
4.3 Operating Principle of the RSOA-FCL ......................................................................... 75
4.4 Theoretical Modelling of RSOA-FCL ........................................................................... 76
4.5 Characterization ......................................................................................................... 79
Experimental Setup ......................................................................................... 79
CW Results and Discussion .............................................................................. 80
4.6 Data Transmission Comparison.................................................................................. 83
4.7 Conclusion ................................................................................................................. 86
5 Bit- and Power-Loaded DMT for Maximum Bit Rate .............................. 87
5.1 Introduction ............................................................................................................... 88
5.2 DMT Signal Generation and Reception ...................................................................... 90
5.3 Experimental Setups .................................................................................................. 92
5.4 SNR Characterization ................................................................................................. 93
Electrical Bandwidth of the RSOA ................................................................... 94
RSOA-FCL Performance ................................................................................... 94
Performance Comparison of the Three Transmitter Schemes ........................ 98
5.5 Bit- and Power-Loaded DMT for Highest Bit Rates..................................................... 99
Maximizing the Capacity of the RSOA-FCL ...................................................... 99
Comparison of Maximizing Capacity ............................................................. 101
5.6 Conclusion ............................................................................................................... 102
6 Summary and Outlook ........................................................................ 103
6.1 Summary .................................................................................................................. 103
6.2 Outlook .................................................................................................................... 104
Appendix A: Improved Euler’s Method .................................................... 107
Appendix B: Additional RSOA Simulation ................................................. 109
B.1. Static Modelling ........................................................................................................... 109
B.1.1. Spatial Distribution of Carrier Density, ASE and Signal Powers ........................... 109
B.1.2. Gain Saturation ................................................................................................... 111
B.2. Dynamic Modelling ...................................................................................................... 114
B.2.1. Small Signal Modulation Response...................................................................... 114
B.2.2. Direct Modulation and Modulation Cancellation ................................................ 116
Appendix C: DMT Signal Processing ......................................................... 119
C.1. Bit- and Power-Loading: Chow’s Algorithm ................................................................. 120
C.1.1. Transmitter SNR Characterization ....................................................................... 120
C.1.2. Bit-Loading .......................................................................................................... 121
C.1.3. Power-Loading .................................................................................................... 124
C.2. Real-Valued DMT Signal Generation ............................................................................ 125
C.3. DMT Receiver DSP ....................................................................................................... 126
C.3.1. DMT Timing Synchronization............................................................................... 126
C.3.2. Frequency Synchronization ................................................................................. 127
Appendix D: Polarization Retracing in FCLs .............................................. 129

iv
 Content

Glossary ....................................................................................................... 133

Greek Symbols .................................................................................................................... 133
Latin Symbols...................................................................................................................... 134
Acronyms .................................................................................................... 139
References................................................................................................... 143
Acknowledgement ....................................................................................... 161
List of Publications ....................................................................................... 163
Journal Papers .................................................................................................................... 163
Conference Contributions................................................................................................... 164
Curriculum Vitae .......................................................................................... 167

v
Abstract
Low-cost, high bandwidth optical access networks are key to cope with today’s and future traffic
demand. While high definition (HD) on demand video services such as Netflix, Hulu, etc.
dominate current traffic demands, ultra HD resolution videos, virtual reality, and even more yet
unknown bandwidth hungry services are expected to escalate data rates even further in the near
future. This development is further augmented by the exploding number of high performance
smart phones and tablets (increase by 563 million in 2015) and dawn of the era of the internet
of things (more than 3 connected devices per person). As per Cisco, this has resulted in 74%
growth in mobile data traffic to 3.7 Exabyte per month in 2015 alone. It is forecasted that it will
even reach 30.6 Exabyte by 2020.
Optical fiber communication is already used in the core network, which enables transmission
with highest capacity across continents, countries and metropolitan areas. Core networks utilize
state-of-the-art and future-proof dense wavelength division multiplexed (DWDM) technology.
The large traffic arriving through the core network is first distributed by the metro-ring to the
service providers’ office. Afterwards, an access network delivers the data to the target end user.
With the current traffic growth rate, access networks need to deliver more than 1 GBit/s at an
affordable price. This has prompted the evolution to optical fiber based access network, which
has already been widely deployed.

The discussion to integrate optical fiber in the access network started in the 1980s and
standardization were already set in the 1990s. However, cost sensitivity of access network forced
service providers to push for solutions, which avoid expensive amplifiers and repeaters in the
network. This makes passive optical networks (PONs) the favorable choice. Further, the relatively
simple and cost-efficient time division multiplexing (TDM) scheme has been dominantly
deployed to cut the price of access networks by allowing subscribers to share the major part of
the network infrastructure. Yet, TDM-PONs lack the capacity, flexibility and security that is
expected in next generation access networks. On the other hand, considering the success of
WDM in core and metro networks, it might also be an option for access networks. In WDM-PONs,
a single wavelength channel is assigned to each subscriber, which essentially establishes a virtual
point-to-point connection. This allows for secure link with high capacity, longer reach, and high
flexibility.

Despite these attractive features of WDM-PON technology, a widespread deployment of WDM-

PON in access network is hindered by the high cost of wavelength specific transmitters and the
attached inventory complication. The solution is to develop transmitters whose transmission
wavelength is not a-priori fixed, and it can be set externally. Such devices are generally called
colorless transmitters and they can operate in a broad wavelength range. Thus, mass-produced
identical colorless devices can be installed across all users, which lowers the overall cost and
highly simplify the inventory.
The focus of this thesis is to solve the puzzle of making WDM-PONs economical by investigating
simple and inexpensive colorless upstream transmitters. Specifically, it aims at Reflective
semiconductor optical amplifier (RSOA) based colorless transmitters. The major part deals with
a detailed study of a self-seeded RSOA fiber cavity laser (RSOA-FCL). A mathematical model is
implemented to explain its operation principle. The model is used to explain the long cavity lasing
process and the impact of cavity components, i.e. the dispersive and lossy long fiber cavity and
filter bandwidth. Data transmission simulations also demonstrate the feasibility of transmitting

vii
Abstract

beyond 2.5 GBit/s amplitude shift keying (ASK) signals. It is found that the speed is limited by the
modulation bandwidth of the RSOA. To overcome this limitation, bit- and power-loaded discrete
multi-tone (DMT) modulation format was implemented and line rate up to 20 GBit/s is
demonstrated. Further, the RSOA-FCL is experimentally compared with alternative colorless
transmitters – Amplified spontaneous emission (ASE) spectrum-sliced source and externally
seeded RSOA. Such qualitative comparison helps to simplify the decision making on which
transmitter to pick when designing access networks.
This thesis is structured as follows.
Chapter 1 presents a top-to-down introduction to optical access networks and the multiple
generations of PONs. The motivation behind the evolution towards WDM-PONs, the
accompanying cost issues, and the proposed state-of-the-art colorless solutions are discussed.
Chapter 2 deals with SOAs and RSOAs. It briefly discusses how these devices impacted the optical
fiber communication field and their purpose in current optical access networks. The
fundamentals of light amplification in SOA and RSOA are presented.
Chapter 3 presents mathematical modelling results and comparison with experimental
characterization for the self-seeded RSOA-FCL. The study concentrates on the spectral and
temporal characteristics of the RSOA-FCL output power. Simulation shows that under
appropriate operation setting, the output of the RSOA-FCL is significantly narrower with a
relatively constant power. For example, despite the 100 GHz bandwidth of the WDM filter in the
RSOA-FCL, the output settles with ~20 GHz bandwidth for short cavities or long dispersion
compensated fiber cavities. The relatively constant output power can then be used for data
transmission. Open eye diagrams are observed for 2.5 GBit/s ASK modulation in both simulations
and experiments.
These results are obtained by integrating a mathematical RSOA model into a long dispersive fiber
cavity model. First, a bidirectional multi-mode RSOA was realized by taking into account both
spatial and temporal evolution of the travelling waves and carrier density. ASE was also included
to better simulate the strong gain saturation in RSOAs. The multi-mode RSOA model was then
included in the dispersive standard single mode fiber (SSMF) cavity model.
Chapter 4 presents a detailed comparison of an RSOA-FCL with alternative RSOA based colorless
transmitters namely an RSOA ASE spectrum-sliced source and an externally seeded RSOA. It was
demonstrated that RSOA-FCL performance seats between the two alternatives. Quantitatively,
when switching from the ASE spectrum-sliced source to RSOA-FCL, one obtains an increase of
more than 5 dB in signal-to-noise ratio (SNR). Similarly, the advantage of switching from RSOA-
FCL to externally seeded RSOA is almost 5 dB in SNR. It is also demonstrated that RSOA-FCL is
viable for 5 Gbit/s transmission over 10 km SSMF and 10 GBit/s over 2 km SSMF, which are
interesting for short reach transmission in a mobile fronthaul scenario. The 5 GBit/s transmission
can be extended to 25 km if nonzero dispersion shifted fiber (NZDSF) is used to minimize the
dispersion penalty.
Chapter 5 demonstrates how the capacity of such low-quality and low-cost colorless transmitters
can be maximized through bit- and power-loaded DMT modulation. Adaptive DMT enables to
exploit the narrow modulation bandwidth of the RSOA to the limit. The bit- and power-loaded
DMT was implemented with an RSOA ASE spectrum-sliced source, an RSOA-FCL, and an
externally seeded RSOA. In back-to-back, the three schemes supported 6, 20 and 30 GBit/s line

viii
 Abstract

rates, whereas each transmitted 4, 10 and 25 GBit/s over 25 km NZDSF. These results indicate
the positive impact of combining adaptive DMT with low-quality source to boost their capacity
beyond expectation. Further, the bit rate reduction due to transmission over dispersive fiber can
be avoided by operating these schemes in the O-band.
Chapter 6 summarizes the results of this thesis and presents an outlook for future work.

ix
Zusammenfassung
Optische Zugangsnetze mit hoher Bandbreite und zu einem niedrigen Preis sind zentrale
Elemente der Telekommunikation, um gegenwärtiges und künftiges Verkehrsaufkommen zu
bewältigen. Während High Definition (HD) On-Demand-Videodienste wie Netflix, Hulu, etc.
derzeit das Verkehrsaufkommen dominieren, werden in naher Zukunft ultra-HD Videos, Virtual
Reality und noch unbekannte, bandbreitenintensivere Anwendungen die Datenraten weiter
steigern. Diese Entwicklung wird zudem verstärkt durch die explodierende Zahl an
Hochleistungs-Smartphones und -Tablets (Anstieg um 563 Millionen in 2015) und dem Beginn
einer Ära der Internet der Dinge (mehr als drei verbundene Geräte pro Person). Laut Cisco, hat
das alleine zu einem Wachstum des mobilen Datenverkehrs von 74% zu 3.7 Exabyte pro Monat
in 2015 geführt. Bis 2020 werden sogar 30.6 Exabyte prognostiziert.
Optische Kommunikation mit Glasfasern wird schon jetzt in Zugangsnetzen eingesetzt, was die
Übertragung mit höchster Kapazität über Kontinente, Städte und Ballungsräume hinweg
ermöglicht. Diese Zugangsnetze beruhen auf hochmodernen und zukunftsfähigen
Wellenlängenmultiplexverfahren (engl. Wavelength Division Multiplex, WDM). Das große
Verkehrsaufkommen, das durch das Backbone-Netz ankommt, wird zunächst vom Metro-Ring zu
dem Service Provider verteilt. Danach liefert das Zugangsnetz die Daten an den Endkunden. Für
Zugangsnetze, bei denen der Abstand zwischen den Kunden kürzer als 100 m ist, können
moderne kupferbasierende Technologien noch die nötigen 1 Gbit/s liefern. Mit den heutigen
Wachstumsraten an Datenvolumen ist jedoch mehr als 1 Gbit/s zu einem viel niedrigeren Pries
erforderlich. Das hat die Entwicklung zu einem optischen Zugangsnetz mit Lichtwellenleitern
veranlasst, das bereits verbreitet eingesetzt wird.
Optische Lichtwellenleiter in Zugangsnetze zu integrieren wurde bereits in den 1980er Jahren
diskutiert und in den 1990ern standardisiert. Dennoch hat die hohe Kostensensitivität der
Zugangsnetze die Service Provider gezwungen, Lösungen ohne teure Verstärker und Repeater
voranzutreiben. Hier sind Passive Optische Netze (PON) die bevorzugte Wahl. Die Kosten pro
Kunde werden zudem gesenkt, indem sich mehrere Nutzer Glasfaser und zugehörige
Netzinfrastruktur teilen, ermöglicht durch Multiplexverfahren. Bis vor kurzem wurde das relativ
einfache und kosteneffiziente Zeitmultiplexverfahren (engl. time division multiplexing, TDM)
eingesetzt. Obwohl diese Technologie ausgereift ist, kann TDM nicht mit dem explodierenden
Zuwachs an benötigter Bandbreite mithalten. Daher verfolgt die Community disruptive
Technologien. Angesichts des Erfolgs von WDM, das Hochgeschwindigkeitsübertragung in
Backbone-Netzen und Metropolitan Area Networks ermöglicht, wird es auch als eine Option für
Zugangsnetze angesehen. In WDM-PONS, wird jedem Kunden ein eigener Wellenlängenkanal
zugewiesen, was im Grunde eine virtuelle Punkt-zu-Punkt Verbindung herstellt. Anders als TDM-
PON, wo ein Leistungsteiler das Downstream-Signal an alle Kunden verteilt, basiert WDM-PONs
auf WDM-Filtern, die eine bestimmte Wellenlänge einem bestimmten Nutzer zuordnen. Das
ermöglicht Sicherheit, hohe Kapazität, große Reichweite und hohe Flexibilität.
Dennoch, die Migration zu WDM-PON steht vor einigen Herausforderungen. Im Falle von
Brachflächen, wo Altsysteme basierend auf TDM-PONs bereits installiert sind, müssten durch
den Einsatz von WDM-PONs die Leistungsteiler an einem Remote Node (RN) durch WDM-Filter
ersetzt werden. Service Provider haben jedoch schon in Leistungsteiler-RNs investiert und sind
noch nicht bereit diese wieder wegzuwerfen. Als Konsequenz hat das Full Service Access Network
(FSAN) Forum das hybride TWDM-PON in der neusten Generation PON2 (NG-PON2) empfohlen.

xi
Zusammenfassung

In TWDM-PONS, werden verschiedene TDM-PON-Wellenlängenkanäle kombiniert, um so die

Bitrate der Nutzer zu steigern. Die Wellenlängenselektion wird durch einen einstellbaren Filter
am Benutzer vorgenommen. Dadurch wird das Leistungsteiler basierte optische Verteilernetz
(engl. optical distribution network, ODN) erhalten und gleichzeitig die Übertragungskapazität
erhöht.
TWDM-PON ist ein erstes Zeichen dafür, dass WDM-PON eines Tages TDM-PON ablösen wird.
Das ist der Grund, weshalb NG-PON2 für spezielle Anwendungen zusätzlich WDM-PON
empfiehlt. Die Hauptanwendungen, die von WDM-PON profitieren werden, sind Fiber To The
Business/Office (FTTB/O) und Fiber To The Antenna (FTTA). FTTO bedient Unternehmen und
Geschäftsstellen, die eine schnelle, sichere Verbindung benötigen, wofür WDM-PON eine ideale
Lösung ist. In der nächsten Generation von Mobilnetzen, stellt man sich kleine Zellen und dichte
Zentralbüros innerhalb des Cloud-Funktzugangsnetzes (engl. cloud radio access network, C-RAN)
vor. In einem C-RAN, benötigt das mobile Fronthaul (MFH) Netzwerk zwischen den Remote Radio
Heads (RRH)s an den Antennen und Base Band Units (BBUs) am Zentralbüro eine
Hochgeschwindigkeitsverbindung. WDM-PON basiertes FTTA ist ein geeigneter Kandidat für C-
RAN.
Die Herausforderung, die WDM-PON daran hindert, in FFTB/O und FTTA die Vorherrschaft zu
übernehmen, sind die hohen Kosten eines wellenlängenspezifschen Upstream Senders und die
damit einhergehenden Bestandskomplikationen. Die einzige Lösung dafür ist ein Sender, dessen
Transmissionswellenlänge nicht von vornherein fix ist und die extern innerhalb eines erlaubten
Trasnmissionswellenlängenbereichs gesetzt werden kann. Derartige Bauelemente heissen
farblose Sender und arbeiten in einem breiten Wellenlängenbereich. Auf diese Weise, können in
Massenproduktion hergestellte, identische, farblose Geräte für alle Benutzer installiert werden,
was die Gesamtkosten senkt und den Bestand vereinfacht. Reflektive Halbleiterlaserverstärker
(engl. reflective semiconductor optical amplifier, RSOA). sind eine mögliche Lösung für farblose
Sender aufgrund ihres breiten optischen Spektrums und ihrer geringen Kosten.
Der Fokus dieser Thesis ist das Rätsel zu lösen, wie ein WDM-PON wirtschaftlich sinnvoll gemacht
werden kann, indem einfache und kostengünstige farblose Upstream Sender untersucht werden.
Insbesondere zielt die Arbeit auf RSOA basierte farblose Sender ab. Der Hauptteil behandelt eine
detaillierte Untersuchung von Self-Seeded RSOA Fiber Cavity Lasern (FCL). Um Physik und
Funktionsprinzip zu erklären, wird ein mathematisches Modell aufgestellt. Dieses wird benutzt,
um den Lasing Prozess in einer langen Kavität zu erklären sowie den Einfluss von
Kavitätskomponenten, d.h. die Effekte der dispersiven langen Faserkavität und der Filter-
Bandbreite. Simulationen der Datenübertragung zeigen, dass es machbar ist ein 10 Gbit/s Signal
mit Amplitudenumtastung (engl. amplitude shift keying, ASK) zu übertragen. Die Ergebnisse
zeigen zudem, dass die Geschwindigkeit durch die Modluationsbandbreite des RSOAs begrenzt
ist. Um diese Bandbreitenbegrenzung zu überwinden, wurde ein Bit- und Power-Loaded Discrete
Multi-Tone (DMT) Modulationsformat implementiert und so Bitraten bis 20 Gbit/s demonstriert.
Außerdem wird der RSOA-FCL mit alternativen Sendern verglichen – Spectrum Sliced Source und
Externally Seeded RSOA. Der qualitative Vergleich hilft bei der Entscheidung, welcher Sender
beim Entwurf des Zugangsnetzes gewählt werden sollte.
Diese Thesis ist wie folgt strukturiert:
Kapitel 1 behandelt eine Top-Down Einführung in optische Zugangsnetze und die verschiedenen
PON-Generationen. Die Motivation hinter der Entwicklung in Richtung WDM-PONs, die damit
einhergehende Kostenfrage und die vorgeschlagene moderne farblose Lösung werden diskutiert.

xii
 Zusammenfassung

Kapitel 2 behandelt SOAs und RSOAs. Es diskutiert kurz wie diese Bauelemente die optische
Glasfaserkommunikation beeinflusst haben sowie ihr Einsatzgebiet in heutigen optischen
Zugangsnetzen.
Kapitel 3 präsentiert die Ergebnisse der mathematischen Modellierung und ein Vergleich mit
experimenteller Charakterisierung für den Self-Seeded RSOA-FCL- Die Studie konzentriert sich
auf die spektralen und zeitabhängigen Eigenschaften der RSOA-FCL Ausgangsleistung.
Simulationen zeigen, dass, auch wenn ein RSOA breitbandig verstärkte spontane Emission (engl.
amplified spontaneous emission, ASE) aussendet, ist bei geeigneten Betriebseinstellungen der
Ausgang des RSOA-FCL signifikant schmalbandiger mit relativ konstanter Leistung. Ein Beispiel:
Trotz einer 100 GHz 3 dB Bandbreite des WDM-Filters im RSOA-FCL, pendelt sich der Ausgang
mit ~20 GHz Bandbreite für kurze Kavitäten oder lange dispersionskompensierenden
Faserkavitäten ein. Ausserdem unterdrückt die saturierte RSOA-Nichtlineartität das ASE
Ampitudenrauschen, was zu einer konstanten Ausgangsleistung führt, die dann für
Datenübertragung verwendet werden kann. Sowohl in Simulation wie auch im Experiment
werden offene Augendiagramme für 2.5 Gbit/s ASK Modulation beobachtet.
Die Ergebnisse wurden durch Integration eines mathematischen RSOA Models in ein Modell
einer langen dispersiven Faserkavität erzielt. Zuerst wurde ein bidirektionaler Multimoden-RSOA
modelliert unter Berücksichtigung der räumlichen und zeitlichen Ausbreitung der
elektromagnetischen Welle und der Ladungsträgerdichten. ASE des RSOAs wurde ebenfalls
einbezogen, um die starke Verstärkungssättigung in RSOAs besser modellieren zu können. Das
Multimoden-RSOA-Modell wurde dann in ein Kavitätsmodell mit dispersiven Standard-
Monomodefasern (engl. standard single mode fiber, SSMF) eingefügt.

Kapitel 4 präsentiert einen detaillierten Vergleich eines RSOA-FCL mit alternativen RSOA
basierten farblosen Sendern, namentlich RSOA ASE Spectrum Sliced Source und ein Externally
Seeded RSOA. Es wurde gezeigt, dass die RSOA-FCL Performanz zwischen der der beiden
Alternativen liegt. Quantitativ erzielt man mehr als 5 dB Steigerung des Signal-zu-Rausch-
Verhältnisses (engl. signal-to-noise ratio, SNR), beim Vergleich zwischen einer ASE Spectrum
Sliced Source und einem RSOA-FCL. Zudem wird gezeigt, dass RSOA-FCL für 5 Gbit/s Übertragung
über 10 km SSMF brauchbar sind und 10 Gbit/s über 2 km SSMF, was für Übertragung über kurze
Strecken in einem MFH Scenario interessant ist. Übertragung von 5 Gbit/s kann man bis 25 km
erwarten, wenn eine nonzero disperion shifted fiber (NZDSF) benutzt wird, um
Dispersionsnachteile zu minimieren.
Kapitel 5 zeigt wie die Kapazität eines solchen niederwertigen und kostengünstigen farblosen
Senders maximiert werden kann durch Bit- und Power Loaded DMT Modulation. Adaptives DMT
ermöglicht die schmale Modulationsbandbreite des RSOAs bis ans Limit auszureizen. Bit- und
Power Loaded DMT wurde implementiert mit einer RSOA ASE Spectrum Sliced Source, einem
RSOA-FCL und einem Externally Seeded RSOA. Back-to-back unterstützten die drei Systeme 6,
20 und 30 Gbit/s Bruttodatenraten, während sie 4, 10 und 25 Gbit/s über 25 km NZDSF
übertrugen. Dieses Ergebnis lassen auf einen positiven Einfluss von der Kombination adaptives
DMT mit niederwertigen Quellen schließen, um so ihre Kapazität jenseits ihrer Erwartungen zu
steigern. Ausserdem kann die Bitraten-Reduzierung durch Übertragung über eine dispersive
Glasfaser verhindert werden, indem die Systeme im O-Band betrieben werden.
Kapitel 6 fasst die Arbeit der Thesis zusammen und präsentiert einen Ausblick in zukünftige
Arbeit.

xiii
Challenge and Achievements of This Work
The challenge addressed in this work is the development of an economical colorless upstream
transmitter that is decisive for the success of wavelength division multiplexing (WDM) based
passive optical networks (PONs). The integration of WDM technology in an access network is
marred by the high cost of wavelength specific components which is accompanied by inventory
complication. Cost is even more critical for equipment that is installed at the subscriber, as the
latter has to cover the expenses. Thus, the development of economically feasible colorless
upstream transmitters is mandatory if WDM-PONs are to be deployed in the access network.
As a promising colorless solution for WDM-PON, this work investigates the self-seeded reflective
semiconductor optical amplifier (RSOA) fiber cavity laser (FCL). A theoretical model was
implemented to investigate how such long cavity laser can function as directly modulated
transmitter. The simulation results are compared with experimental results for an RSOA-FCL,
which showed both qualitative and quantitative agreements. Further, data transmission
experiments are performed starting with simple amplitude shift keying (ASK) modulation, with
which up to 10 GBit/s transmission is demonstrated. These performance results are compared
with alternative low-cost RSOA based colorless transmitters: amplified spontaneous emission
(ASE) spectrum-sliced source and externally seeded RSOAs. For all schemes, the capacity is
boosted 2 to 3 fold by utilizing advanced modulation formats with adaptive discrete multi-tone
(DMT) signal.
The achievements of this work are briefly summarized as follows:
 Bidirectional and multimode RSOA mathematical model
A simple and yet illustrative RSOA model was implemented to be integrated in the simulation of
the self-seeded RSOA-FCL. It had to be adapted to include the following properties that are
obligatory for an accurate model. First, it takes into account bidirectional multimode amplified
spontaneous emission (ASE) and optical signal fields, with both phase and amplitude. The
multimode nature is necessary to realize the multimode characteristics of the RSOA-FCL. Second,
both the ASE and signal are spatially resolved within the RSOA, as RSOAs are known to have
spatially dependent carrier and gain distribution. This is important to describe the nonlinear
properties of an RSOA that is needed for the proper functioning of the RSOA-FCL.
 Self-seeded RSOA-FCL as a directly modulated transmitter

The RSOA-FCL model consists of a multimode, bidirectional RSOA, a dispersive fiber cavity and
an optical bandpass filter (OBPF). The OBPF mimics the WDM filter in a WDM-PON. This model
is used to investigate how long cavity laser can function as directly modulated transmitter. First,
even if the self-seeding starts from a broad band filtered noisy ASE of the RSOA (filtered by
100 GHz OBPF), the RSOA-FCL output settles with an almost constant output power and
significantly narrower spectrum (~20 GHz). In fact, the output consists of a large number of
fluctuating longitudinal modes, which still sum up to give a constant power. The constant output
power permits one to encode amplitude shift keying (ASK) signal by directly modulating the bias
current of the RSOA. Further, the narrower spectrum means RSOA-FCL suffers much lower
dispersion penalty compared to a spectrum-sliced scheme. The direct modulation capacity is
demonstrated through both simulation and experiment by open eye diagrams at 2.5 GBit/s ASK
modulation. For longer dispersive cavities, the RSOA-FCL becomes extra highly multi-mode and

xv
Challenge and Achievements of This Work

the chromatic dispersion of the fiber cavity induces strong mode partition noise. In case of longer
than 500 m standard single mode fiber (SSMF) cavity, the output signal spectrum did not narrow
down.
 Self-seeded RSOA-FCL versus alternative colorless schemes
The RSOA-FCL is compared with alternative RSOA based colorless transmitters. It is
experimentally shown that an RSOA-FCL outperforms an ASE spectrum-sliced source by more
than 5 dB in signal-to-noise ratio (SNR). However, externally seeded RSOA has an advantage of
~5 dB SNR over the RSOA-FCL. An RSOA-FCL offers higher SNR than an ASE spectrum-sliced
source due to the lasing process achieved by forming the Fabrey-Perot resonator cavity. Lasing
sets in once the RSOA gain is able to compensate for the resonator cavity losses. Thus, the RSOA-
FCL output power is >8 dBm higher than that of an ASE spectrum-sliced source. In addition, the
relative intensity noise (RIN) of the RSOA-FCL is up to 10 dB/Hz lower since the saturated RSOA
suppresses the amplitude noise of the seeding ASE. Data transmission experiments with the
RSOA-FCL demonstrated 5 GBit/s ASK transmission over 8 km SSMF and 25 km non-zero
dispersion shifted fiber (NZDSF). 10 GBit/s ASK is also transmitted over 2 km SSMF with an RSOA-
FCL. It is found that transmission fiber dispersion poses a challenge, which can be solved by O-
band operation.
 Bit- and power-Loaded DMT for up to 20 GBit/s
Bit- and power-loaded DMT modulation is used to achieve record high line rates for the three
RSOA based colorless transmitters. 6, 20, and 30 GBit/s line rates are demonstrated in back-to-
back for an ASE spectrum-sliced, RSOA-FCL, and externally seeded RSOA sources, respectively. 4,
10 and 25 GBit/s line rates are transmitted over 25 km NZDSF for the three schemes.

xvi
1 Introduction
Within the last few decades, the telecommunication networks have evolved to become a
necessity in the present global and highly connected society. The impact of telecommunication
is visible in social and economic developments around the world. People connect through social
media, share files, pictures, videos, news and so on. Education facilities and businesses are able
to send and receive vast amounts of data in an instant. Data centers and super computers are
connected to gather and analyze vast amount of raw data that needs to be changed into useful
information. However, the number of services and connected devices and therefore the required
amount of traffic capacity is still growing exponentially [1]. To cope with this, the
telecommunication service providers, vendors, and researchers are always in search for the next
big technology.
Typically, telecommunication systems have a hierarchical structure. Traditionally, such systems
consist of three network sections that differ by transmission capacity and reach that they offer –
core, metro and access. The beginning of this chapter briefly introduces these network sections.
The second subsection concentrates on the access network. It presents the fundamentals of the
technologies implemented and state-of-the-art solutions.

1
Introduction

1.1 Telecommunication Network

The overall telecommunication network architecture has three sections, see Figure 1-1 [2]. The
first section is the core network, which interconnects large cities, countries and continents. The
second section, a metro or regional network, connects the central offices (CO) in metropolitan
areas or regions. The third section, an access network, connects the end users to the COs of the
service providers. Each of these sections has different requirements with respect to bandwidth
capacity and costs.
The core network carries huge amount of data, up to Terabit/s across oceans and continents, see
Figure 1-2 [3]. Present core network relay on state-of-the-art optical fiber communication
technologies – dense wavelength division multiplexing (DWDM) with all-optical regenerators,
optical amplifiers and coherent reception. The distances in the core network extend from
hundreds to thousands of kilometer. For example, submarine fiber optic cables laid across the
Atlantic and Pacific oceans are around 6000 and 10000 km long, respectively [3]. The core
network is not easily and frequently upgradable, thus it is well-planned and future proof. Since
different service providers and all users of the network share installation and operating expenses
of the high capacity links in this section, higher cost for state-of-the-art technology can be
tolerated as long as high capacities, reach, and reliability can be achieved.

Figure 1-1. The global telecommunication network has three sections, the core network, the metro/regional network and the
access network. The core network, on the top represent the long haul and high capacity (TBit/s) networks. The metro network
aggregates the signal from multiple central offices (CO) for transmission over the core network. It also distributes the signal from
the core network to the COs. Finally, the access network section connects the end users (subscribers) to the COs. The access
network is a multi-technology zone, where different mediums such as copper, wireless (radio frequency, RF), optical fiber (fiber to
the x, FTTx), or any combination is deployed [2].

The second network section is the metro-network, which distributes the large quantity of data
from the core network into metropolitan areas as shown in Figure 1-1. It also aggregates and
transports the traffic from the access network to the core network. In many cases, it has a ring
topology that links multiple central offices (COs) and provides resilience towards cable failures.

2
 Introduction

Here also, state-of-the-art wavelength division multiplexed (WDM) networks with multiple tens
of gigabit capacity and switches provide a future proof infrastructure.

Figure 1-2. Undersea optical fiber cable capacity growth through time: The capacity grew from below ~1 Gbit/s with coaxial cable
and RF amplifiers to 100 Tbit/s over optical fiber link since the introduction of amplified coherent transmission. This is more than
100,000 fold over 70 years [3]

The last section of the telecommunication network is the access network, which connects the
subscriber to the service provider COs, see Figure 1-1. In general, access networks face
challenges to distribute the vast traffic from the high capacity core and metro network to the
destination user [4]. However, the fact that the cost of an access link has to be covered by the
user prohibits expensive solutions. This has forced researchers to concentrate on budget
schemes with reasonable capacity. This capacity discrepancy between the well-developed core
metro networks and the access network is often referred to as the last mile problem. Different
service providers implement various economical technologies to deliver the needed bandwidth.
The implemented technologies also vary based on the type of the subscriber, which could be
residence, business, office, and so on. Due to the large variety of technologies and standards in
the access network, this section is also referred to as a multi-technology zone [5].
In general, a large number of phenomena dictate the evolution path of telecommunication
networks, specifically of the access networks [6]. For example, the availability of high
performance smartphones and tablets at a constantly reducing price means that an increasing
number of customers requires high quality and capacity connections, see Figure 1-3(a) [7]. In
addition, the main content of mobile data transmission is no longer voice data, as shown in
Figure 1-3(b) [7, 8]. Rather, high bandwidth consuming services such as high definition video-on
demand and file sharing require the bulk of the network capacity. This necessitates an
economical high capacity optical access network, which can also satisfy the growing demand in
residential areas, business and also in the mobile networks.
Practically, copper wire technologies are no longer able to cope with the ever-increasing capacity
requirements. Wireless links are still in operation with continuing improvements such as using
higher carrier frequencies and better directivity.

3
Introduction

Figure 1-3. Global devices and data type evolution based on Cisco [7, 8]: (a) shows that the main contributor to the overall device
communication traffic are Machines (M2M, machine-to-machine). The number of smartphones and tablets is also increasing
while the PC number is reducing, indicating that future users will be mobile. (b) shows that by 2019, 72 % of the mobile traffic will
be mobile video.

Novel microwave-photonic technologies promise data rates beyond 100 Gbit/s [9]. Fiber-optic
access networks, however, promise highest data rates and reliability for future networks. The
vast research undertaken and experience obtained from exploiting optical fiber for core and
metro networks has paved the way for fiber optics in the access network. The costs of optical
components have reduced to a level, which is acceptable for the access network.

Since the introduction of optical fiber in the access network, there have been multiple
generations of optical access network technologies and standardizations. The next subsection
presents the details of these technologies and generations.

1.2 Optical Access Network

This subsection briefly introduces the fundamentals of optical access networks. It discusses the
different nomenclatures and technical concepts. The multiple generations of optical access
networks are also presented. For each generation, the state-of-the-art technology is reported.

Fiber to the x (FTTx)

This subsection discusses the most common names used to describe optical access networks.
The generic name fiber to the x (FTTx) incorporates all optical fiber based access networks, where
x replaces the termination point. The x describes how deep fiber penetrates into the last mile or
customer premises [10, 11]. The most common alternatives of FTTx are listed next. One should
note that these definitions are based on books and literature [10, 12-15], and some may vary
between sources. Therefore, a brief definition is given to minimize confusion.

Fiber to the Node (FTTN) or Fiber to the Neighborhood (FTTN)

In FTTN, see Figure 1-4(a), the fiber link is only up to a remote node (RN), which is located in the
neighborhood. Copper wire based xDSL (Digital subscriber line) technology connects the
subscriber with the RN. In case of cable television service providers (which are also called multi
service providers, MSO), a variant of FTTN called hybrid fiber coax (HFC) is deployed to deliver
triple play service. In HFC, a coaxial cable, which was once used only for video broadcast is
exploited to deliver voice and data service as well. In FTTN, the distance from the node to the
user end may range from hundreds of meters to tens of kilometers. FTTN is the most economical
as it reuses the majority of the already installed copper network [16].

4
 Introduction

Fiber to the Cabinet (FTTCab) or Fiber to the Curb (FTTC)

The idea behind the FTTC is to shorten the length of the copper link to maximum of few hundred
meters by installing a cabinet between the RN and the user premises as shown in Figure 1-4(b).
Based on [17], the difference between the FTTCab and FTTC is that, in FTTC, the copper cable
length to the end user is much shorter. xDSL access multiplexers or switches are placed at the
cabinet to aggregate the neighborhood traffic [10]. In addition, by utilizing upgraded twisted pair
links, improved bandwidth can be achieved and bit rates up to 100 Mbit/s are available [16].

Fiber to the Building (FTTB)

In FTTB, the fiber is terminated at a building. The building could be residential or business, see
Figure 1-4(c). In case of residential building, MDU (Multi dwelling point) is placed at the building
to distribute the broadband service to households. In case of business building, a higher capacity
MTU (multi-tenant unit) is used to convert the optical signal into electrical form. The link from
the MDU/MTU to the subscriber is through xDSL or coaxial cable (the latter is for HFC).

Fiber to the Office (FTTO)

FTTO has a similar topology as FTTB, the only difference being the termination box, which in this
case is a single business unit (SBU), is intended for a single office or business as shown in
Figure 1-4(d). The capacity could reach 1 GBit/s depending the customers need. Details are
available in [11, 15]

Fiber to the Home (FTTH)

In FTTH, each subscriber has a dedicated fiber connection to the service provider CO or to a RN.
However, copper based or wireless links are used within the household to connect user
equipment, see Figure 1-4(e). Depending on the need of the household, bit rates of up to 1 GBit/s
or more are being deployed.

Fiber to the Desk (FTTD)

FTTD refers to the a scheme where an optical fiber is used up to the outlet at a desk or personal
computers (PCs) as shown in Figure 1-4(f). The devices or PCs will need fiber network compatible
cards. Such scheme is especially important for large institutions such as data centers with a large
amount of data sharing . For standard business and house hold, the cost of all optical
implementation still needs further reduction [18]. In case of institutions, e.g. data centers,
capacities beyond 10 GBit/s are becoming common.

Fiber to the Antennae (FTTA)

In the next generation wireless networks, the backhaul and the newly emerged fronthaul are
based on optical fibers. For mobile backhauling, the fiber terminates at the cell-site backhauling
unit (CBU) positioned at a base station. On the other hand, in the mobile fronthaul case, the fiber
terminates at the antenna, as shown in Figure 1-4(g).

5
Introduction

Figure 1-4. Types of optical access networks based on the penetration of the optical fiber link (from the central offices (COs) to
the end user’s premises. The link is called fiber to the x (FTTx), x being the termination point. (a) in fiber to the node (FTTN) the
optical network ends at a remote node (RN) in the neighborhood after which copper based links are used. In fiber to the cabinet
(FTTC), (b), the fiber goes a bit deeper up to a cabinet to shorten the copper link thus improving the performance. (c) shows a fiber
to the building (FTTB, either a residential or business building). The fiber terminates at an MDU (multi dwelling unit) or MTU
(multi-tenant unit) which convert the optical signal to electrical domain for distribution to the families or tenant businesses,
respectively. (d) shows the fiber to the office (FTTO) where single business (with single business units (SBUs)) are optically
connected to the COs. In a fiber to the home (FTTH), see (e), the fiber ends at the household inside the single family unit (SFU).
The household devices are connected to the SFU through copper or wirelessly. If the fiber terminates at the device, the scheme is
called fiber to the desk (FTTD), see (d). In a mobile network, the antennae are also connected with fiber to the CO, called fiber to
the antennae (FTTA). The cell-site backhauling unit (CBU) converts the optical signal to electrical domain with the appropriate
protocol.

FTTx Technologies
FTTx deployments topology are either point-to-point (PtP) or point-to-multipoint (PtMP). As the
name implies, PtP dedicates a single fiber to each subscriber as shown in Figure 1-5(a) and (b).
PtP links are mainly based on Ethernet technology, and are also called home run Ethernet [19,
20]. Practically, multiple fibers from an Ethernet switch at the CO can be bundled together up to
a remote location with a splicing box, from which dedicated fibers branch out to the subscribers,
see Figure 1-5(b). In PtP scheme, each fiber to the subscriber can be exploited to its full capacity,
thus offering gigabit speed. The downside of PtP topology is its relatively high cost due to large
number of dedicated devices per subscriber, i.e. dedicated transceiver per user at the CO and
dedicated fiber per user. This has generally limited the deployment of home run Ethernet to
business and communities that can afford the finance.
The cost of FTTx can be lowered by sharing section of the fiber up to a certain location, i.e. a RN
positioned in the vicinity of the users as depicted in Figure 1-5(c) and (d). This architecture is
referred to as PtMP. PtMP access networks can be further classified based on the network
components installed at the RN. Active optical networks (AOMs) utilize powered components at

6
 Introduction

the RN, the most common one being an Ethernet switch. Such AOMs are called active Ethernet
and the schematics is shown in Figure 1-5(c). Compared to PtP home run Ethernet, active
Ethernet reduce the amount of fiber that should exit from the CO. Yet, installing a powered
equipment at a RN is not convenient for service providers, as it also means the installation of
power source and related extra infrastructure.

Figure 1-5. FTTx implementation topologies: (a) point-to-point (PtP) or home run Ethernet means that each subcarrier has a
dedicated fiber link to the central office (CO). (b) is also another flavor of PtP approach, where the fibers are bundled together up
to a remote location with a splice box. Each subscriber is then connected to its dedicated port at this location. (c) and (d) show
point-to-multipoint (PtMP) topologies. (c) depicts an active optical network (AON) or active Ethernet approach where an
electrically powered Ethernet switch is positioned at the remote node to distribute and aggregate the downstream and upstream
Ethernet signals, respectively. (d) shows a passive optical network (PON) where the RN consists of only passive components.

For access networks where the distance from the RN to the subscriber is below 20 km, simpler
and economical topology were found to be passive optical networks (PONs). The topology of a
PON is shown in Figure 1-5(d), where a passive device is deployed at the RN to distribute the DS
signal and combine the US signal. The removal of any active components and related system
simplification at the RN significantly reduces the installation cost. Therefore, PONs have been
economically viable for large-scale deployment of fiber based access network.

Passive Optical Networks (PONs)

PtMP PONs can be implemented based on different multiplexing technologies. The following
subsections discuss the two dominant schemes and the components of a standard PON [21].

1.2.3.1 Multiplexing Technologies

This section presents the multiplexing and multiple access control (MACs) protocols
implemented in PONs. The MAC protocols are needed to manage how each user receives the
signal that is addressed to it. In addition, they also manage how each user can send its US signal
without any collision in the shared fiber link.

7
Introduction

In PON, multiple schemes can be used to multiplex the signals for the multiple users in the shared
fiber. The two dominant ones are time division multiplexing (TDM) and WDM.

Time Division Multiplexed (TDM)

TDM has been the major multiplexing scheme used in PONs due to its simplicity and cost-
efficiency, see Figure 1-6(a). It allocates specific time slots to each user, and the transmission is
made over a single wavelength for all subscribers. In the downstream, the signal from the CO is
broadcasted to all users. The user equipment, called optical network unit (ONU), detects the
address and receives the frame meant for the user. In the US direction, the signals from all the
users have to be synchronized to avoid any collision. The timing synchronization must take into
account the different distances of users form the RN. The synchronization also considers the
possibility that different users may be allocated different time slot sizes based on their
bandwidth requirement, technology called dynamic bandwidth allocation. In general, the
stringent timing synchronization and the dynamic slot allocation can all be done with relatively
low-cost electronics. This has made TDM the most dominant multiplexing scheme in the access
networks.
However, it has some drawbacks. The broadcasting of the downstream signal to all the users
makes it insecure. In addition, the power loss of the power splitter at the RN highly limit the
reach of TDM-PONs. Further, the dynamic time slot allocation for maximum bandwidth will not
suffice to cope with the Gbit/s/user requirements.

Wavelength Division Multiplexing (WDM)

The next inline technology is WDM. In WDM-PONs, specific wavelengths are assigned to the each
user as shown in Figure 1-6(b). This way, the link functions as a virtual point-to-point link even if
the infrastructure is PtMP. The dedication of a single wavelength per user and the replacement
of the power splitter with a WDM filter at the RN allows for long reach PONs by improving the
power budget. In addition, WDM-PONs have higher capacity per user compared to the TDM-
PONs.
However, the transition from TDM-PON to WDM-PON is hindered by two factors. One is the high
cost and complicated inventory of wavelength specific transceivers. Second, TDM-PONs depend
on power splitter at the RN, which have to be replaced by WDM filters in case of WDM-PONs,
which is an expensive process. The former issue is solved by developing wavelength agnostic
transceivers. Identical devices can then be installed across all users, which simplify the inventory
and lower the cost.
There are also other multiplexing schemes. In code division multiplexed access (CDMA), the
subscribers are assigned orthogonal codes. At the receiver, the addressed signal can be
separated by convolving the overall signal with a signature code of the targeted user. However,
in order to accommodate an appropriate number of subscribers, the codes have to become
longer and complex. This translates into unnecessary higher line rate requirement without any
substantial coding gain. Further, the spectral broadening due to longer codes results in higher
dispersion penalty. Thus, CDM has not captured that much interest so far. Electrical subcarrier
multiplexing (SCM) is another scheme where the users are assigned specific electrical RF carrier.
In orthogonal frequency division multiplexing access (OFDMA), the subscribers are assigned
specific number of orthogonal subcarriers. All these schemes are still only on research level to
be implemented in optical access network.

8
 Introduction

Due to the maturity of its technology, TDM has been adopted as a standard for multiple
generations of PONs. In general, It also benefits from inexpensive and simple TDM transceivers.
The issue with TDM is the missing scalability when it comes to upgrading the bitrates of users.
Even with complicated algorithms for dynamic bandwidth allocation and ranging, TDM does not
offer a future proof solution. Further, the subscribers have to be equipped with a receiver
operating at the full rate of the TDM-PON even if they only receive for a specific time slot thus
lowering the effective bit rate they really receive. And in case of an evolution to a higher speed
TDM-PON, all user equipment may need to be replaced.

Figure 1-6. Point-to-multipoint (PtMP) passive optical networks (PONs) multiplexing technologies: (a) shows a time division
multiplexed (TDM) PON. In TDM-PON, a specific time slot is assigned to the multiple subcarriers. In downstream (DS), the signal
for the central office (CO) is broadcasted to all users by the power splitter at the remote node (RN). However, the receiver chooses
the assigned time slot to detect. In the upstream (US), the signal from the subscribers is combined by a power combiner at the RN
and transmitted to the CO. (b) shows a wavelength division multiplexed (WDM) PON. Here a specific wavelength is assigned to
each subscriber establishing virtual point to point link.

In general, the decision to choose a multiplexing scheme depends on multiple parameters. The
splitting ratio or the planned number of subscribers should be considered. In general, it is
preferential if the network can serve more users with the same infrastructure. However, in case
of TDM-PONs, large number of users means that the power budget is highly reduced, thus
limiting the reach. For example, 1:32 power splitter induces around 15 dB power attenuation. In
addition, the available time slot for the individual users also diminishes, thus lowering the per
user data rate. In case of WDM-PON, scaling is done by allocating additional channel to new
users. Of course, that available optical spectrum must be used effectively to accommodate more
users. Since there is no power splitter in WDM-PONs, one needs only to consider the insertion
loss of the WDM filter at the RN. Further considerations include data security of the multiplexing
scheme. With this respect, TDM-PON sufferes as the DS information is literally broadcasted to all
users, thus one users confidential data is available at all users. This is not the case in WDM-PON
where each subscriber has an independent path to the CO. In case of a network upgrade, it is
preferred to implement the scheme which will utile majority of the already existing
infrastructure.

9
Introduction

1.2.3.2 PON Components

This subsection discusses the main components of a standard PON as depicted in Figure 1-7. A
standard PON has the following building blocks.

Optical Line Terminal (OLT)

OLT is placed at the central office, and it is where the DS signal processing and electro-optical
conversion takes place. One OLT serves multiple subscribers by multiplexing the signals by using
either TDM or WDM schemes. On the other side, it also interfaces the PON with the backbone
metro network.

Optical Network Unit (ONU)

ONU is the termination point of the optical section of a PON on the subscriber side. In general, if
the ONU is placed at the subscriber premises, it is called optical network terminal (ONT). The
term network interface unit (NIU) has also been used in some cases [22]. ONU includes the
components for the electro-optical conversion and signal processing. It interfaces the optical
domain of the PON with the electrical signal format such as Ethernet used at the subscriber [23].
Twisted pair copper wires, coaxial cable, wireless or optical fiber links can be used to connect the
user equipment with the ONU.
Depending on the requirement or characteristics of the end user, the ONU can differ [24]. The
single-family unit (SFU) ONU is the most common one. The operator may own and control the
SFU, or it can be considered a customer premises equipment (CPE) which is owned by the
subscriber. In addition, the subscriber covers the power consumption cost of the SFU. In case of
fiber to the building, the multi dwelling unit (MDU) ONU is used to serve multiple residences in
a single building. The overall cost of the MDU including power consumption are covered by the
operator. There are also special ONUs for business and offices, namely multitenant unit (MTU)
and single business unit (SBU), respectively.

Optical Distribution Network (ODN)

ODN includes all the optical components between the ONU and OLT. It includes the feeder fiber
(FF), the remote node (RN), the distribution fiber (DF), and the drop fiber (DrF) which are all
shown in Figure 1-7. The FF links the OLT with the first optical distribution point, or the first RN.
The fiber connections between the multiple levels of the cascaded RNs are called distribution
fibers. The drop fiber connects the last access network RN with the ONU or the ONT.

Remote Node (RN)

The main application of the RN is to multiplex and demultiplex the US and DS signals,
respectively. The specification of the device used depends on the type of PON. In TDM-PON, a
power splitter is used to broadcast the DS signal and combine the upstream signal. In case of
WDM-PON, a WDM filter, for example an array waveguide grating (AWG) is used to mux/demux
the US and DS signals. In practical systems, the access network may contain cascaded RNs for
larger number of connected subscribers.

10
 Introduction

Figure 1-7. Nomenclatures of components of a standard passive optical network (PON) access network: at the central office (CO)
the downstream (DS) transmitter and upstream (US) receiver are placed inside the optical line terminal (OLT). The DS signal is
distributed to the destination termination point by the remote node (RN) which consists of either a power splitter in case of TDM-
PON and TWDM-PON, and wavelength Mux/DeMux in case of WDM-PON. In practical cases, the PON may consist of cascaded
RNs. In general, the feeder fiber (FF) connects the CO and the first RN. The multiple cascaded RNs are linked by the distribution
fiber (DF). At the last termination point, which could be a curb (FTTC), at home (FTTH) or building (FTTB), the US transmission and
DS reception is performed by the optical network unit. The fiber connecting the last RN and the ONU is called drop fiber (DrF). If
the final termination point is the a home (FTTH), the ONU could also be called optical network terminal (ONT).

PON Generations
PONs have evolved through multiple generations. The driving forces behind the evolution is the
need to deliver high bandwidth at lower cost and energy consumption. However, whenever
there is a plan to advance to the next level, there are always stringent requirements. The new
emerging technique should co-exist with the predecessor techniques, referred to as seamless
integration [25]. Subscribers who will not upgrade should be able to continue with their service.
This is one of the main limitation of TDM-PON as seamless integration is not possible. In addition,
the new scheme must, as much as possible, reuse already existing ODN infrastructure [26].

In general, the PON generations can be split to three groups, Legacy PONs, next generation PON2
(NG-PON2) and beyond NG-PON2. The “Legacy” PON is used to refer to the generations of PON
which have already been standardized till date [27]. These are the broadband PON (BPON),
gigabit PON (GPON) , Ethernet PON (EPON), Gigabit EPON, and XG-PON (10G-PON). XG-PON has
two phases XG-PON1 and XG-PON2, however only the first phase has been standardized. In the
nomenclatures of the international telecommunication union (ITU-T), XG-PON is a subclass of
NG-PON1. As NG-PON1 uses the same ODN and TDM scheme, it is considered as a Legacy PON.
On the other hand, NG-PON2 was planned to use disruptive techniques to enhance the access
capacity. In the end, NG-PON2 was standardized to mix TDM and WDM. Research is still going on
to come up with the next generation which should offer much improved performance with low
cost and energy consumption, which we will discuss under the title beyond NG-PON2.
Figure 1-8 shows the overall roadmap of the PON evolution [28, 29]. The x-axis shows the years
of the standardization of the PONs and the y-axis indicates the standardizing body, institute of
electrical and electronics engineers (IEEE) or ITU-T. This subsection goes through these PONs. It
discuss briefly their specifications including: the multiplexing scheme, protocol, US and DS line
rates, planned reach, power budget and splitting ratio. The type of optical transmitters and
receivers they use and their wavelength allocation are also introduced.

11
Introduction

Figure 1-8. Road map of generations of passive optical networks (PON) through the years from around 1995 50 2015 and beyond.
The y-axis shows the two standardization bodies: the IEEE and the ITU-T/FSAN. BPON: broadband PON, GPON: gigabit PON, EPON:
Ethernet PON, GE-PON: gigabit Ethernet PON, NG-PON1: 1st group of next generation PON, XG-PON: 10 Gbit/s PON (Latin X for
10), 10GE (XE)-PON: 10 Gbit/s Ethernet PON, NG-PON2: 2nd group of next generation PON [22]

1.2.4.1 Legacy PON

BPON
Broadband PON (BPON) is the first generation of PONs, which was internationally standardized
by ITU-T. It used asynchronous transfer mode PON (ATM) protocol, which was developed by the
full service access network (FSAN) and British telecom (BT). BPON uses the DS and US wavelength
bands 1260-1360 and 1480-1580 nm, respectively. Such broad wavelength ranges, specifically
100 nm, was assigned so that low cost multi-mode Fabrey-Perot laser diodes (FP-LDs) can be
used at the ONU. The OLT uses distributed feedback (DFB) lasers as DS transmitters. The DS and
US line rates are 155.52-1244.16 and 155.52-622.08 Mbit/s. The maximum ODN power budget
classes range from 20 – 33 dB. The maximum reach is 20 km with a splitting ratio of 1:128. BPON
uses TDM technology share the fiber between the multiple subscribers. It employs ranging to
synchronize the time difference from multiple subscribers with different distances form the CO.
The final version of BPON included dynamic bandwidth allocation (DBA), where the OLT can
allocate bandwidth to users based on their current traffic. However, the ATM protocol was fading
out and Ethernet and internet protocol (IP) were growing faster. Therefore, one needed an
interface between the PON ATM and Ethernet or IP protocols used at the subscribers devices.
This additional interface increased the cost of the ONU. In addition, further increase in bandwidth
requirement forced the ITU-T to consider a newer scheme.

GPON
The gigabit capable PON (GPON) came in to play to simplify the evolution to the more successful
Ethernet and IP protocols. It also satisfies the higher bandwidth requirement at the user end. The
GPON encapsulation protocol was developed to accommodate the various protocols with
minimum effort and bandwidth efficiency [2, 22]. GPON uses the same wavelength plan, power
budget and reach as BPON. It also includes a second wavelength line for analog video broadcast
transmission. The GPON allows for 2.488 and 1.244 Gbit/s DS and US respectively. To support
these high line rates, the OLT uses high power DFB transmitter laser and avalanche photodiode
(APD) as a direct receiver. The APD receivers are chosen to have high sensitivity when receiving
TDM burst signal at such high line rates. Similarly, at the ONU DFB lasers and APD photodiodes
are used. The use of forward error correction (FEC) is also specified in the ITU-T GPON standards
for extended reach or to use relatively low cost transceivers.

EPON/GEPON and 10GEPON/XEPON

With further success of the Ethernet protocol in the local area networks (LANs) and metro area
networks (MANs), a new generation of PON was studied by IEEE. The Ethernet in the first mile

12
 Introduction

(EFM) study group established by IEEE proposed the Ethernet PON (EPON), which utilizes the
Ethernet protocol in the full optical access network.
EPON provides symmetric 1.25 Gbit/s DS and US line rates. For the purpose of easy timing
recovery, 8B/10B codding scheme is implemented in EPON. This reduces the user bit rate to
symmetric 1 Gbit/s DS and US. EPON uses high power DFBs and APDs at the OLT, similar to that
of GPON. The ONU still consists of FP-LD or DFB lasers.
Further development to 10 Gbit/s was provided by Gigabit EPON (GEPON or 10GEPON). The
GEPON maintained similar specifications so that users can enjoy backward compatibility with
EPON. The proposed power budget ranges from 20 to 29 dBm. In GEPON, the OLT is designed
with externally modulated laser (EML) transmitter. The receiver is an APD with an option of
semiconductor optical amplifiers (SOAs) as preamplifier for better power budget and reach. At
the ONU, directly modulated laser (DML) with APD or PIN (p-type-intrinsic-n-type) photodiode
are deployed. US wavelength rage is 1260-1280 nm and DS range is 1575-1580 nm. Compared to
the previous generations, narrower US and DS bands are allocated because now single mode DFB
lasers instead of multi-mode FP-LDs are deployed. This allows for narrow spacing and effective
utilization of the optical spectrum to accommodate several PONs and subscribers. GEPON also
uses an improved codding scheme, the 64B/66B, and FEC to improve the reach and bandwidth
efficiency.

NG-PON1
Based on the nomenclature used in ITU-T standardization, next generation PON (NG-PON) refers
to the PON generations beyond GPON [13, 14]. NG-PONs have two phases, the NG-PON1 and
NG-PON2. NG-PON1 was planned to maximize the line rate per user, reach and splitting ratio
while keeping the same ODN as BPON and GPON. Thus, it is just an evolutionary progress of the
G-PON and its standardization started in 2010. NG-PON1 is split to two subclasses: XG-PON 1 and
2. XG-PON pushes the limit of TDM-PON scheme to the limit and delivers line rates of 10 Gbit/s
in the downstream direction. In the US, 2.5 or 10 Gbit/s are planned for XG-PON1 and XG-PON2,
respectively. The main limitation of XG-PONs is the availability of TDM burst mode receivers at
bit rates of 5 Gbit/s and above [30].
ITU-T recommends a minimum of 1:62 splitting ratio and maximum reach of 60 km. The allocated
US and DS wavelength range for XG-PON is shown in Figure 1-9. For the requirement of the back-
compatibility, the GPON and EPON US wavelength range is reduced to 1290 – 1330 nm and
further narrowed down to 1300-1320 nm. At the CO, a WDM filter is used to distinguish the G-
PON and the XG-PONs. It is advised that the WDM filters are installed with the G-PON installation
than with the XG-PON upgrade so that the evolution is done without service outage for the G-
PON subscribers [25].

Considering the high line rate, reach and split ratio expected in XG-PON, only single longitudinal
mode (SLM) lasers are required at both the ONU and OLT [31]. Thus the OLT has high output
power EML with APD. Depending on the planned power budget, SOA can be inserted as pre-
amplifier in front of the APD. The ONU has directly modulated DFB with PIN photodiode or APD
for longer reach.

13
Introduction

Figure 1-9. Wavelength allocation of the multiple generations of PONs: the x-axis shows the wavelength in nm with O, E, S, C and
L-bands. The y-axis shows the different generations of PONs, the green boxes showing the IEEE standards whereas the yellow
boxes represent the ITU-T FSAN standards. GEPON, BPON and GPON upstream (US) and downstream (DS) wavelength bands are
the same and are indicated by the US and DS boxes. For the XEPON and XG PON1, the US and DS bands are relatively narrower
as shown by the blue and red boxes. To accommodate XEPON and XG PON, the US band of the GPON and EPON are narrowed
down as indicated by the blue patter boxes. In case of NG PON2, the TWDM-PON US and DS wavelength band are indicated by
the orange and red boxes, respectively. The PtP WDM-PON overlay is assigned the expanded spectrum.

1.2.4.2 NG-PON2
Initial publications on the proposed techniques for NG-PON2 suggested that it would bring in
revolutionary technologies by making legacy coexistence non-mandatory [25]. Initial
investigations envisioned shifting from TDM to WDM and dense WDM (DWDM) or adopt
relatively new access multiplexing techniques such as OFDM, SCM or other hybrid technologies
[27]. Even if TDM was well developed and relatively inexpensive, its bandwidth limitation and
security issues made it not the ideal future-proof solution. Therefore, NG-PON2 was expected to
be highly scalable, flexible, reliable and efficient with respect to bandwidth and power [32].
However, considering the ongoing vast installation of legacy PONs, specifically BPONs and G/E-
PONs, operators were not willing to rule out back compatibility. They still want to recover their
investment on the power splitter based TDM infrastructure, which may need to be replaced if
they decide to adopt the promising WDM technology. Therefore, in 2012, it was decided by FSAN
that NG-PON2 should use the ODN of the legacy PONs. In addition, the same subscriber legacy
PON transceivers must also be used by the NG-PON2. This also means that NG-PON2 must cope
with the power budget class implemented for the previous generations. In the back
compatibility, the wavelength allocation must also be carefully designed so that NG-PON2 does
not either linearly or nonlinearly affect the legacy PONs.
Extra features envisioned for NG-PON2 included the integration of extension boxes or boosters
[33]. Extension boxes are used to compensate for a large attenuation at the power splitter when
having large splitting ratio. This means that, less number of COs can be used to serve relatively
large number of users. This helps to reduce the cost of the access network [34, 35]. Further, NG-
PON2 must facilitate the convergence of multiple access applications including mobile
backhauling, fronthaul and other newly emerging applications [26, 36]. This has been discussed
as the fixed mobile convergence (FMC) [37].
NG-PON2 installation is planned in two scenarios. One is the deployment to areas where the
legacy PONs have already been installed, a scenario called PON brownfield. In this case, the
migration should consider co-existence with minimum service interruption (exchanging or
placing new components such as WDM filters) and keeping all legacy services. The second
scenario is the PON greenfield where the installation takes places where no PON has been
installed before. Greenfield scenario includes new residential or business developments areas
[13].

14
 Introduction

With all these requirements in mind, FSAN NG-PON2 standardization recommends the hybrid
TWDM-PON for the residential users with PtP WDM-PON overlay for specific applications.
TWDM-PON Wavelength allocation is presented in Figure 1-9 [36, 38]. Both TWDM-PON and PtP
WDM-PON are discussed in the next paragraphs.

TWDM-PON
For the residential based optical access network, NG-PON2 combines WDM with TDM in the
hybrid TWDM concept [28, 38]. This enables the exploitation of the existing power splitter based
infrastructure of legacy PONs without extra investment. WDM is included to upgrade the data
rate per user by using additional wavelengths channels. Per user data rate is then the rate
delivered by the TDM system multiplied by the number of wavelength channels. With such a
scheme, the targets are DS 40 Gbit/s and US 10 Gbit/s transmission with expected reach of 40 km
with a differential reach of 40 km. If needed, the network should also support 60 – 100 km by
using reach-extending boxes. Of course, the reach extenders are preferentially placed either as
a booster at the OLT or a pre-amplifiers at the ONU to keep the ODN passive. The system should
still be kept transparent such that the presence of the extenders should not disrupt the service.
NG-PON2 could be designed to operate with the existing 1:16 to 1:128 splitting ratio legacy PONs.
However, it should support at least 1:256 splitting ratio.
To lower the cost of wavelength specific transceivers, FSAN proposes the use of colorless ONUs.
This reduces the inventory expenditure and simplifies wavelength management [13]. The initial
colorless solutions deploy tunable transmitters and receivers at the ONUs and OLTs [39, 40]. The
tuning can be done by coordination between the OLT and the ONU or independently by each
[12]. However, the tuning speed and mechanism are still in study phase.
Figure 1-10 shows the general architecture of TWDM-PON. The OLT has four separate 10 Gbit/s
DS transmitters and four separate 2.5 Gbit/s burst mode receivers. The colorless transmitters at
the OLT could be narrow band tunable DFB and DBR laser [41]. At the ONU, a tunable direct
detection receiver detects only a single wavelength, which is carrying a TDM signal. The ONU
tunable filter could be an AWG or thermally tunable thin film filters, which have different
insertion loss.

Figure 1-10. Time wavelength division multiplexed (TWDM) passive optical network (PON) architecture. This hybrid architecture
uses four TDM wavelength lines for a downstream (DS) transmission. At the subscribers’ optical network unit (ONU), a tunable
receiver is used to receive one of the four TDM wavelengths at a time.

This insertion loss is critical as it influences the power budget. For US transmission, the ONU can
also use colorless tunable sources. For higher power budget and reach extension, bidirectional

15
Introduction

optical amplifiers can be installed at the OLT [42, 43]. The OAs boos the DS signal and pre-amplify
the US carrier. Different architectures to include SOAs in NG-PON2 have are being [44].

WDM-PON in FTTB/O and FTTA

WDM-PON can offer high bandwidth with high security and flexibility by establishing virtually
point-to-point link [45]. Thus, WDM-PON within the NG-PON2 is nominated for specific high
capacity applications such as fiber to the business/office (FTTB/O), mobile backhaul and the
newly emerging mobile fronthaul (MFH). MFH is also called fiber to the antennae (FTTA). As it is
a relatively novel concept, the MFH/FTTA is discussed in the next paragraphs.

The spreading of high performance smart devices and a surge in bandwidth hungry services are
putting pressure on the mobile network. The future mobile network is evolving to dense base
station distribution with planned macro cells and unplanned overlaid small (micro, pico and
femto) cells. When considering small densified cells, the reduction of power consumption, size,
capital expenditure (CAPEX) and operational expenditure (OPEX) is a must. To achieve this, the
cloud radio access network (C-RAN) concept proposes to split the traditional base station
Figure 1-11(a) into the remote radio head (RRH) and the base band unit (BBU), as shown in
Figure 1-11(b) [46]. The RRH now includes the antennae, amplifiers and analog to digital/digital
to analog (AD/DA) convertors. The BBUs consists of baseband and higher layer computations
through ASICs, FPGAs and general purpose processors. The BBUs are then centralized to the
central office. This removes the major power consumer, which is the BBU, from the base station.
Aside from the reduced power consumption of the base station, it also minimizes its physical
size. The centralization of the BBUs has further application. It enables flexible sharing of
processing resources by coordinating the BBU which are now at the same location. It also
simplifies the implementation of coordinated multi-point (CoMP) which is an LTE-advanced
feature. This optimizes interference mitigation and beam forming. Efficient maintenance (OPEX)
and hardware pooling gain (less investment, CAPEX) are also other advantages.
For the optimum operation of the C-RAN, the link between the RRH and BBU should be able to
deliver high bandwidth. Technologies including DRoF (digital radio over fiber), WDM-PON,
microwave, and dark fiber are being investigated to carry the high speed digital signal between
the RRH and BBU [47]. The common public radio interface (CPRI) protocol chosen for the MFH
has stringent specifications, such as small latency, near zero jitter, and near zero BER at bit rates
up to 10 Gbit/s over 40 km reach. For example, TDM multiplexing and demultiplexing are time
consuming process which may not fit the latency requirement of C-RAN [35]. On the other hand,
WDM-PON is considered to be the best alternative as shown in Figure 1-11(b). However, WDM
technology is relatively expensive when considering the need for wavelength specific
transmitters and receivers. Especially in the case of MFH where a large number of antennae sites
must be interconnected, wavelength specific transceivers are not the way to proceed. The
development of wavelength-agnostic or colorless transceivers is a key [48].

16
 Introduction

Figure 1-11. Evolution of mobile base station (BS): in the traditional BS shown in (a), the remote radio head (RRH) and the base
band unit (BBU) are positioned in a rack near the antennae tower. The antennae and the RRUs are connected by a coaxial cable,
which highly limit the link performance. Recently, the cloud radio access network (C-RAN) is being implemented. In C-RAN, the
traditional BS is slit where the RRH is move close to the antennae to minimize the length of copper cable used. The BBUs are then
moved to the central office, which simplifies the BBU coordination. However, one now needs a high speed link between the BBUs
at the CO and the RRHs, and this new section of the mobile network is called mobile fronthaul.

1.2.4.3 Beyond NG-PON2/XE-PON

The existing roadmap of the two standardization bodies ITU-T and the IEEE has already run out,
see Figure 1-8. Both organizations are planning for the next step. IEEE has established a work
group to study next generation EPON (NG-EPON) [49, 50], while ITU-T FSAN is also discussing the
way forward [51]. NG-EPON targets capability to support 40 Gbit/s in both US and DS with
scalability to 100 Gbit/s. In case of ITU-T, NG-PON2 already has a potential to support 40 Gbit/s
and this can be scaled up by adding more wavelength channels in the hybrid TWDM-PON.
As it stands, residential FTTH is not the main requirement driver as already standardized GE-PON,
XEPON, XGPON and NGPON2 have the capability to satisfy the demand, at least in the short term.
The legacy PONs are still on the market and NG-PON2 just got standardized and is being deployed
[52, 53]. However, there are specific applications with stringent bandwidth and quality
requirements. As stated in [54], there is no one-size-fit-all solution in an access network. Each
application, mainly FTTB/O, FTTA (MFH) and mobile backhauling need respective optimum
solutions.
During the standardization process of NG-PON2, initial indications suggested it would be a
disruptive technology targeting higher bit rate, longer reach, and scalability. In the end, the
decisive point was the issue of cost and the push by the operators who still have to recover the
investment on the legacy PONs. Cost and back-compatibility will continue to influence the
decision-making in the future, especially in the brownfield scenario.
Research wise, multiple technologies are suggested for the next generation access. The strait
forward step is to improve the performance of the legacy TDM-PON technology for cost effective
upgrade. 25 to 40 Gbit/s TDM transmission with reasonable improvement in power budget and
reach have been compared in [55]. Advanced modulation formats such as multi-level pulse
amplitude modulation (PAM) and electrical or optical duo binary (EDB, ODB) are used with
electronic dispersion compensation (EDC) and digital signal processing (DSP). APDs are also
utilized for improved sensitivity and longer reach. In case of WDM-PONs, one issue is the
compatibility with legacy PONs, mainly with respect to sharing the available wavelength band.
One way to solve this is to use narrower grid, in the range of 50, 25, 12.5 and 6.25 GHz. To deploy

17
Introduction

narrow band WDM-PON while keeping the power splitter based infrastructure, one needs to find
a way for the ONU receiver to choose the wavelength dedicated for it. An attractive solution is
to use cost efficient coherent receiver, where an inexpensive local oscillator laser with the same
wavelength is placed at the ONU [56, 57]. When the DS multi-wavelength signal beats with the
local oscillator on the photodiodes, only the signal on the dedicated wavelength will be down
converted to the base band.
There are also discussions on the utilization of spatial division multiplexed (SDM) WDM-PONs
with multicore fiber [58] or few mode fiber [59] as a feeder fiber. Further, as the cost of DSP is
minimizing, advanced modulation formats with higher spectral efficiency and performance could
be in the road map of the future access network [60].

State-of-the-Art: Colorless Upstream Transmitters for WDM-PON

As discussed in the previous subsection, WDM-PON is still the forerunner solution to bring by
future-proof solution to the ever-increasing bandwidth need in the last mile. However, the cost
and inventory issue must be significantly reduced. This is achieved by developing mass-
producible colorless transmitters. Colorless transmitters are devices with broad optical operation
band. However, their specific operation wavelength is determined by external factor or device.
Colorless transmitters are mainly deployed for US transmission, as that is where the main cost
limitation is required. In this subsection, we discuss state-of-the-art colorless solutions. These
include tunable lasers, ASE spectrum-sliced sources, externally seeded and injection locked
devices, wavelength reuse systems and self-seeded source.

1.2.5.1 Tunable Lasers

Tunable lasers (TLs) are one of the colorless US transmitters solutions for WDM-PON [35]. Mass
produced identical devices can be installed at all subscriber ONUs. The operation wavelength of
the ONU is set during the installation of the TLs.
Tunable lasers have been studied and used in long haul and metro WDM networks to achieve
dynamic wavelength reconfiguration capability [61]. However, for TLs to be selected in the access
network, their cost should be further lowered. In addition, simpler tuning control and efficiency
would be required. Tunable laser candidates include distributed Bragg reflector (DBR) and DFB
laser diodes, vertical cavity surface emitting lasers (VCSELs), and external cavity lasers (ECL), see
Figure 1-12. The selection of one of these depends on the simplicity of the tuning method, the
tuning range (covering the C/L band) and the frequency stability. The tuning methods could be
electrical current controlled (mainly varieties of DBRs), thermally controlled (the varieties of DFB)
and mechanically controlled (mainly the VCSELs and the ECL).
There have been several experimental demonstration of tunable lasers in NG-PON2. 1 Gbit/s
Ethernet PON field demonstration using a special DBR diode operating up to temperatures of
90 °C while transmitting up 10 Gbit/s has been presented in [62, 63]. In these works, a Mach-
Zehnder modulator (MZM) is monolithically integrated with the tunable DBR laser. Tunable DFB,
with a tuning range compatible with the 50 GHz ITU grid, integrated with electro absorption
modulator (EAM) has demonstrated 10 Gbit/s transmission [64]. In cases where tunablity beyond
a single optical band is required, combining multiple DFBs with different wavelength bands has
been suggested. A multi section DFB covering both C and L band with continues tunablity and
1.25 Gbit/s direct modulation was presented in [65]. Because of their low cost, tunable VCSELs
are also another option [66].

18
 Introduction

Figure 1-12. Tunable laser (TL) based colorless optical network unit (ONU) for WDM-PON: even if identical devices are installed at
all the users, the specific operation wavelength of each ONU is set by tuning the TLs during installation. Common TLs are DBR
(distributed Bragg grating) DFB (distributed feedback) laser diodes and VCSELs (vertical surface emitting lasers), The upstream
(US) signal could be a directly modulated onto the TL. For higher reach and speed, external modulators such as electro absorption
modulator (EAM) and Mach-Zehnder Modulators (MZM) could be integrated with the TL.

They can offer 10 Gbit/s direct modulation, where the performance is limited by their strong
chirp [67]. For higher bit rates beyond 20 Gbit/s, EAM can be integrated with the VCSELs [68].
General comparison of the different tunable sources pros and cons are discussed in [67].
The latest ITU-T standard for NG-PON2 TWDM-PON uses only 4 to 8 distinctive wavelengths with
the 100 GHz grid. However, this will be increasing as one evolves to fully WDM or DWDM-PONs.
Therefore, research will continue to obtain continuous tunability over multiple wavelength
bands. The cost of tunable sources is still high for general residential use. However, they could
soon be deployed for high-end access networks to business and mobile front/backhauling.

1.2.5.2 Spectrum-Sliced Transmitters

The idea behind spectrum-sliced transmitters is to use simple and low costs colorless WDM
transmitters. In this scheme, relatively inexpensive broadband light sources (BLS) are placed at
each ONU. They could be amplified spontaneous emission (ASE) sources or multimode FPLDs.
The US signal is then directly modulated onto the broadband light. At the RN, the WDM filter
slices the ASE and passes only sub bands from each ONU at their assigned band, as shown in
Figure 1-13.

Figure 1-13. WDM-PON colorless upstream transmitters based on tunable lasers (TL) or broadband light sources (BLS). In case of
TL, the upstream data can be directly modulated. However, to avoid chirp and bandwidth limitation, external modulators such as
electro-absorption modulators (EAM) or Mach-Zehnder Modulators can be integrated with the TL. Directly modulated BLS such
as light emitting diodes (LEDs), superluminecent LEDs (SLEDs), and reflective semiconductor optical amplifiers (RSOAs) also simple
and a very low cost transmitters.

19
Introduction

Examples of BLSs include light emitting diodes (LEDs) [69-72], superluminecent LEDs (SLEDs) [73],
reflective semiconductor optical amplifiers (RSOAs) [74], and FP-LDs [75] as shown in Figure 1-13.
Despite their simplicity and inexpensiveness, the reach and speed of such transmitters is limited.
One reason is the high attenuation due to filtering, which reduces the power budget by up to
10 dB. The large bandwidth of the sliced ASE is highly sensitive to chromatic dispersion. In
general, ASE sources have high relative intensity noise (RIN). Further, the ASE-ASE beat noise
during reception by a photodiode incurs extra amplitude noise. The bit rates of such schemes are
limited to below 1 GBit/s therefore has not attracted noticeable attention.

1.2.5.3 Externally Seeded and Injection Locked Transmitters

The externally seeded and injection locked transmitters were suggested as an alternative to the
spectrum-sliced schemes to improve the bit rate and transmission distance. The idea is to use
identical optically broadband and faster reflective modulators (RMs) at all ONUs. The operating
wavelength of each ONU is determined by the wavelength of the CW (continuous wave) light
sent from the CO, see Figure 1-14. At thse CO, a BLS sends a broad spectrum light. At the RN, a
WDM filter distributes the sub bands to the respective ONUs. Then, the RMs at the ONUs encode
the US signal onto this light and reflect it back to the CO as shown in Figure 1-14. The BLS could
be LEDs, SLEDs or EDFAs. Multimode FP-LD lasers could also be used, where each mode is used
for one ONU. If more quality and capacity is needed a bank of lasers, each for one ONU could
also be deployed at the CO. At the ONU, the possible RM options are RSOAs, reflective MZMs
and EAMs, or FP-LDs. Of course, MZMs and EAMs offer wider modulation bandwidth compared
to RSOAs and FP-LDs. The high insertion loss of MZMs and EAMs can be mitigated by integration
with SOAs or RSOAs. Figure 1-14 summarizes the possible combinations of BLSs and reflective
modulators.

There have been several experimental investigations of externally seeded colorless WDM-PONs.
By using EDFA as an ASE source at the CO, a MZM at the ONU was able to encode line rate of
622 Mbit/s in the upstream direction [76]. One limitation of this scheme is the increased intensity
noise after filtering a broadband ASE. One can use saturated SOA [77, 78] or RSOA [79] at the
ONU which are used to nonlinearly suppress the intensity noise of the sliced ASE from an EDFA.
Another architecture is to use SLED instead of EDFA at the CO, and RSOA instead of MZM at the
ONU [80]. In this scheme, the RSOA simultaneously amplifies and directly encodes the US data
onto the seeding ASE slice. Such scheme has shown to support up to 5 Gbit/s transmission [81].
A comparison of ASE seeded RSOAs and FP-LDs is presented in [82].
Significant leap in the achieved bit rate, performance and reach was demonstrated by integrating
fast external modulators (EAM, MZM) with SOAs or RSOAs. The integration of EAMs and SOAs or
RSOAs offers a better performance for two reasons. One, EAMS offer larger electrical bandwidth
and extinction ratio than directly modulated SOAs or RSOAs. Two, the SOA could purely be used
for amplification thus boosting the power budget by compensating for the insertion loss of the
EAM. Further, if operated in the nonlinear region, the SOAs and RSOAs can be used to suppress
the amplitude noise on the seeding ASE slice. Thus far, integration of EAM with RSOA have shown
10 Gbit/s transmission [83, 84]. A reflective EAM integrated with one or two SOAs (SOA-REAM)
offers bit rates of 10 Gbit/s with 25 km transmission [85, 86].
The other type of colorless upstream sources are FP-LDs. The FP-LDs are placed at the ONUs as
directly modulated upstream transmitters. A spectrally sliced ASE from the CO seeds the FP-LDs.
If the mode spacing of the FP-LDs and the WDM filter overlap, then the FP-LDs emission will be
locked to the seeding wavelength [87].

20
 Introduction

Figure 1-14. WDM-PON topology with externally seeded or injection locked colorless upstream transmitters. A broadband light
sources (BLSs), such as light emitting diodes (LEDs), superluminecent LEDs (SLEDs), Erbium doped fiber amplifiers (EDFAs), Fabrey-
Perot laser diodes (FPLD) and a laser array, at the central office produce the seeding light. in the case of LEDs, SLEDs and EDFAs,
it is an amplifier spontaneous emission, filtered by the WDM filter that is used for upstream tramission. The laser array provides
distinct laser lines. At the optical network unit (ONU), a reflective modulator, such as reflective semiconductor optical amplifiers
(RSOAs), SOA integrated with elecro-absorption modulator (EAM) encode the upstream data onto the seeding laser. an injection
locked FP-LD, wavelength locked to the seeding light can also be used as an upstream transmitter.

The seeding sources discussed so far were based on ASE sources. Substantial improvement in bit
rate and transmission distance when using bank of lasers instead of a sliced ASE. In this case,
each laser line is dedicated to one ONU. RSOAs, SOA-REAM and FP-LDs are used as reflective US
transmitters Electrical bandwidth limited RSOA seeded with a CW laser line have demonstrated
10 Gbit/s transmission without any optical or electrical signal processing [88, 89]. Different types
of amplitude modulation formats were tested for improved performance, for example return to
zero (RZ) [90], 4 PAM with polar and nonpolar formats [91, 92]. To mitigate the impact of chirp
and chromatic dispersion on the modulation bandwidth, offset filtering using delay line
interferometers at the receiver were used [93-95]. Beyond this, a significant bit rate boost was
obtained by using discrete multi-tone (DMT) modulation format with adaptive bit- and power-
loading. 25 GBit/s signal was squeezed into a 1 GHz bandwidth RSOA through DMT while using
offset filtering [96]. With broader bandwidth RSOAs, bit rates up to 40 Gbit/s are also measured
[97, 98]. On the other hand, due to the larger electrical bandwidth of EAMs, SOA-REAM can
achieve 40 Gbit/s with a standard non-return-to-zero (NRZ) amplitude shift keying (ASK)
modulation and simple signal processing [99, 100].
The performance of injection-locked FP-LD is also boosted by using a laser line instead of ASE.
Different designs of FP-LDs have also been investigated such as weak resonant cavity (WRC) FP-
LDs. The performance of different injection schemes of WRC-FPLDs, such as sliced ASE, sliced
multimode WRC-FPLD or DFB laser array is discussed in [101]. With the implementation of
advanced modulation format, specifically 16QAM and OFDM, up to 42 Gbit/s has been recorded
with DFB seeded WRC-FPLD [102]. AS an alternative, mode locked laser comb generators can be
used instead of deploying laser array source at the CO [103].
Despite the relatively high bit rates offered by both an externally seeded and injection locked US
transmitters, they have some downsides. One is that fact that they need an extra source placed
at the CO. In order to cope with attenuation of the seeding light as it propagates downstream,
one needs high power source. The alternative is to integrate an amplifier, SOA, with the reflective
modulators. These technologies are cost-wise challenging to achieve. The technological

21
Introduction

challenge is that the Rayleigh scattering of the downstream travelling cw laser line will interfere
with the US signal, thus degrading the performance.

1.2.5.4 Wavelength Reuse Technique

In wavelength reuse scheme a DS signal carrier is reused for an US transmission. Thus, an optically
broadband colorless US transmitter can be installed at the ONU. The US transmission wavelength
is determined by the DS signal carrier wavelength. At the ONU, the modulated DS laser line is
split into two. A DS direct detector receives part of the DS signal. The other part is fed to the
reflective modulator to be used for US transmission, see Figure 1-15. This scheme highly lowers
the cost of the WDM-PON since no extra source is required. However, one needs an efficient way
to remove the DS signal from the carrier so that it does not interfere with the US signal. In case
of simple ASK modulation DS transmission, a saturated SOA at the ONU can suppress the DS
signal and encode the US signal [104]. Due to the higher gain saturation offered by RSOAs
compared to SOAs, RSOAs have been investigated more in the wavelength reuse scenario [105].
In addition, the data erasing and remodulation capacity of simple FP-LD for up to 10 Gbit/s US
transmission have also been investigated [106].
Another option is to use orthogonal modulation format for the US and DS in order to avoid any
interference between the US and DS signals. For example, in [107], frequency shift keying (FSK)
was used for DS and ASK for US. Downstream binary phase shift keying (BPSK) modulation and
ASK for US were used to demonstrate a symmetric 10 GBit/s bidirectional transmission [108].
Direct phase modulation capacity of SOAs for up to 25 Gbit/s has been measured in [109].

Figure 1-15. Wavelength reuse WDM-PON schematics: the upstream (US) signal is modulated onto the downstream (DS) carrier,
thus avoiding the need for extra source. At the optical network unit (ONU), a power splitter sends part of the DS signal to the DS
receiver (Rx) while the remaining part of the DS signal is sent to the US reflective modulator (RM). The RM could be a reflective
semiconductor optical amplifier (RSOA), a reflective electro absorption modulator (REAM), a combination of SOA and REAM (SOA-
REAM) or Fabrey-Perot laser diode (FP-LD).

One can also encode the DS signal on an intermediate frequency to prevent the spectral overlap
between the DS and US signals [110, 111]. At the RN, a common delay line interferometer can
be used to separate between the US signal subcarrier frequency and the carrier wavelength.
Then the carrier can be re-modulated by an RSOA at the ONU. Of course putting a dedicated
delay interferometer and two WDM filters increases the cost. In addition, the DS and US signals
could be separated by using different polarization and subcarrier frequency, with further reach
and power budget improvement by using low cost self-homodyne coherent receivers [112-114].
Another solution is to use advanced signal processing for better modulation suppression at the
ONU. Adaptive OFDM was used with directly modulated RSOA at the ONU in [115]. Coding and

22
 Introduction

equalization has also enabled up to 40 Gbit/s transmission in a wavelength reused WDM-PON

[116].

1.2.5.5 Self-Seeded Transmitters

Despite the high bit rates recorded by the externally seeded RSOAs, SOA-REAMs and FP-LDs, they
are still limited to high end users that can afford the cost. Therefore, the need for a solution that
can offer bit rates in the range of 10 Gbit/s without expensive external lasers is still high. This is
especially true in applications such as mobile backhaul and fronthaul, which will need to install
large quantity of WDM-PON ONUs. In such scenarios, self-seeded transmitters offer a promising
alternative. The idea behind self-seeded transmitters is to avoid an external seed source, and
instead use the light from the US transmitter itself as a seeding source. A feedback mechanism
to the ONU is established by placing a reflector in the ODN, so that part of the US transmission
is reflected back to the ONU to automatically lock the ONU’s transmission wavelength. The
reflector could be placed at either the OLT, RN or ONU as depicted in Figure 1-16. For example,
an RSOA can be placed at the ONU and fiber Bragg grating (FBG) at the output of the RSOA, as
shown by ONU 1 in Figure 1-16 (a). The FBG reflects back a filtered slice of the ASE from the
RSOA. The reflected ASE slice re-enters to the RSOA and is amplified dominantly locking the
emission wavelength of such ONUs. This implementation have demonstrated up to 2.5 GBit/s
transmission over 40 km standard single mode fiber (SSMF) [117-120]. Similar to this
architecture, tunable filters could be placed at the ONU in front of the FP-LDs instead of FBGs.
The tunable filter selects a single mode from the FP-LD output and a circulating loop is used to
couple back part of the selected mode back to the FP-LD [121]. This has demonstrated 2.5 GBit/s
transmission over 85 km SSMF.
In Figure 1-16 (a), ONU 2 shows an alternative implementation of colorless self-seeding by
placing the FBGs at the RN rather than at the ONU. Each ONU’s FP-LD is now connected to an
FBG, which has a reflection peak at the required operation wavelength. The FBG reflects back
part of the mode at its reflection peak, thus locking the FP-LD emission to this mode and
transmitting 1.25 GBit/s over 10 km [122]. However, these schemes do not fully simplify the
transmission system as the specific design and implementation of FBG or tunable filter and
additional circulators complicates the system and increases cost and loss. All these extra
components are needed for each individual ONU, which becomes an inventory issue as the
number of ONUs increases. A relatively simplified implementation of self-seeded colorless
transmitter based on RSOA was suggested in [123]. Here, a circulator loop after the WDM filter
at the RN provided the feedback for the self-seeding, as shown in Figure 1-16(b). Here a common
circulator loop is used for the entire ONUs. The wavelength selection was automatically done by
the WDM filter, so no additional FBG or tunable filter was needed. Despite the system
simplification, the insertion loss of the reflective path, which includes a circulator and an optical
bandpass filter, degraded the feedback power. Thus, the high amplitude noise of the seeding
filtered ASE allowed only 1.25 GBit/s.

23
Introduction

Figure 1-16. Self-seeded upstream transmitter implementations for WDM-PON: here, a reflector is placed in the optical
distribution network, either at the remote node (RN) or at the optical network unit (ONU). The optical source, which is either a
reflective semiconductor optical amplifier (RSOA) or a Fabrey-Perot laser diode (FP-LD), is positioned at the ONU. The frequency
selective reflector reflects back a sliced sub band of the output of ether the RSOA or FP-LD. The re-entering sub-band will lock the
emission wavelength of the sources. For instance, in (a) two implementations are depicted. In case of ONU 1, a fiber Bragg grating
(FBG) is placed at the output of the RSOA or FP-LD. The FBG reflects a specific sub band of the amplified spontaneous emission
(ASE) from the RSOA, which is dominantly amplified by the RSOA. In case of a FP-LD, the FBG reflects back part a selected mode
from the multimode emission of the FP-LD. Alternative option is to place the FBG at the RN, as it is for ONU 2 in figure (a). In (b),
a reflective path is established at the RN using a coupler and an optical circulator. The reflected wavelength is now chosen by the
WDM filter at the RN and the path just establishes a feedback mechanism..

1.3 Summary
Optical fiber based access networks are indispensable to fulfil the growing bandwidth
requirement at the last mile. Current access networks rely on the cost-efficient however, a
capacity limited and un-scalable TDM-PON. Nevertheless, a future-proof path would be to adopt
WDM technology, which is highly scalable and secure. However, the cost of wavelength specific
transceivers is not yet acceptable for access network. Since access network is a high volume
business, wavelength specific components also complicate the inventory. Therefore developing
cost-effective, high capacity colorless upstream transmitters is a necessity.

Cost-efficient, high-speed colorless transmitters are critical if one wants to take advantage of the
attractive capacity of WDM technology at an acceptable investment. With this respect, the state-
of-the-art colorless transmitter technologies summarized in this chapter lack the ability to
address both capacity and cost challenges simultaneously. For instance, tunable lasers offer the
highest speed and transmission distance yet they are not yet financially feasible. Similar
conclusion applies to the externally seeded or injection locked approaches, which require extra-
dedicated source. On the contrary, the relatively low-cost schemes, specifically the self-seeded
techniques offer a promising combination of simplicity and inexpensiveness. Their performance
is not yet adequate to satisfy the foreseeable growth in bit rate requirement in the range of
10 GBit/s.
This work investigates an RSOA fiber cavity laser (RSOA-FCL) – a further simplified self-seeded
scheme as proposed in [124]. RSOA-FCL tackles both capacity and cost issues concurrently. Unlike

24
 Introduction

tunable lasers, it requires only RSOAs, which are relatively inexpensive mass-producible devices.
Further, it avoids the need for any external source, which simplifies and lowers its cost compared
to externally seeded approaches.
RSOA-FCL utilizes a simple mirror at the RN to establish a feedback mechanism. A Fabrey-Perot
cavity is formed between the RN mirror and the RSOA reflective facet, and the cavity is
embedded in the PON distribution fiber. The RSOA acts as the gain medium of this laser, whereas
the WDM filter at the RN performs the automatic wavelength allocation. In case of RSOAs with
highly polarization dependent gain (HPDG), the system performance is degraded since the
distribution fiber birefringence alters the polarization of the reflected light. This issue has been
solved by integrating Fraday rotator (FR) and FR mirror (FRM) after the RSOA and the reflector,
respectively [125].
Sine RSOA is an important part of the self-seeded RSOA-FCL, chapter 2 is dedicated to introduce
the operating principle of SOAs and RSOAs. An RSOA mathematical model is implemented in
chapter 2 and further RSOA simulation results are given Appendix A. An illustrative theoretical
diagnosis of an RSOA-FCL is presented in chapter 3, which also includes experimental verification.
Chapter 4 qualitatively measures how an RSOA-FCL performs against alternative RSOA based
colorless transmitters. In chapter 5, advanced signal processing techniques are exploited to push
beyond 10 GBit/s target and test the maximum achievable bit rate with an RSOA-FCL and
alternative techniques.

25
2 Semiconductor Optical Amplifiers (SOAs) and
Reflective-SOAs (RSOAs)
The development of SOAs and RSOAs has been integral to several of the access network
technologies discussed in the preceding chapter. From legacy to future PON generations, SOAs
and RSOAs play an important role. Therefore, it is appropriate to introduce these devices in more
detail.
Briefly, SOAs are optoelectronic devices with a similar structure as Fabry-Perot semiconductor
lasers, however with antireflective facets on both longitudinal sides of the active region. The
RSOAs on the other hand, have only one reflective facet, see Figure 2-1. A semiconductor active
region provides the gain to amplify the input light. The energy needed for the amplification is
delivered in the form of an electrical current. In case of SOAs, the light is fed in and collected on
opposite sides of the active region whereas in RSOA, a single port is used for both input and
output, as shown in Figure 2-1.

Figure 2-1. Simple schematics of semiconductor optical amplifier (SOA) and reflective-SOA (RSOA): In case of an SOA, the input
light travels through the amplifying active medium and is collected on the opposite side. In case of an RSOA, the input light is
injected and collected through a single input/output facet. The opposite side is designed to have high reflectivity.

This chapter presents the fundamental backgrounds of SOAs and RSOAs. It starts with a brief
historical development of the two devices and discuss how RSOAs came to be vital components
in optical access communications. It continues with the discussion of specific applications of SOAs
and RSOAs in optical access networks. The final subsection goes deeper into the optoelectronic
process behind the functionalities of SOAs and RSOAs. The material presented here is based on
multiple semiconductor optical device textbooks and thesis [126-131].

27
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

2.1 Background of SOAs and RSOAs

Major investigation and integration of SOAs into optical fiber communication was kindled by
their development at the telecommunication wavelength of 1.5 µm in the 1980s [132-134].
Optical amplifiers were attractive to replace the inefficient regenerators used in the long distance
optical fiber transmission systems at that time. SOAs were then deployed as pre-amplifiers and
in-line amplifiers. During this time, the challenge was the high polarization dependent gain
(HPDG) of the SOAs, which showed several dBs of difference between the transverse electric (TE)
and transverse magnetic (TM) polarizations. One suggested solution was to amplify the two
polarization separately. This was accomplished by using a single HPDG SOA, which is fitted with
a 45° FRM on one side, thus having a similar structure as an RSOA [135, 136]. When light is
injected into the HPDG SOA, only one polarization is amplified, for example the TE. As the light
exits the SOA on the opposite side, an FRM simultaneously rotates the polarization by 90° and
reflects the light back into the SOA. Due to the polarization rotation, now the originally TM mode
(rotated to TE by the FRM) is amplified. In the end, both polarizations experience the mean of
the TE and TM gains. Such SOA-FRM architecture resulted in the first conceptual development
of RSOAs.
By the end of the 1980s, there was already an idea to use an optical fiber in access networks to
deliver digitized video to subscribers [137]. However, in order to consider optical fiber
deployment in an access network, there needs to be a significant cost reduction to an acceptable
level. For instance, avoiding any expensive light sources at the subscribers end is attractive
option. Thus, an initial solution was to place reflective modulators at the subscriber ONU of an
access network. For example, Lithium-Niobate configured as reflective modulators were
investigated [138, 139]. However, these were not permanent solutions due to the high insertion
loss of such devices, which degraded the access network power budget. Therefore, a device that
could simultaneously function as an amplifier and modulator was needed. In 1996, an RSOA, then
called single-port laser amplifier modulator (SLAM) was used as a reflective modulator in WDM-
PON architecture to achieve 100 Mbit/s transmission [140]. The SLAM was a modified version of
the traveling-wave SOA built from bulk InGaAsP operating at 1.5 µm. One facet of the SOA was
re-cleaved for higher reflectivity, similar to the RSOA in Figure 2-1. This work paved the way for
further study on RSOAs as attractive WDM-PON reflective modulators. The research topics
concentrated on enhancing the electrical modulation bandwidth, gain and optical bandwidth
while further lowering their production cost [141, 142].

Optical communication systems have been relaying on both SOAs and RSOAs for multiple
applications. These applications could be split into two based on the two operation regimes of
SOAs and RSOAs, namely linear and nonlinear. The operation regime mainly depends on the
magnitude of the input optical power. When SOAs and RSOAs are used as optical amplifiers, they
are operated in their linear regime. On the other hand, their nonlinear behavior have been
exploited in all-optical signal processing. These include regenerative wavelength conversion, all-
optical demultiplexing, and many more as summarized in [143-145]. In case of RSOAs, they
exhibit strong nonlinear gain saturation due to the bi-directional propagation of light. Thus, they
have been widely used to nonlinearly suppress amplitude noise when using low quality, low-cost
sources in access networks.
General advantage of SOAs and RSOAs is their small footprint, allowing for their integration with
other optoelectronic devices (eg. Lasers and modulators) and the possibility for mass production

28
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

[146]. Their electro-optical efficiency has been improving over the past 20 or more years; their
threshold current have been significantly reduced, and the coupling efficiency to standard single
mode fiber (SSMF) has seen huge improvement. They also offer more than 40 nm optical
bandwidth. However, they also face some challenges. SOAs and RSOAs have relatively large noise
figure (>6 dB) when used as inline or preamplifiers. In addition, the polarization dependence of
the gain is also another obstacle.

2.2 Application of SOAs and RSOAs in Optical Access Networks

The required characteristics of SOAs and RSOAs depends on the specifics of the target
application. The applications of SOAs and RSOAs in the access networks can be summarized into
three categories: 1) amplification, 2) directly modulated transmission and 3) noise suppression
[34, 143].

Boosters, Inline and Pre-amplifiers

SOAs are used as booster, inline and pre-amplifier mainly for reach extension and splitting ratio
improvement in PONs [33]. For example, in TDM-PONs, achieving 60 km reach when the
downstream (DS) signal power has to be split for 128 subscribers, the network would require a
high power budget beyond 30 dB. In this case, SOAs are interesting solutions considering their
broadband gain and cost-efficiency.
SOAs can be incorporated in an access network at three typical positions. First, as a booster, SOA
can be placed just after the DS transmitter at an OLT [44] or also after the upstream (US)
transmitter at an ONU. Second, an SOA can also be positioned within an ODN at a remote node
(RN), to be used as an in-line amplifier [147]. Third, an SOA can be deployed as a pre-amplifier;
just before a direct detection photodiode at the ONU [34, 42, 44], or before the receiver at the
OLT [44]. In all these applications, the SOAs or RSOAs must offer the following qualities:
 Low noise figure (NF): NF is the ratio of the signal-to-noise ratio (SNR) of the input
optical signal to the SNR of the output amplified optical signal. In other words, NF
quantizes how much noise an amplifier adds to the optical signal during amplification.
The strong amplified spontaneous emission (ASE) in SOAs and RSOAs degrade their NF
performance. The relatively high NF of SOAs, >6 dB, is one of the drawbacks that
hindered their deployment in long haul transmission. The research is still going on to
lower the NF, and 3.8 dB has been achieved in [148]. Low NF is specifically important
when the SOA is used as a pre-amplifier.
 High saturation power: Saturation power quantizes the power at which the SOA inters
its nonlinear region. High saturation power is necessary especially when an SOA is used
as a booster. In case of multi-wavelength operation where the total input power is the
summation of all channels, the saturation power determines if the SOA can manage
to stay linear. SOA with saturation power of 16.5 dBm with < 6 dB NF was
demonstrated in [149]. 11 dB saturation power for RSOAs was achieved in [150].
 A broad input power dynamic range (IPDR): Practically, the performance of SOAs
highly depends on the input power. It is already discussed that high input power will
push the SOA into nonlinear regime, which must be avoided when they are used as
amplifiers. On the other extreme, to low input power means that the output is
dominated by ASE noise. Therefore, the input power must be within these two limits
to have an acceptable SOA performance. This range is now called IPDR, which

29
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

quantifies the input power range between which a predefined minimum performance
quality is obtained. The nonlinearity point can also be determined by the saturation
power. For example, quantum dot (QD) SOAs have higher saturation power and
broader IPDR compared to bulk SOAs [144]. Broader IPDR is required for in-line
amplifiers.
 Low polarization dependent gain (PDG): PDG measures the difference between the TE
and TM polarization gain inside an SOA or RSOA. In optical fiber where the polarization
of the light varies due to birefringence, inline or preamplifier SOAs should have PDG
maintained below 1 dB [151]. On the contrary, in case of booster SOAs, HPDG is a
bonus to avoid noise on unused polarization while amplifying the single polarization
output of the transmitter source.
 Integration with other devises: Boosters or pre-amplifier SOAs are commonly
integrated with transmitter sources and direct receivers, respectively, in optical
access. There have been demonstrations of an external modulator, for example EAMs
integrated with SOAs [86, 152, 153]. As pre-amplifier, SOA can be integrated with a
PIN photodiodes or APDs [154, 155].
 Burst mode transmission: The dynamic characteristics of the SOA must support burst
mode transmission in TDM-PONs. Power equalization of a burst mode upstream
transmission in TDM-PON is the other applications. In TDM-PON, the power available
at the RN from the multiple subscribers varies due to the different distance of the
subscribers from the RN. Therefore, SOAs with controlled bias or SOAs in gain
saturation could be used to equalize these powers [156, 157].

Directly Modulated Transmitters

SOAs and RSOAs are used as directly modulated US transmitters mainly in WDM-PONs. This is
because they offer a colorless feature thanks to their broad optical bandwidth [140]. The general
requirement of directly modulated transmitters is a broad modulation bandwidth. SOAs and
RSOAs have relatively narrow modulation bandwidth, 1 – 3 GHz compared to laser diodes which
offer >10 GHz. The other characteristic requirements depend on the configuration scheme of the
SOAs and RSOAs in the WDM-PON. There are three alternative configurations, which determine
the required properties of the device.
 The SOA/RSOA could be injected with unmodulated laser seed light, i.e. the externally
seeded architecture depicted in Figure 1-14 and discussed in section 1.2.5.3. Here the
BLS is a laser array. Then the device only serves to amplify the seed line and encode
the US signal onto it. Here, it is important that the RSOAs have higher gain and
saturation power. As example, 30 dB gain with 11 dB saturation power is presented in
[150]. Multi-section RSOAs have also been tested to improve the gain and reduce
noise as in [157].
 The RSOA or SOA could be seeded by a noisy spectrum-sliced ASE [80]. The
architecture is similar to the one presented Figure 1-14 with an ASE source as a BLS.
Even if the seeding light is not modulated, it is still necessary to suppress the extra
intensity noise of the seeding ASE. In this case the device should have lower saturation
power so that it is easily pushed into the nonlinear regime to suppress the ASE noise.
Therefore, the device performs three functionalities simultaneously; noise
suppression, modulation and amplification.
 In wavelength reused WDM-PONs, also discussed in Figure 1-15 and section 1.2.5.4,
the SOA or RSOA are seeded with a modulated light [115]. Therefore, the device has

30
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

to erase the DS modulation before encoding the US signal. Thus, similar to the second
application, the device should have low saturation power for nonlinear suppression of
the DS modulation. Input saturation power of below -13 dB at gain of 13 dB was
presented in [105].

Suppression of Intensity Noise

The ability to design SOAs and RSOAs to have dominant nonlinear characteristics opens up other
application route. Specifically, the strong gain saturation makes them attractive for suppressing
amplitude noise in PONs, which rely on low-quality seeding sources. RSOAs are more dominant
as they offer lower saturation power due to the bidirectional travel of light in the device [158,
159]. Gain saturated SOAs [77, 78, 104], or RSOAs [79] were used to suppress the excessive
amplitude noise of an ASE that was used in spectrum-sliced WDM-PONs. The nonlinear noise
suppression and RIN reduction capacities of RSOAs and SOAs are discussed and compared in
[160]. Saturated SOAs and RSOA are also used to suppress back-scattered amplitude noise in
single fiber access loop [161, 162].

2.3 Operation Principle of SOAs and RSOAs

This subsection presents the light amplification principle used in SOAs and RSOAs. As RSOAs are
basically SOAs with one reflective facet, the same principle applies for both. Light-matter
interaction in semiconductors, specifically emission and amplification processes are discussed.
In addition, device structural designs for efficient amplification and device characteristics are
presented. The last section presents mathematical models for gain and field propagation in the
device.

Light Absorption and Emission in SOAs and RSOAs

Light can be absorbed or emitted by semiconductor materials. In SOAs and RSOAs, light emission
is the dominant process which is exploited for amplification. A photon is emitted when a free
electron in the conduction band recombines with a hole in the valence band of a direct bandgap
material. In case of SOAs and RSOAs, the emitted photon can have identical properties as an
incident photon, leading to light amplification. The electron-hole pairs can be created by either
optical or electrical pumping, the latter being common in semiconductor optoelectronic devices.
In general, there are two mechanisms of transition of carriers from the conduction and valence
band: radiative and non-radiative transitions. Radiative transitions involve the radiative transfer
of energy in a medium, mainly by absorption or emission of a photon. On the contrary, as the
name indicate for non-radiative transitions, there is no interaction with a photon. Radiative
transition process can further be classified as stimulated absorption, spontaneous emission,
stimulated emission. The occurrence probabilities of these processes are related by Einstein
relations. The major non-radiative transitions are Auger recombination and trap assisted
recombination. Figure 1-2 shows schematics of the different types of transition processes for a
simple two level system. In the two level system, the valance and conduction bands have
energies . E1 . and E2 , respectively. The average numbers of carriers in the two energy levels
E1 and E2 are N1 and N 2 , respectively. Figure 1-2(a) and (b) depict radiative and non-
radiative transitions, respectively. These processes are briefly introduced as follows [127]:

31
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

 Stimulated absorption

In this excitation process, an incident photon with frequency   ( E2  E1 ) h is absorbed by an

electron in the valence band. The photon energy excites the absorbing electron to the
conduction band as shown in Figure 1-2(a.1). For an incident photon with energy density  , the
number of carriers at the higher state E2 can be increased by stimulated absorption at a rate
r12,stim given by

r12,stim  B12  N1 (2.1)

where B12 is the Einstein-coefficient for stimulated absorption. One can see that the stimulated
absorption rate directly relates to the occupation of the lower level E1 . Highly occupied lower
level means there is a higher probability that an electron absorbs a photon of energy
h  E2  E1 to jump to level E2 . It also directly relates to the incident light photon density.
This process is used in photodiodes.
 Spontaneous emission
As depicted in Figure 1-2(a.2), electrons and holes recombine spontaneously and emit light with
random characteristics, i.e. emission direction, polarization, momentum and phase. The emitted
photon has an energy that is equal to the energy difference between the two levels: E2  E1 .
The rate of the spontaneous emission r21,spon , which expresses the transition rate of electron
from the conduction to the valance band is given by

r21,spon  A21 N 2 . (2.2)

Here, A21 is the Einstein-coefficient for spontaneous emission. r21,spon depends only on N 2 as it
is the density of the higher level that determines the rate of spontaneous relaxation of an
electron to the valence band. For efficient optical amplification, this process should be
minimized.
 Stimulated emission
A photon stimulates the electron-hole recombination which ends up with an emission of a
second photon, see Figure 1-2(a.3). The emitted photon is coherent with the stimulant photon,
i.e. it is an identical replica of the incident photon, having the same frequency, phase,
propagation direction and polarization. For stimulated emission, the energy of the incident
photon must equal the energy difference of the two energy levels: E2  E1 , and its frequency
being   ( E2  E1 ) h , where h is Plank’s constant.

The stimulated emission rate depends on two parameters: the energy density of the incident
light  and the occupation of the higher energy level N 2 . It is given by

r21,stim  B21  N 2 , (2.3)

where B21 is the Einstein-coefficient for stimulated emission. The stimulated emission rate
increases with both higher input photon energy density and larger N 2 . Stimulated emission is
the fundamental process behind SOAs, RSOAs and lasers.
 Trap assisted recombination (Shockley–Read–Hall process):

32
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

An impurity may cause a trap state level between the conduction and the valence band as
depicted in Figure 1-2(b.1). The electron from the conduction band then falls through this trap
without emitting a photon.

Figure 2-2. Light matter interaction for a simple two level system with energies E1 and E2 : (a) and (b) show radiative and

non-radiative transitions, respectively. (a.1) Stimulated absorption: an incident photon of frequency   E2  E1 / h can 
also be absorbed to excite an electron from E1 to E2 . (a.2) Spontaneous emission: an electron-hole pair recombines
 
spontaneously and a photon of frequency   E2  E1 / h is emitted. (a.3) Stimulated emission: the electron-hole pair
recombination is stimulated by an incoming photon. The generated photon and the incoming, stimulating photon are identical.
(b) Non-radiative recombination mechanisms. (b.1) Trap assisted recombination: The excited electron relaxes trough an
intermediate impurity level. (b.2) Auger recombination occurs when the energy emitted by electron-hole recombination is
absorbed by a third carrier. If the third carrier is a hole, it will be excited to higher energy in the valence band. In case an electron
in the conduction band absorbs the emitted energy, it will be excited to higher conduction band energy level.

 Auger recombination:
The energy emitted by the electron-hole recombination may be absorbed by a third carrier. The
third carrier will be excited to higher energy state within its original band or split-offs, see Figure
1-2(b.2). This causes carrier heating. The probability of Auger recombination is relatively high in
long wavelength materials, e.g. in InGaAsP, which leads to higher temperature sensitivity of
related devices [127].
From previous calculations, the probability for the occurrence of the radiative transition
processes depends on the number of carriers in the two levels. For a non-degenerate two level
system at thermal equilibrium, the ratio of the occupation probability in the two states with
energies E1 and E2 is given by the Boltzmann statistics as [131]

N2  ( E2  E1 ) 
 exp  , N  N1  N 2 . (2.4)
N1  kBT 

k B and T are the Boltzmann constant and the system temperature, respectively. In Eq. (2.4),
the right hand side is a negative exponential, which is less than 1. This indicates that at thermal

33
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

equilibrium, the higher energy level is less populated than the lower level. This relative
population state, quantized by the occupation ratio N 2 N1 , determines which of the radiative
processes dominates.
For the case of B12  B21 at thermal equilibrium [163], the relation between the Einstein-
coefficients for spontaneous and stimulated emissions, A21 and B21 ., is given by

A21 8 n 3h 3
 , (2.5)
B21 c3

where n , and c are the refractive index of the material and the speed of light in vacuum. With
derivation given in [126], the stimulated emission rate is now written as

A21c3  ( ) N 2
r21,stim  ,  ( )   l ( )
8 n 3h 3 (2.6)
A c  l ( ) N
3
 21 3 3 2 .
8 n h

 ( ) is the incident photon density at frequency  .  is the energy density of the light that
caused the transition which resulted in an emission of photon with frequency  . l ( ) is the
transition line shape function and l ( )d gives the probability a single spontaneous emission
results in an emission of a photon with frequency in the range between  and   d . In case
of a single wavelength incident light into a material with cross sectional area of Axs , the power
generated within the differential length dz and thus volume Axs dz is given by

dP   r21,stim  r12,stim  h Axs dz. (2.7)

r21,stim  r12,stim is the net rate between the stimulated emission and absorption, and in case
r21,stim  r12,stim , then dP is positive and the power increases. Then, inserting Eq. (2.1) and (2.6)
in (2.7) and using (2.5), one can derive

dP
 g m ( ) P , (2.8)
dz

where, g m ( ) is the material gain coefficient given as

A21c 2l ( )  N 2  N1 
g m ( )  , (2.9)
8 n 2 2

with P   ( ) Ac n . Eq. (2.9) is the material gain for a simple two level system. A further
expanded version for semiconductors is presented in section 1.3.6.1, page 44. In Eq. (2.8), a
positive dP / dz indicates light amplification. Eq. (2.9) proves that N 2  N1 is a necessary
requirement to achieve net gain and thus amplification. The phenomena N 2  N1 is called
population inversion. However, this is not achievable with simple two level systems.
Interestingly, semiconductors allow a design to achieve population inversion. The discretely
spaced energy levels in the conduction and valence bands of semiconductors can be considered
as multi-level systems as shown in Figure 2-3. The highest energy level in the conduction band,
N
ECB has the lowest occupancy probability. On the contrary, the occupancy probability of the

34
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

lowest energy level of the valence band is the highest. This means, the excitation of an electron
N N
from level EVB to level ECB is highly possible, see the blue transition line in Figure 2-3.

Figure 2-3. Density of state vs. energy for a direct band gap semiconductor: The conduction band and valance band energy levels
1 N 1 N
are given by ECB ECB and EVB EVB . qV is the energy needed to excite an electron from the bottom of the valence
band to the top of the conduction band (blue line). The electron then instantaneously decays to the bottom levels of the conduction
band before falling to the valence band by emitting a photon which is identical with the incident photon.

This phenomenon can be exploited to achieve population inversion between the highest
1
unoccupied level of the valence band, EVB and the lowest occupied level of the conduction band,
1 N N
ECB . With the appropriate amount of energy, electrons can be pumped from EVB to ECB . In
the optical pumping case, one can shine light with photon energy of ha  ECB  EVB to the
N N

system. The other pumping alternative is electrical injection of carriers that is common for
optoelectronic devices such as semiconductor lasers and SOAs. Here the applied voltage V must
fulfill eV  ECB
N
 EVB
N
, where e is the electron charge. After the electron is excited to ECB N
, it
1
quickly relaxes to ECB , indicated by the green transition in Figure 2-3. Therefore, with suitable
N 1
pump energy, more electrons could be exited to ECB which relax to ECB , creating population
inversion between ECB and EVB , thus, fulfilling N 2  N1 in Eq. (2.9). Here, N1 and N 2 are the
1 1

1 1
carrier numbers on EVB and ECB , respectively.
The probability of obtaining population inversion is higher for pn junction structures. This is
because doping introduces higher carrier densities, i.e. electrons in the conduction band of an n-
type and holes in the valence band of a p-type semiconductor.

pn Junction Structures
This subsection discusses the different pn junction architectures implemented in semiconductor
optical devices, also mainly in SOAs and RSOAs. It starts with the classical pn monojunction and
proceed to the modern heterostructure.

2.3.2.1 pn Monojunction
The pn monojunction interface is created between two layers of the same material, thus same
band gap energy, E g , however with different doping types – p and n, as shown in Figure 2-4(a).
A pn junction simplifies the process of achieving and maintaining population inversion due to the
high density of carriers [164]. Figure 2-4((a) shows a pn monojunction without bias voltage at
thermodynamic equilibrium. In this case, the quasi Fermi levels of the two sides, EFN and EFP
are on the same level and continuous throughout the system. Thus, the junction band diagram
bends as shown.

35
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

Figure 2-4. Energy band diagram of pn junction: (a) shows the band diagram without a bias voltage. The system is bent to maintain
constant Fermi-levels, EFP and E FN in the p and p regions. ECB and EVB are the bottom of the conduction band in the n-
layer and the top of the valence band in the p-layer, respectively. E g is the bandgap of the material. The blue and the red circles
represent the holes and electrons. (b) shows the band structure after the application of a forward bias voltage V . The Fermi
levels shift according to the bias voltage, creating a region with high density of electrons and holes. This increases the probability
of electron hole recombination to emit a photon.

N N
The bottom of the conduction band, ECB is below the top of the valence band, EVB , with
minimum chance of an electron-hole recombination. The holes in the p-type and the electrons
in the n-type region are visualized by the blue and red circles, respectively. However, a forward
bias voltage V , across the pn junction will cause the Fermi energies of both n and p-type regions
to shift proportionally as shown in Figure 2-4((b). Here, the voltage with qV  EFN  EFP causes
the Fermi levels to split in the active region. This will cause the migration of electrons and holes
into the active region, thus creating a high density of both carrier types. The high concentration
means higher probability for an electron-hole recombination, which results in the emission of a
photon with frequency  e . The emission process fulfills

Eg  he  EFN  EFP  qV (2.10)

The drawback of such simple pn monojunction is that the electrons and holes may drift away
from the interface without recombination and thus without emitting light. Such a two layer
structure is also not optimum to confine the light within the active region. Thus, monojunctions
have very low electrical to optical conversion efficiency.

2.3.2.2 Heterostructures
A three-layer Heterostructures design, first suggested in [165, 166], mitigates the inefficient
carriers and light confinement issue of the two-layered homojunctions. To do so, it adds one
more layer, in this case intrinsic material layer, between the p and n-layers, see Figure 2-5. The
additional layer adds extra degree of freedom to achieve both carrier and light confinement.
 Carrier confinement: In the three-layered structure, a small band gap material layer is
sandwiched between two larger bandgap materials, see Figure 2-5(a). This forms a
potential well in the active region in which the carriers are trapped once they enter
during forward bias as shown in Figure 2-5(b). The confined carriers cannot drift to the
p- and n-layers thus improving the probability of radiative recombination, and
therefore, the electro-optical efficiency. The higher bandgap of the cladding materials
also avoids recombination outside the active region, thus restricting the gain or light

36
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

amplification region. Further, no re-absorption of the emitted light takes place in the
outer cladding layers due to the larger bandgap.
 Light confinement: The three-layered structure also enables refractive index
optimization to enhance the light confinement through wave guiding effect. The active
layer is designed to have higher refractive index compared to the cladding to confine
the light by total internal reflection. The refractive index profile is shown in
Figure 2-5(c).
Interestingly, the carrier and light confinement are complementary in a heterostructure, i.e. both
can be achieved simultaneously in the intrinsic layer. This is because bandgap energy and
refractive index are generally inversely related for direct bandgap materials [130, 167]. The high
concentration of carriers in the active region cause band-shrinkage in materials, for example
InGaAsP, which induces a positive refractive index change. Thus, intrinsic InGaAsP will have
narrower bandgap and higher refractive index, compared to the dopped p- and n-layers.

Figure 2-5. Energy band diagram of a double heterostructure: (a) shows the band diagram without bias. An intrinsic layer with
bandgap energy Eg ,1 is sandwiched between the p- and n-layers of bandgap energy E g ,2 where E g ,1  E g ,2 . The electrons
and holes are represented by the red and blue circles, respectively. (b) shows the band diagram when a forward bias voltage of
V is applied. The Fermi levels are rearranged and a barrier is created. The electrons and holes migrate into the active region and
are confined within the barriers. The recombination probability is higher, thus boosting the chance of recombination and photon
emission. (c) shows the refractive index profile of the three layers. The intrinsic material has higher refractive index so that the
photons are confined to the intrinsic region.

A band diagram for a heterostructure is shown in Figure 2-5. Since the bandgaps of the materials
are different, we call the junctions heterojunctions. The whole structure is also called double
heterostructure, since it has two heterojunctions. Figure 2-5(a) shows a double heterostructure

37
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

band diagram without a bias voltage. At thermal equilibrium, the energy bands of the layers bend
to maintain the same Fermi level throughout the structure. The red and blue circles represent
the free electrons and holes in the conduction and valence band, respectively. When a forward
bias voltage of V is applied, the Fermi levels split as shown in Figure 2-5(b). The electrons are
injected from the n-type material into the active region, and the same amount of holes are
injected into the active region through the p-side. The injected electrons from the n-type region
are not able to pass to the p-type region since their movement is blocked by the potential barrier
due to the bandgap difference between the active material and the cladding. Then, the trapped
carriers recombine in the active region with higher probability (higher rate).
The carrier confinement and thus the improved electro-optical conversion efficiency of the
double heterostructures lowered the threshold current, thus the power consumption of such
devices. Low electrical power consumption also means simpler temperature control. The carrier
confinement can be further improved by reducing the thickness of the active region, d shown
in Figure 2-5(a) and (b), further increasing the gain within the confined volume. However due to
diffraction limit, after a minimum thickness in the range of 0.1 μm, the optical mode will no more
be confined. Rather it spreads out into the cladding. In other words, the coupled control of both
carrier and light confinement results in a minimum thickness of the active region, preventing any
further efficiency enhancement.

2.3.2.3 Separate Confinement Heterostructure (SCH)

In the separate confinement heterostructure (SCH), the carrier and light confinement control are
uncoupled [164]. In this case, the active region is sandwiched between two additional layers,
which are again surrounded by the cladding material, see Figure 2-6.

Figure 2-6. Separate confinement heterostructure (SCH): (a) SCH band diagram after a forward bias voltage V is applied. Here
the carrier confinement and light confinement are controlled separately. The carriers are confined within the active region with
band gap energy Eg ,1 and thickness d CC . The electrons and holes migrate to the active layer during forward bias and are
confined there by barrier created due to E g ,1  E g ,2 . The emitted photons are confined within the light confinement region
with thickness d LC The refractive index difference between the light confinement layer with bandgap E g ,2 and the cladding
(band gap E g ,3 ) guides the light. The refractive index profile of the three layers is depicted in (b).

The active region has bandgap of Eg ,1 which is lower than that of the light confinement layer
characterized by Eg ,2 . The carriers are confined within the narrow bandgap active region with

38
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

thickness d CC due to the potential well created since Eg ,2  Eg ,1 . The light is confined within
d LC by the refractive index difference between the cladding, with bandgap energy of Eg ,3 and
the light confinement layer. The refractive index profile is plotted in Figure 2-6(b). Thus the
carriers are confined within a region narrower than the waveguide where the light is confined.
This enables the separate control of the carrier and light confinements. Further reduction of the
active region thickness dCC of SCH to few nanometer offers even higher carrier confinement,
efficiency and thus lower threshold current.

Figure 2-7. Quantum well structures: (a) shows the density of states comparison of bulk (SCH) and quantum well (QW) materials.
In case of QWs, the density of state has a step like structure which depends on the dimension of the QWs, e.g. the carrier
confinement layers thickness d CC . This limits the movement of the carriers to a plane. (b) shows the band structure of a forward
biased single quantum well structure. Eg ,1 , E g ,2 and E g ,3 are the bandgap energies of the quantum well, the light
confinement layer and the cladding, respectively. The different colored lines in the light confinement layer, thickness d LC , show
that the bandgap change may be designed to be step like (black), linear (red) or quadratic (green). This is also reflected on the
refractive index profile (graded index, GRIN) plotted below. (c) shows the band diagram for a multi-quantum well (MQW)
structure. As in (a), the different colors indicate the possibility to design step like (black), linear (red) or quadratic refractive index
profile for the refractive index and gain of the light confinement layer.

39
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

2.3.2.4 Quantum Well (QW) structures

If the active region thickness is in the range of 4 – 20 nm, such SCH are called quantum wells
(QW): quantum because at such size, quantum effects come into play and well because of the
potential well created around the active section. Such thickness of the active layer results in a
different characteristics compared to a bulk semiconductors. The distribution of the density of
states (DOS) is strongly influenced by the quantum effects. The DOS in bulk semiconductor was
discussed in Figure 2-3, where it shows a continuous characteristics. However, when the
thickness is in the few nanometer range, the so called quantum size effects (QSE) kick in [168].
The energy states are quantized and the density of state has a step like characteristics [126], as
shown in Figure 2-7(a). The band diagrams of QW structures are plotted in Figure 2-7(b) and (c)
for single and multi-quantum wells (MQW), respectively. For better optical confinement, one can
also use graded index (GRIN) light confinement layer as depicted by the colored band diagram
lines. The red and green lines show linear and parabolic GRIN, respectively as shown at the
bottom of Figure 2-7(b) and (c) [169].
The main improvement of QW compared to the bulk based SCH is due to their step like density
of state. This contributes to the relatively flat optical gain bandwidth [168], low threshold current
with minimum dependence on temperature [170] and higher gain. The thickness of the single or
multiple quantum wells or the barrier height can be designed to a specific application, for
example to tune the emission wavelength.

Lateral Confinement and Polarization Sensitivity

As discussed in the previous subsection, one driving force for the evolution of pn junction was
efficient light confinement in the vertical direction, x -axis in Figure 2-8(a). In the following
section, multiple design options for light confinement in the lateral dimension ( y -axis) are
discussed. Any imbalance between the vertical and lateral confinements will result in polarization
sensitivity, which is also a topic of this subsection.

2.3.3.1 Lateral Confinement

The gain structures discussed so far, SCHs and QWs, offered light and carrier confinement in the
vertical x -axis direction as shown in Figure 2-8(a). However, the confinement of light in the
lateral y -axis is also as important. There are a few methods to achieve lateral light confinement.
The three schemes are given in Figure 2-8(b) – (d). In the simplest case, one can design the system
in a way that only a limited area in the lateral direction offers gain. This is done by physically
restricting the electrical contact area, i.e. using strip contact, as shown in Figure 2-8(b). Only the
photons propagating in this area are amplified due to the formed gain profile, thus confining the
light. This method is called gain guided as the gain is used for confining the light. Another option
is to introduce an insulator to block the carrier injection. For example silicon dioxide SiO 2 is
etched bellow the electrical contact so that only the strip contact can inject carrier into the p-
layer cladding. This blocks the carrier flow, as shown in Figure 2-8(c) by the ridge waveguide
structure. Since SiO2 is not a good heat conductor, proton bombardment has also been used to
block the carriers and thus confine the carrier injection, while offering good heat conductivity.
The gain-guided structures are relatively simple to fabricate. However, the confinement is not
that efficient and suffer from higher threshold currents. Also the carrier concentration in the
active layer change will change the refractive index, thus the light confinement will vary leading
to multimode operation and non-uniform light beam [171].

40
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

Figure 2-8. Typical SOA structures and lateral light confining structures: (a) shows a 3D schematics of a typical heterostructure
SOA with p, intrinsic and n layers. The pump current is injected through the top electrode (black plate). The active region has a
width of w and thickness d . The light is confined within the active region in the ( x, y) plane while propagating along z -
axis. (a) Gain guided: the relatively simpler way to restrict light within the active region along the ( x, y) plane is to limit the
carrier injection region by using a strip electrode. Only photons below the strip-electrode will experience gain and are amplified.
The arrows show the carries flow. The carrier injection region can also be limited by adding an insulating layer below the electrode
as shown in (c). The highest light confinement is achieved by adding lateral refractive index difference by introducing the burying
layers around the ridge active region, (d).

Higher light confinement can be achieved by introducing refractive index change in the lateral
direction as well. The idea is to bury the high refractive index active layer in a low refractive index
burying materials. This method is called index guiding using buried ridge active region. A general
structure of the buried ridge structure is given in Figure 2-8(c). The burying layer, most of the
time p-type material with lower refractive index, is used to create total internal reflection for
lateral confinement. Similar to the gain guided ridge waveguide structure, the burrier ridge
method uses dielectric material under the p-contact to restrict the carrier injection to the active
area. The added refractive difference between the p-type cladding and the dielectric material
contributes to additional light confinement.
Now that both lateral and vertical confinement methods are discussed, next comes the
important parameter, the confinement factor. The confinement factor quantifies how much of
the light is really confined within the gain producing region in relation to the total power. If the
lateral profile of the propagating electromagnetic field is F ( x, y) (the solution of the wave
equation as presented in 1.3.6.2), the 2-dimensional confinement factor  is given as [163]
 w/2  d /2
 
2
F ( x, y ) dxdy
  w/2  d /2
 
  x y , (2.11)
 
2
F ( x, y ) dxdy
 

where d and w are the thickness and width of the active region.  x and  y are the
confinements in the x and y dimensions, respectively. Thus, in practice, the confinement
factor can also be controlled by carefully designing the width and height of the active region.
Another explanation is that, the confinement factor quantizes the photon density in the active
region that leads to stimulated emission. Thus, higher confinement means strong saturation due

41
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

to strong carrier depletion. In addition, high confinement factor leads to high ASE, which
increases the noise figure of SOAs and RSOAs [172]. Further, difference between  x and  y
causes polarization dependence of the gain which will be discussed next.

2.3.3.2 Polarization Sensitivity

When SOAs/RSOAs are used as inline or pre-amplifiers in optical fiber communication,
polarization independence of the gain is a critical point. However, the different light confinement
factor in the vertical and lateral directions means that the two polarizations will experience
different gain. The lateral and vertical confinement difference emanates from the material
structure difference in the two dimension. In addition, for the gain-guided structure, the carrier
injection changes the refractive index of the cladding above and below the active region. Of
course this change is not identical as the one in the lateral x -axis. This means the x-and y-
polarization experience different confinement factor and thus different gain. Moreover, the
different reflectivity of the cleaved facet of the SOAs could be different for the two polarizations.
The polarization dependent gain (PDG) is given as

   g mat,TM 
PDG  G  1  TM . (2.12)
 TE  g mat,TE 


The parameters g mat,TM and g mat,TE are the material gains for the TM and TE polarizations, in the
x and y dimensions, respectively. In addition,  TM and  TE are the TM and TE confinement
factors in the x and y dimensions, and  TM   x and  TE   y as in Eq. (2.11). The polarization
dependent gain can be mitigated using different methods, by playing with either the
confinement factors or the material gain. The confinement factor depends on the geometry of
the active region. Therefore, one can design the geometry of the system carefully so that the two
polarizations experience the same confinement factor. The simplest solution suggested was to
use square shaped (symmetrical shaped) active region to have equally high confinement for both
polarizations and achieve PDG less than 1 dB [173]. This method results in huge divergence of
the light at the output thus making coupling to a fiber a difficult task. Long tapers were also used
to improve the fiber coupling with low PDG [174]. In case of gain guided ridge waveguide
structures, the dimensions of each layer are controlled to get equal light confinement for the
two polarizations. Especially the active layer is made relatively thicker to offer this flexibility
[126].
The second parameter determining the PDG was the material gain for the two polarizations,
g mat ,TM and g mat ,TE . The material strain in semiconductors can be used to engineer the bandgap
and its material gain coefficient in each dimension. Therefore, the difference in the confinement
factor can be compensated by modifying the material gain in each dimension using strain
material SOAs [173-175].

Additional Design Challenges

Additional general requirements of SOAs and RSOAs are discussed as follows.
 The facet reflectivity is an important parameter influencing the gain and spectral
characteristics. Initially, SOAs where referred to as Fabry-Perot semiconductor laser
amplifiers due to the relatively high reflectivity, 10-2, of the facets. This resulted in high
gain ripples (up to 10 dB) which was visible in the ASE spectrum. This strongly limits

42
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

the optical bandwidth. However, improved antireflection coating and tilting the device
waveguide by 5° to 10° has allowed for lower reflectivity. Combination of the two
schemes and windowing (spot size conversion to broaden the mode size so that the
reflection is minimized) [176, 177] have resulted in the reflection of the facets below
10-5. Such SOAs with low gain ripple <1 dB are called travelling wave (TW) SOAs. The
impact of facet reflectivity on an SOA performance have been studied in [178]. With
this respect, the only difference between SOAs and RSOAs, is that in the case of RSOAs,
one facet is reflective. The reflective facet of the RSOA is built by high reflectivity
coatings with the materials SiO2 and TiO2.
 SOAs must allow only a single transverse mode. Single mode operation of the SOA is
controlled by designing the size of the active region cross section accordingly. This is
important for the light confinement and avoid mode dependent gain. Plus, it is
necessary for coupling to a SMF.
 The coupling efficiency form the output facet of an SOA or RSOA to a single mode fiber
must be kept maximum to have high fiber to fiber gain allowing for values of even
more than 25 dB [174]. The first step to lower the coupling loss is to adapt the mode
size of the amplifier waveguide to match that of a SMF. This is done by using mode
size adapter tapers which have lowered the coupling loss between 2 – 3 dB.

Materials
By the 1960s, compound direct bandgap materials such as GaAs, GaAsP, AlGaAs, InAs and InP
have already been used in semiconductor lasers. When it comes to SOAs, the quaternary
semiconductor InGaAsP active region with InP cladding was demonstrated to work at
wavelengths of 1.3 and 1.5 µm. This was a significant step towards the incorporations of
semiconductor lasers and SOAs in optical communication.
There are several material characteristics that have to be taking it into account for the SOA
fabrication. In the processing of semiconductors, lattice matching between different layers for
epitaxial growth is a necessity. For example, InGaAsP is lattice matched to InP enabling crystal
growth [179], and for this reason InGaAsP/InP is widely used in lasers and SOAs. The refractive
index of each layer is also designed accordingly to obtain the optimum light confinement. In
general the composition of the material indicated by the subscripts x and y, e.g. as In1-xGaxAsyP1-
y, determine both the bandgap energy (the emission wavelength), the refractive index and also
lattice constant. The composition also enables to tune the strain of the material. Modifying the
strain allows for optimizing the performance of the device [180]. Strain engineering has been
used to mitigate polarization dependence of the gain, to improve threshold current and tune the
operation wavelength (band gap).

Additional parameter for a material is its temperature tolerance. It is suggested that the low
electron confinement and high Auger recombination at high temperature degrade the
performance of InGaAsP/InP devices [181, 182]. Thus the packaging becomes complicated as it
has to include cooling system. Using the materials AlGaInAs/AlGaInAs and InGaAsN/GaAs has
shown to offer better temperature independence even enabeling uncooled operation [183]. In
such material devices, higher carrier confinement is obtained due to larger bandgap offset
between the active and the confinement layers.
New material compounds such as ones based on Bismuth are being investigated as they offer
higher efficiency because of their low non-radiative recombination [184]. The highly reduced

43
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

inter valence band non-radiative recombination in Bismuth alloys, e.g. GaBiAs was reported in
[185]. GaBixAs1-x/GaAs laser structure has been discussed in detail in [186]. GaNyAs1-x-yBix alloy is
another material being investigated for better optical efficiency with lower threshold current
[187, 188].

Details on Material Gain and Light Propagation in SOAs and RSOAs

This subsection discusses the material gain calculation and the wave propagation inside a
semiconductor active medium in more detail.

2.3.6.1 Material Gain

The simplified 2-level system and the requirement of population inversion for light amplification
were discussed in the subsection 2.3.1. The material gain at an emission frequency  is derived
to be

A21c 2l ( )  N 2  N1 
g m ( )  , (2.13)
8 n 2 h 2

Where A21 is the is the Einstein-coefficient for spontaneous emission which can also expressed
as A21  1  sp ,  sp being the spontaneous transition transmission life time. In semiconductors,
which are multilevel systems with defined DOS, the above gain equation needs to be adapted.
This is done as follows. First of all, the population inversion component, N 2  N1 , can be
expanded to the net transition from the conduction band to valence band. The transition rate
from an energy level EC in the conduction band to energy level EV in the valance band within
the frequency interval d0 is given by

dRCBVB  CV (0 ) f C ( EC ) 1  f V ( EV )  d0 (2.14)

where CV (0 ) is the optical joint density of state at the transmission frequency  0 . The
occupation probability of an electron with energy E in the conduction and valence band, f C ( E )
, and f V ( E ) , respectively, follow Fermi-Dirac statistics and are given as

1 1
fC ( E )  , fV (E)  (2.15)
 E  EFC   E  EFV 
1  exp   1  exp  
 kBT   kBT 

EFC and EFV are the quasi-Fermi levels in the conduction and valence bands, respectively. k B
is the Boltzmann constant and T is the temperature.
Eq. (2.14) states that the rate of an electron transition from the conduction to the valence band
is determined by the joint density of states, the occupation of the conduction band and the
inoccupation of the valance band. Similarly, the rate for the opposite transition rate from the
valence band to the conduction band is given as

dRVBCB  CV (0 ) f V ( EV ) 1  f C ( EC )  d0 (2.16)

Here the upward transition depends on the density of states, the occupation of the valence band
and the inoccupation of the conduction band. Then, the net transition rate is given by the
difference

44
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

d ( N 2  N1 )  dRCBVB  dRVBCB
(2.17)
 CV (0 )  f C ( EC )  f V ( EV ) d0

Substituting Eq. (2.17) into Eq. (2.13), the gain coefficient dg m ( ) will be

A21c 2
dg m ( )  l ( ) CV (0 )  f C ( EC )  f V ( EV )  d0 , (2.18)
8 n 2 h 2

The total gain is then the integral over the whole frequencies written as

A21c 2
g m ( )   l ( )  (0 )  f C ( EC )  f V ( EV )  d0 , (2.19)
8 n 2 h 2
CV


l ( ) represents the spectral broadening due to the intraband relaxation. The common function
used is a Lorentzian, which represents the homogeneous broadening in optical semiconductors,
and is given as


l ( )  , (2.20)
2 h (  0 )2    2  
 2
 

where  is the frequency spreading that is due to the intraband relaxation time  in,relax and
they are related as

1
  (2.21)
 in,relax

In practice, the material gain given by Eq. (2.19) is complicated to solve. Typically, for sufficiently
narrow or single wavelength operations, a simple linear dependence of the gain peak on the
carrier density is used, which is given by

g m  a0  N  N 0  (2.22)

where a0 is the differential gain coefficient and N 0 is the carrier density at transparency. This
linearization is applicable for relatively slow (picosecond and higher) dynamic operations.
Further, when an optical mode field is propagating in an amplifier, the effective modal gain is
calculated my multiplying the material gain by the confinement factor  as

g  g m  a0  N  N 0  (2.23)

2.3.6.2 Field Propagation

Here, the optical field propagation in an SOA active medium is discussed. Before going to detail,
it is important to state the assumptions to simplify solving the general Maxwell’s equations. The
medium is assumed to be non-magnetic and it has sufficiently low free charge carriers. The latter
applies as long as the active region is not doped. Further, the active medium is assumed to be
isotropic, i.e., the material property does not depend on propagation direction.
The propagation behavior of an electromagnetic field is determined by the material
characteristics, i.e. its relative permittivity  r . As an electromagnetic field E travels in a

45
Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

medium with the mentioned assumptions, its propagation is governed by the wave equation
[189]

 r 2 E
E   0, (2.24)
c 2 t 2

where the complex relative permittivity  r of the active medium is related to the complex
refractive index as
r  n 2. (2.25)
Here, it is assumed that the propagating field has a sufficiently narrow spectral range or it is a
monochromatic field and the medium is weakly attenuating. Thus it is acceptable to assume
 r   r (0 )  n (0 ) 2  n 2 at an optical angular carrier frequency 0  20 . In practical
scenario, when a field propagates through a medium, the refractive index that it experiences is
influenced by the carrier density of the active medium. The complex refractive index taking this
effect into account is given by [190]

g ( N )
nNL  n  n ( N )  j , (2.26)
2k0

where n is the background refractive index which takes into account the lateral 2D index profile
of the waveguide in ( x, y) dimension. n ( N ) and g ( N ) represent the refractive index and
gain change induced by the carrier density N . k0  0 c is the vacuum wave number. The
dielectric constant will then be
2
 g ( N ) 
 r   n  n ( N )  j 
 2k0 
2
g ( N ) n ( N )g ( N )  g ( N ) 
 n 2  2n n ( N )  j n  n ( N ) 2  j   (2.27)
k0 k0  2k0 
g ( N )
 n 2  2n n ( N )  j n
k0

In Eq. (2.27), the final result is obtained by neglecting higher order terms which are relatively
small [190]. In another notation, the carrier density dependent part of the dielectric constant can
be expressed by the susceptibility     N 

r  n 2   (N )
g(N ) (2.28)
 ( N )  2n n ( N )  j n
k0

In Eq. (2.26), the real and imaginary parts of the refractive index are related by the Kramers-
Kronig (KK) relation [191, 192]. KK relates the carrier density dependent refractive index change
with the carrier density dependent gain change. It was phenomenologically found to be
represented by the linewidth enhancement factor  H (also called Henery’s factor or Alpha
factor), which is given as

46
 Semiconductor Optical Amplifiers (SOAs) and Reflective-SOAs (RSOAs)

nrl N n N n ( N )
H   2k0 rl  2k0 (2.29)
nim N g N g ( N )

It can be approximated by taking the differences instead of the derivatives.  H depends on the
active material characteristics, wavelength of the propagating field and current density.
However, constant  H can be assumed when operating in a sufficiently narrow wavelength
range and when the carrier density modulation is small.
To this end, Eq. (2.24) - (2.29) are used to solve the propagation of the optical field in the amplifier
medium. The electrical field in the Cartesian coordinate is given as
E ( x, y, z, t )  e F ( x, y ) E ( z, t ) exp  j(k0 z  0t )   c.c..
1
(2.30)
2
Here, e is the polarization unit vector. The lateral waveguide mode distribution is represented
by F ( x, y) whereas the E ( z, t ) is the slowly-varying envelope. Substituting (2.30) into (2.24),
and integrating in the lateral dimension ( x, y) yields

E ( z, t ) 1 E ( z , t ) 1  k 
  E ( z, t )  j 0    int  (2.31)
z vg z 2  nb 

 2 F ( x, y )  2 F ( x, y )
   n 2  nNL
2
 k02 F  0 (2.32)
x 2 y 2

In (2.31), the group velocity is given by vg  c ng with the group index ng  n  0  n   .

Concerning the lateral profile of the optical field, (2.32) gives transverse profile and it can also be
used to calculate how much of the power is confined within the active region (width w and
thickness d ) by calculating the confinement factor 
 w/2  d /2
 
2
F( x, y ) dxdy
  w/2  d /2
 
(2.33)
 
2
F( x, y ) dxdy
 

Eq. (2.31) determines the propagation wave equation. It includes the internal scattering loss
through  int and the lateral profile is included through the confinement factor  . Substituting
the susceptibility from Eq. (2.28) and including the alpha factor from Eq. (2.29), one can reach

E ( z, t ) 1 E ( z, t ) 1
  E ( z , t )   int  g ( N ) 1  j H   , (2.34)
z vg z 2

In Eq. (2.34), the carrier density dependent of the field propagation is clearly visible. Further, in
case of dynamic operation with modulation slower than the intraband relaxation time, nonlinear
effects (such as carrier heating and spectral hole burning) result in the gain compression. This
nonlinear gain compression is phenomenologically included in the gain part of Eq. (2.34) as

E ( z, t ) 1 E ( z, t ) 1  g(N ) 
  E ( z, t )   int   1  j H   (2.35)
z vg z 2  1   NLGC Ptot ( z, t ) 

 NLGC is the nonlinear gain compression factor and Ptot  z, t  is he total optical power at position
z and time t . Eq. (2.35) is now used to model the RSOA in chapter 2.

47
3 Reflective Semiconductor Optical Amplifier Fiber
Cavity Laser (RSOA-FCL)
This chapter presents a descriptive theoretical analysis of the self-seeded reflective
semiconductor optical amplifier (RSOA) fiber cavity lasers (FCLs).
RSOA-FCLs are attractive colorless, self-seeded, self-tuning and directly modulatable sources for
WDM-PONs. They comprise of an RSOA in the ONU as the active element, a distribution fiber as
the laser cavity, a waveguide grating router (WGR) and a common reflective mirror with the latter
two positioned at the RN. In this chapter, we introduce a model and perform simulations to
elucidate the recently discovered successful operation of this new WDM-PON source. The model
is further used to predict the optimum operation conditions and configuration. The impact of the
km-long fiber cavity dispersion, loss and other related parameters are presented with support
from experiments.
The chapter is organized as follows. Section 3.1.2 starts by discussing a WDM-PON scenario
where an RSOA-FCL is deployed as an upstream transmitter. In section 3.1.3.1, the wave
propagation equation derived in section 2.3.5.2 is adapted to accommodate the multimode
nature of an RSOA-FCL, which is integrated into the overall RSOA-FCL model in section 3.1.3.2.
Finally, the simulation results including experimental verification are presented in section 3.1.4.
Section 3.2 investigates an alternative self-seeded scheme, where the cavity is equipped with an
additional RSOA to form the Fabrey-Perot cavity.
Section 3.1 has been published in the Journal of Selected Topics in Quantum Electronics in
September 2014 and it is placed here word by word. Section 3.2 is direct copy of the publication
in 2014 conference on lasers and electro-optics (CLEO). Some variable names have been modified
so as to keep the same nomenclature in this document.

49
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

3.1 Reflective-SOA Fiber Cavity Laser as Directly Modulated

WDM-PON Colorless Transmitter
Simon A. Gebrewold, Lucia Marazzi, Paola Parolari,
Sean P. O. Duill, Romain Bonjour, David Hillerkuss,
Christian Hafner, Juerg Leuthold

Journal of Selected Topics in Quantum Electronics, vol. 20; issue 5;

2014 [193].

Introduction
Reflective semiconductor optical amplifiers (RSOAs) are already deployed as colorless
modulators in wavelength division multiplexing (WDM) passive optical networks (PONs). They
further hold promise to not only act as colorless modulators but also to act as colorless laser
sources if the access network fiber trunk is used as a laser cavity [194, 195].
RSOAs so far are largely deployed as externally seeded upstream transmitters. In these
transmitters, the RSOA modulates a light that originates from a different point in the network.
And even though a common external broadband seeding source has shown to work well, the
reach and speed of such schemes are limited by amplified spontaneous emission (ASE) beat noise
and chromatic dispersion due to the broad spectrum [196]. In addition, they require high power
and broadband light sources, which makes them cost-inefficient. Alternatively, RSOA based self-
seeded WDM-PON upstream transmitters were suggested to eliminate the external seeding
source. Initial implementations of self-seeded sources employed feedback mechanisms including
fiber Bragg gratings (FBGs), [120, 197]. Unfortunately, these methods require separate and
accurately tuned wavelength selective components in every optical network unit (ONU), thus
preventing colorless operation and increasing the complexity and cost.
Recently, a more simple and cost-effective network topology was introduced that incorporates
self-seeded RSOA-fiber cavity lasers (RSOA-FCLs) [194, 198, 199]. The resonant laser cavity is
formed by the distribution network between the RSOA’s reflective facet and a mirror at the
remote node. The gain of the cavity is provided by the RSOA whereas the waveguide grating
router (WGR) at the remote node automatically selects the operating wavelength. The WGR thus
not only confines the optical bandwidth of the RSOA-FCL but also allows for an automatic
wavelength assignment that is easily scalable so that many users within the network can transmit
information at pre-defined wavelengths. The amount of users is limited by the number of output
ports of the WGR
For proper functioning of the overall system, the RSOA in the cavity should perform three
operations. Firstly, the gain has to compensate the overall cavity losses for the system to lase.
Secondly, the RSOA should offer a sufficient modulation bandwidth. Thirdly, the residual
modulation on the optical field in the upstream path of the cavity must be nonlinearly suppressed
upon return to the RSOA to provide a clean carrier for new upstream data transmission. Properly
designed and operated RSOAs are able to offer the above three functionalities simultaneously.
In particular, large gain bulk RSOAs can meet these specifications when the active region length

50
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

and the confinement factor are appropriately optimized [150]. So far, the modulation
cancellation capacity of saturated RSOAs has been demonstrated theoretically and
experimentally in [158] and [200]. Also, the electro-optical bandwidth of quantum-well RSOAs
have been explored and 10 Gbit/s direct modulation has been demonstrated [88]. The entire
RSOA-FCL system has been experimentally demonstrated in a WDM-PON for beyond 2.5 Gbit/s
upstream transmission [198, 201].Despite the amount of experimental demonstrations, so far
no theoretical and numerical treatment of the RSOA-FCL system has been offered, thus the
working principle is poorly understood.
In this paper we present a theoretical investigation, with supporting experimental results, of the
RSOA-FCL with a cavity formed by the distribution fibers in the order of a few meters to tens of
kilometers length. The RSOA-FCL excites a large numbers of longitudinal modes due to its extra-
long cavity. In order to properly simulate the interaction of these modes, we construct a
bidirectional RSOA model that accounts for: (i) all the counter-propagating laser modes; (ii) the
dynamic carrier injection in order to simulate the direct modulation capability and (iii)
spontaneous and stimulated emission. Using this model, we discuss the effects of chromatic
dispersion and mode partition noise. We show how the RSOA in a fiber cavity can be used as
directly modulatable, colorless, self-seeded upstream transmitter. It will be shown, how the
interplay between the nonlinear RSOA together with a properly designed fiber cavity suppresses
the intensity fluctuations arising from the multi-mode nature of the laser and also allows for
efficient cancelation of the modulation component of the recirculating signal in the FCL. In
particular, the performance of the RSOA-FCL with cavity lengths up to 10 km as a directly
modulated upstream transmitter at 2.5 Gbit/s is discussed.

WDM-PON Architecture
Before proceeding with the analysis of the RSOA-FCL, we briefly review the RSOA-FCL’s PON
topology. The topology of a WDM-PON access network with a central office (CO) connected to
the remote node (RN) and the ONUs through the feeder and distribution fibers, respectively, is
shown in Figure 3-1. The ONU only comprises of an RSOA, a downstream receiver (DS Rx) and a
C/L-band WGR for the C-band signal in the up- and the L-band signal in the downstream,
respectively. The RSOA serves as an active medium of the fiber laser cavity, which is formed by
the RSOA mirror on one side and the Faraday rotator mirror (FRM) at the remote node on the
other side.

Figure 3-1. SOA-FCL based self-seeded WDM-PON Architecture: CO (central office), US (upstream), DS (downstream),

51
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

The cyclic WGR multiplexer at the remote node (RN) acts as a frequency selective filter through
which the laser “self-tunes” to the particular WGR connection port. The RSOA further allows
encoding of the user upstream (US) data by direct modulation of the RSOA injection current.
The Faraday rotator (FR) in the ONU and the RN in the cavity are needed to enable lasing even
when the RSOA should have some polarization gain dependence [125].

RSOA-Fiber Laser Modeling

This section deals with the overall mathematical realization of the RSOA-FCL model. Firstly, a
multimode, bidirectional travelling wave RSOA model is presented. Then we introduce models
for the passive cavity, which consists of an optical fiber and an optical bandpass filter (OBPF).
Finally, the two are combined to form the entire RSOA-FCL system.

3.1.3.1 RSOA Model

The RSOA model is based on a bidirectional SOA implementation [158, 202]. Figure 3-2 depicts
the simulation procedure where the RSOA is sliced into m spatial sections. The temporally
resolved carrier density and gain are calculated within each section. The optical field propagation
is described in terms of travelling wave equations, and is solved for all the counter-propagating
multimode contributions to the signal and ASE fields. The consideration of ASE is necessary to
generate an RSOA-FCL lasing field that is self-seeded from the ASE and also to accurately account
for RSOA gain saturation.

Figure 3-2. RSOA simulation schematics where the counter propagating signal and ASE fields and carrier density are spatially
resolved. The symbols are defined in the text.

The carrier density rate equation is given by

N ( z, t ) I
  Rspon ( z, t )  Rstim ( z, t ) (3.1)
t qV

where N ( z, t ) is the carrier density and I is the bias current. Rspon and Rstim represent carrier
depletion due to spontaneous and stimulated emission, respectively. The expression for Rspon is
given by

N  z, t 
Rspon  z , t  
C (3.2)
 AN  z , t   BN  z , t   CN  z , t  ,
2 3

52
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

where  C is the effective carrier life time which is related to the nonradiative recombination
rate A , spontaneous recombination coefficient B and Auger recombination coefficient C .
Rstim includes the carrier depletion due to the signal rs ( z , t ) , and the ASE rase ( z , t ) and is given
by

1   rs  z , t   rase  z , t  
Rstim  z, t  
dw hc 1   NLGC Ptot  z , t  

 
2

rs  z , t   k g k  N   Ek  z , t   Ek  z , t   (3.3)

k

 
2

rase  z , t   2   j g j  N   j  z, t    j  z, t  
k

where we define E  z, t  and Ek  z, t  as the forward and backward propagating complex

k
amplitude of the k th longitudinal mode. Similarly  j  z , t  and  j  z , t  are the fields of the j th
ASE mode. The modes can be further decomposed using amplitude and phase coefficients such
that

Ek  z, t   Ek  z  e j2k t

(3.4)
 j  z, t    j  z  e
j2 j t
,

where  , dw  NLGC and h are the confinement factor, effective area of the RSOA active
region, the nonlinear gain coefficient and the Plank’s constant, respectively. The indexes k and
j refer to the longitudinal modes for the signal and ASE, respectively. For the considered
simulation bandwidth of 4 nm, this paper assumes a linear gain as

g  N  z , t    a0  N  z , t   N 0  (3.5)

where a0 is the differential gain coefficient and N 0 is the carrier density at transparency.
The wave equation for the counter propagating signal fields is written as

Ek  z, t  1 Ek 1  gk ( N ) 
 z, t   Ek  z, t   int   1  j H  (3.6)
z vg t 2  1   NLGC Ptot  z, t  

where Ptot  z , t  is the total power at the position z and time t given by

Ptot  z , t     Ek  z , t   Ek  z , t       j  z , t    j  z , t  .

2 2

    (3.7)
k   k 

 int and  H are the internal scattering loss of the active region and the linewidth enhancement
factor or the Henry factor, respectively. Since the lasing field extends over a spectral width of at
most 100 GHz within the OBPF, this paper assumes a constant  H -factor. Such an assumption
is reasonable and allows for a sufficiently accurate SOA simulation [189, 203].
The equation for the ASE field is given by.

53
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

 j  z, t 
  int  g j  N  z , t     j  z , t   Rsp,j  N  z , t  
1

z 2
(3.8)

In (3.8), the first part of the right hand side defines the amplification of the ASE as it propagates,
whereas the second part is the spontaneous emission rate Rsp given as

 x  jx2 
Rsp,j  N   nsp g j  N  z , t   M FSR h j z 1  1  (3.9)
 2 

The spontaneous emission is assumed to have white Gaussian noise behavior, [204]. x1 and x2
are Gaussian distributed random number sequences with mean zero and a variance of one. nsp
and  j are the inversion factor and the frequency of the j th ASE mode, respectively. The
spontaneous emission is calculated within the frequency range M  FSR where  FSR is the free
spectral range (FSR) of the RSOA-FCL and M is an integer number. In the next subsection we
will discuss both.
To represent the propagation of the signal and ASE, the time t and position z change with steps

l z
z  , t  (3.10)
m vg

where l is the length of the RSOA active region with effective refractive index ng . vg  c ng is
the group velocity, where c is the speed of light in vacuum. The carrier density is numerically
solved using the modified Euler method as in [203, 205].
The following boundary conditions are applied at the output and reflective facets of the RSOA
shown in Figure 3-2.

Ek (0, t )  Rl Ek (0, t )  RSOA 1  Rl  Ein,k (t )

Ek (l , t )  Rh Ek (l , t )
(3.11)
 j (0, t )  Rl  j (0, t )
 j (l , t )  Rh  j (l , t )

Ein,k (t ) is the input field for the k th signal mode whereas Rl and Rh are the power reflectivity
of the output and the mirror facet, respectively. The coupling from fiber to the RSOA is given by
RSOA . The output signal and ASE are

Eout,k (t )  RSOA 1  Rl  Ek (0, t )

(3.12)
 out,j (t )  RSOA 1  Rl  j (0, t ).

3.1.3.2 Cavity Model

The simulation setup of the RSOA-FCL is shown in Figure 3-3. The overall cavity length includes
the distribution fiber and the RSOA. An optical bandpass filter (OBPF) mimics the WGR at the RN
and performs frequency selection whereas a variable optical attenuator (VOA) represents the
total insertion losses of all the passive constituents in the cavity.

54
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Figure 3-3. The reflective SOA fiber-cavity laser (RSOA-FCL) formed by the RSOA mirror and the remote node mirror. The cavity
consists of the distribution fiber, an optical band pass filter (OBPF) and variable optical attenuator (VOA).

The frequency response of the fiber assuming only group velocity dispersion (GVD) and
attenuation is

 2 2 
H fbr ( )  exp  0.1152 dB/km  j D o   o   L  (3.13)
 c  

Where  dB/km , D and 0 are the fiber attenuation, chromatic dispersion (CD) parameter and
the center wavelength, respectively. The OBPF has a super Gaussian shape. Combining the
Gaussian filter’s frequency response and VOA attenuation gives

 1    0  2 n 
H OBPF ( )   VOA exp     (3.14)
 2    
 

Where

3dB
  (3.15)
22 n ln  2 

With 3dB and n being the 3 dB bandwidth and the order of the Gaussian filter, respectively,
whereas  VOA is the single direction attenuation of the VOA.
The ASE and signal of the laser, which are now transmitted into the network are

 out (t )  cpr F 1 H fbr ( ) H OBPF ( )   out 

(3.16)

Eout,tx (t )  cpr F 1 H fbr ( ) H OBPF ( )  Eout , 
where cpr is the ratio of the out coupled power. Eout and  out are the Fourier transforms of
the output signal and ASE from the RSOA, respectively. F 1   stands for inverse Fourier
transform.
Now that the resonator cavity is explained in detail, we have to discuss the formation of the
multiple longitudinal modes. The FSR of the modes FSR is given by

c
FSR  (3.17)
2  nfbr L  ng l 

where nfbr is the refractive index of the optical fiber. Since the RSOA-FCL extends up to tens of
kilometer, the FSR is in the order of kHz. In other words for an exemplary OBPF bandwidth of 100
GHz we expect excited modes in the order of 106 . Thus solving the wave equation for all modes
would be impractical. Similar to [202] we bundle neighboring modes to relieve the simulation

55
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

load. We introduce M as the integer number of modes bundled together and for which a single
equation is solved. The signal and ASE mode bundle frequencies, are then given by

k  0  kM FSR , k  1, 2,  N m  1 2 (3.18)

where N m and  0 are the total number of modes and the center frequency respectively.

Simulation Results
This section presents the simulation procedure and the results obtained for the RSOA-FCL.
Initially we describe the ASE self-seeding within the cavity. Then, we report on the buildup of the
laser emission and its spectral and temporal properties when the RSOA is unmodulated and
modulated, respectively.

3.1.4.1 Simulation Procedure

The iteration steps of the signal and ASE fields in the RSOA-FCL are summarized in Figure 3-4. In
a first iteration step, i  1 shown in Figure 3-4(a), the RSOA just emits ASE as shown in
Figure 3-4(d). Subsequently this ASE light is traveling through the cavity undergoing dispersion
and being filtered by the WGR. A monitor shows the spectrum in the remote node, which can be
seen in Figure 3-4(e).
In the second iteration, i  2 shown in Figure 3-4(b), the ASE in the fiber cavity is reflected at
the RN mirror and an attenuated fraction is fed back into the fiber, the WGR and into the RSOA
as an input signal field, shown in Figure 3-4(g). This signal is now amplified by the RSOA and
passes through the fiber and OBPF to be coupled out, as depicted in Figure 3-4(f). Here, the ASE
spectral components with index j are coupled to the signal with index k , i.e. j  k . The initial
conditions for the first two iterations are given as
1
Ein, k 0
(3.19)
k   H fbr  k  H OBPF  k  
 2 1 2
Ein, k  RRN (1  cpr )   out,

where RRN is the remote node mirror reflectivity.

The RSOA output passes through the fiber and is filtered twice, thus the fiber and OBPF responses
are squared as in (3.20). After the second iteration, the signal power significantly depletes the
carriers and the ASE power noticeably diminishes. Therefore, for i  3 , we can neglect the
coupling between the signal and ASE. The input signal field is then the RSOA output signal result
from the previous iteration after passing through the channel twice, as per Figure 3-4(c). It is
given by

Ein,i k  RRN (1  cpr )  Eout, k   H fbr k  H OBPF k   .

 i 1 2
(3.20)

56
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Figure 3-4. Simulation procedure followed to simulate the modes in the RSOA-FCL: (a) First iteration i  1 : the RSOA emits ASE
shown in (d) while the input signal is zero (black arrow). The ASE coupled out of the RSOA-FCL and having traveled once through
the fiber cavity and its WGR filter is shown in (e); (b) Second iteration i  2 : the scaled version of the ASE is fed into the RSOA
as a signal, shown in (g). The output signal of the RSOA-FCL laser is shown in (f); (c) After the 3rd step i  1 the signal is again
fed into the cavity and a new iteration is started.

3.1.4.2 Simulation of Unmodulated Operation

Here, we investigate the RSOA-FCL operation when a DC bias current is applied to the RSOA and
the modulation is disabled. This allows us to observe the laser spectrum and intensity waveform
and to initially judge if such a highly multi-mode source could be used to transmit information at
all. The RSOA and cavity parameters used for the system simulations are listed in Table 3-1 and
they are based on [158] and further fitting to experimental results.
At first, we consider an RSOA-FCL with a short single mode fiber (SMF) of 1 m length and a
dispersion parameter D  16 ps/(nm·km). The evolution and the narrowing of the spectral width
as the signal iterates in the cavity is depicted in Figure 3-5 The output spectra after the 10th and
100th iteration are shown in the insets. The 10 dB spectral width is recorded after every 5 th
iteration up to the 100th iteration. The RSOA is biased with a 120 mA DC current and the overall

57
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

bidirectional cavity loss is 16 dB. This includes the insertion losses when twice passing through
the WGR, the remote node mirror and the fiber-to-RSOA coupling losses.

Figure 3-5. Evolution of the signal spectral width versus iteration index in the 1 m long SMF cavity of the RSOA-FCL; the insets
show sample spectra for the 10th and 100th iteration with OBPF frequency response (blue line).

It can be seen that initially the spectral width of the laser is determined by the filter bandwidth,
as plotted in Figure 3-4(e). However, after a few iterations, the laser field builds up and mode
competition leads to a narrowing of the spectrum. Figure 3-5 shows how the spectral width of
the long cavity laser narrows relatively quickly within the first 40 iterations. The Gaussian OBPF
with a 3 dB bandwidth of 100 GHz attenuates the outermost modes, which die out after only a
few iterations. The RSOA-FCL gradually settles with less than 20 GHz spectral width and a red
shifted spectral peak relative to the center of the OBPF. The redshift can be seen in the insets of
Figure 3-5. One might think that the laser spectra should be centered at the OBPF transmission
maximum; however, this is not the case. The lasing spectrum extends to the redshifted transition
band of the OBPF due to the Bogatov effect [206]. The dominance of the redshsifted components
implies that they receive more gain and thus lasing at these modes dominates. A further redshift
of the peak at the low frequency edge of the WGR passband is prevented by an increased
attenuation [207].
In addition, a stable output power waveform is observed because RSOA gain saturation
suppresses fast intensity fluctuations of the seeding ASE and beating between the laser modes.
This will be discussed in more detail below.
The simulation parameters for the RSOA listed in Table 3-1 were fitted to the actual data by
comparing the simulation and experimental results of the power versus bias current (P-I) plot.
The P-I plot of the RSOA-FCL with 6, 11, and 16 dB overall cavity loss is depicted in Figure 3-6.
The P-I plot with a 16 dB loss is shown for both simulation and experiment. The experimental
setup is modelled as the scheme depicted in Figure 3-3, the only difference being that the OBPF
is a Finisar WaveShaper that was programmed to mimic a 110 GHz Gaussian bandpass filter. As
expected, the output power diminishes for larger cavity losses. The threshold current is observed
around 50 mA both in the simulation and in the experiment.

58
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Table 3-1. RSOA-Fiber laser component parameters [158]

Symbol Quaantity Value

a0 Differential gain 2x10-20 m-1

 int Internal scattering loss 5000 m-1
H Linewidth enhancement factor 4
 NLGC Nonlinear gain coefficient 3 W-1

A Nonradiative recombination 4x108 s-1

B Radiative recombination 1x10-16 m3s-1

C Auger Recombination 1x10-42 m6s-1
nsp Inversion factor 2
ng Effective group refractive index of the
3.45
RSOA
N0 Transparency carrier density 5x1023 m-3

l RSOA active region length 1x10-3 m

w RSOA active region width 2x10-6 m
d RSOA active region height 0.1x10-6 m

 Confinement factor 0.2

Rl Output facet reflectivity 1x10-6
Rh Mirror facet reflectivity 0.2
3dB OBPF bandwidth 100 GHz

n OBPF order 2
 dB/km Fiber attenuation 0.2 dB/km
0 Center wavelength 1550 nm
RRN RN mirror reflectivity 0.86

m Spatial sections 20
Nm Total number of modes 966
RSOA RSOA coupling loss 3 dB
cpr Cavity network coupling ratio 0.2

59
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Figure 3-6. Simulation and experimental plot of the output power versus bias current for the RSOA-FCL with 1 m SMF cavity and
6, 10 and 16 dB overall cavity losses. The output power reduces for larger cavity loss.

Subsequently, the long fiber cavity that may extend up to tens of kilometers results in a large
number of longitudinal modes within the 100 GHz bandwidth of the OBPF. Due to the broad
spectrum, there is a strong effect of CD on the system performance. We then investigated two
types of cavities with SMF and dispersion shifted fiber (DSF), where the SMF was simulated with
D  16 ps/(nm·km) and the DSF was simulated with D  0 ps/(nm·km). The signal spectral
width as a function of the fiber cavity length for SMF and DSF are shown in Figure 3-7(a). During
the simulation, the program was executed 50 iterations beyond the steady state operation point
and then the mean of the 50 spectral width values was taken. It can be seen how the signal
spectrum abruptly broadens for longer dispersive cavities due to an accumulation of CD, while it
remains constant for the DSF cavities. The finding agrees well with the measurement results
shown by the black triangles for SMF and blue star for the DSF in the same plot.
Figure 3-7(b)-(e) exemplify the CD induced power fluctuations and the spectral broadening.
Figure 3-7(b) and (d) show the temporal power and spectra for 10 m SMF and DSF cavities,
respectively. In both cases, once the laser field has built up, the intensity remains constant with
small fluctuations only. The “small” thereby refers to the fact that it is small given the highly
multimoded field. When the SMF cavity is short, the CD is insignificant and the nonlinear RSOA
with strong gain saturation suppresses the power fluctuation of the seeding ASE. Thus, the power
converges to an almost constant value with a narrow spectrum as for the DSF case. The spectra
for the two cavities are similar to the steady-state spectrum of the 1 m SMF cavity shown in the
inset of Figure 3-5. When the cavity is extended beyond 0.5 km SMF, the CD alters the delicate
phase relation between the multiple modes that is required to achieve a constant intensity. Once
these modes reach the RSOA, the modes interfere constructively and destructively creating
strong power fluctuations that cannot be suppressed by the RSOA gain saturation; and thus
leading to increased spectral broadening as depicted in Figure 3-7(c). This CD induced mode
partition noise substantially increases the relative intensity noise (RIN) of the laser and renders
the laser useless for direct modulation of an upstream signal. However, when using a 1 km DSF
cavity the phase relation of the modes is mostly preserved and a relatively flat output power with
a narrow spectrum is obtained, as shown in Figure 3-7(e).
It is instructive to plot the total output power of the RSOA-FCL against the power in different
spectral ranges that contribute to this total power. In Figure 3-8(a) and (b) we show the power
of 90 longitudinal mode bundles and the total power for a 10 m SMF cavity RSOA-FCL.

60
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Figure 3-7. Simulated and experimental spectral width as a function of the cavity length for SMF (red line) and DSF (blue line). The
subfigures show simulated output powers with time (left) and spectra (right) for (b) 10 m SMF, (c) 1 km SMF, (d) 10 m DSF and (e)
1 km DSF cavities, respectively.

The power in a mode bundle is obtained by filtering the output field with a Lorentzian filter (517
MHz full width at half maximum) centered at the modal frequencies. In Figure 3-8(a), the colored
lines show how the individual mode powers fluctuate due to strong mode competition while the
total power (blue line) is relatively constant. Figure 3-8(b) then shows that the fluctuations in the
total output power remain low even if the laser is directly modulated. The modes still compete
however; the output power now depends on the extinction ratio of the modulated bias current.

61
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Figure 3-8. Plot for the power of individual longitudinal mode bundles (colored lines) and the total power (blue line) for 10 m SMF
cavity RSOA-FCL; (a) without direct modulation and (b) with direct modulation at 2.5 Gbit/s

The cavity loss is another important parameter in the characterization of the RSOA-FCL. To
observe the impact of cavity losses alone, we simulate a cavity with 1m long SMF having
negligible CD and the output spectra for 6, 11 and 16 dB overall cavity losses are depicted in
Figure 3-9. For low cavity losses or strong feedback into the RSOA, the spectrum is broadened
due to the interaction between the strong multiple modes reflected back into the RSOA.

Figure 3-9. Output signal spectra of RSOA-FCL with 1 m SMF cavity and 6, 11 and 16 dB overall cavity losses. The low cavity loss
has broader spectrum due to the stronger multimode interaction.

62
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

3.1.4.3 Simulation of Directly Modulated Laser

This subsection presents the performance of the RSOA-FCL when used as a directly modulated
transmitter. After the CW output signal reaches steady state a modulated bias current is supplied
to the RSOA to encode data for upstream communications. A non-return to zero (NRZ) signal is
encoded on the bias current, and for every iteration, a new pseudo random bit sequence (PRBS)
is generated. This allows us to show explicitly that the RSOA with saturated gain simultaneously
suppresses the recirculating residual intensity modulation from upstream communications in a
previous iteration while encoding new information onto the field for new upstream transmission.

Figure 3-10. Simulated ((a) and (c)) and measured ((b) and (d)) eye diagrams for 2.5 Gbit/s direct modulation of the RSOA-FCL
with; (a) and (b) 10 m SMF cavity; (c) and (d) 420 m SMF cavity.

Figure 3-11. Simulated eye diagrams for 2.5 Gbit/s direct modulation of the RSOA-FCL with; (a) 10 km SMF cavity and (b) 10 km
DSF cavity. The CD in the 10 km SMF results in complete closure of the eyes.

The eye diagrams for a 2.5 Gbit/s modulated RSOA-FCL are depicted in Figure 3-10. The RSOA-
FCL is biased at 100 mA, and the modulation component swings between 70 and 170 mA. The
minimum current is 20 mA above the laser threshold. This corresponds to 6 dB extinction ratio
in the output laser power. The ideal operation point that allows erasure of previous upstream
signal is at a reduced extinction ratio as outlined in more detail in [158]. The bidirectional cavity
loss is 11 dB. Figure 3-10(a) and (b) show simulation and experimental eye diagrams for a 10 m
SMF cavity whereas (c) and (d) show simulation and experimental eye diagrams for a 420 m SMF
cavity, respectively. The eyes are open since the cavities are relatively short and both CD and loss
are low. Moreover, now the RSOA input for the i th iteration is the modulated output signal from
the  i  1 iteration having different PRBS. To encode a new data on the signal field, the
th

modulation from the previous iteration must be suppressed. Thus, the open eyes in Figure 3-10
are clear indications that the simultaneous modulation cancellation and remodulation
operations are working perfectly.

63
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

In Figure 3-11(a), the simulated cavity is extended to 10 km of SMF and it can be seen that the
eye is completely closed. The accumulated CD impairment is responsible for the poor
performance. On the other hand, if 10 km DSF replaces the SMF fiber, open eyes are obtained as
shown in Figure 3-11(b), though with less power due to the additional fiber attenuation.
Figure 3-12 shows the change of the signal spectral width with the bit rates for 10 m SMF and 1
km DSF. The plot illustrates that there is some small spectral broadening with higher bit-rates.
This may be due to the chirp as a result of direct modulation and suggests that the reach of RSOA-
FCL transmitter might be slightly reduced when increasing the bit-rate [198].

Figure 3-12. Signal spectral width of the RSOA-FCL with 10 m SMF and 1 km DSF cavities modulated at 2.5, 5 and 10 Gbit/s.

Conclusion
This paper has presented a theoretical model for the RSOA fiber cavity laser as a colorless
transmitter for next generation WDM-PON access. The ASE self-seeded laser cavity was found to
narrow the spectrum to a smaller subset of modes with an almost constant output power
allowing for a good operation with short fiber cavities. However, for longer and in particular for
cavities with large chromatic dispersion, the output power strongly fluctuates since CD breaks
the delicate phase relation between the multiple modes that is required to achieve constant
power. The RSOA then can no longer suppress these fast fluctuations and thus the obtained
directly modulated signal presents very poor performance, as evidenced by an almost complete
eye closure. However, when the dispersion in the cavity is sufficiently small, as with a DSF, clear
open eye diagrams are obtained at 2.5 Gbit/s. Thus proving that both direct modulation and
cancellation of the modulation part of the recirculating signal could occur simultaneously even
for cavity lengths of the order of 10 km. Self-seeded RSOA-FCL, as evidenced by this paper, are
thus promising colorless sources for beyond 2.5 Gbit/s access transmission when operated with
proper CD management.

64
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

3.2 Alternative Amplified RSOA-FCL

Simulation results for the standard self-seeded (SS) RSOA-FCL in section 3.1.4 have shown that
chromatic dispersion of the distribution fiber cavity is decisive to the performance at 2.5 GBit/s.
Here, an alternative scheme called amplified self-seeded (ASS) RSOA-FCL is compared with the
standard SS RSOA-FCL for up to 10 GBit/s modulation in simulation and experiment. In the ASS
RSOA-FCL a second RSOA is positioned in the cavity to form the Fabrey-Perot cavity. The 2nd RSOA
in the ASS RSOA-FCL offers extra gain and also offers additional noise suppression.
The simulation procedure and device parameters used in this section are identical to the ones
presented in [193].
This work has been presented at the 2014 conference on lasers and electro-optics (CLEO) and it
is copied here word by word [208].

65
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Colorless Self-Seeded Fiber Cavity Laser Transmitter for

WDM-PON
Simon A. Gebrewold, Lucia Marazzi, Paola Parolari,
Marco Brunero, Romain Brenot, David Hillerkuss,
Christian Hafner, Juerg Leuthold

In Conference on Lasers and Electro-Optics (Optical Society of

America, 2014), paper Stu1J.4 [208]

Introduction
Colorless, self-tuning sources, i.e. able to assign the source wavelength automatically and
passively, are attractive as cost-efficient transmitters for widespread deployment of wavelength
division multiplexed (WDM) passive optical networks (PONs). Such transmitters can be obtained
by externally seeding directly modulatable injection locked Fabry-Perot laser diodes [209], or
reflective semiconductor optical amplifiers (RSOAs) [210], using a broadband light source at the
optical line terminal (OLT). Recently self-seeded (SS) laser sources have been proposed that no
longer need an external source. Instead they use the amplified spontaneous emission (ASE) from
the RSOA itself as the seeding light [123], and the distribution fiber (DF) constitutes the cavity
length. Unfortunately, the power budget is quite tight, particularly when fiber cavities are long.
More recently, an amplified self-seeded (ASS) configuration has been demonstrated that
overcomes the power budget limitation by inserting a second RSOA in the cavity at the OLT [211].
This paper discusses the optimum RSOA fiber cavity laser (FCL) configuration required to obtain
a directly modulatable SS and ASS-FCL WDM PON transmitter. Further, we show that with
chromatic dispersion (CD) control, the access DF can be extended despite the large number of
longitudinal modes.

WDM-PON System Architectures

Figure 3-13 shows two different access network architectures. In Figure 3-13(a), the WDM-PON
is configured with an SS-FCL upstream transmitter, which has a single RSOA as the gain medium.
The resonator cavity is formed by the RSOA mirror at the ONU, by the Faraday rotator mirror
(FRM) at the remote node (RN) and by the DF. Figure 3-13 (b) shows the WDM PON having an
ASS-FCL transmitter with two RSOAs. The cavity consists of both the DF and feeder fiber between
the RSOA mirrors at the ONU and OLT. The cyclic waveguide grating router (CWGR) at the remote
node allocates the operating wavelength of both the SS and ASS-FCLs. A Faraday rotator (FR) is
used to cope with birefringence of the fiber cavities when using RSOAs with highly polarization
dependent gain [125].

66
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Figure 3-13. WDM PON architectures with (a) self-seeded and (b) amplified self-seeded FCL upstream transmitters; OLT (optical
line terminal), CWGR (cyclic waveguide grating router), RN (remote node), DF (distribution fiber), FRM (Faraday rotator mirror),
FR (Faraday rotator)

Simulation and Experimental Results

To simulate the performance of the SS and ASS-FCL a bidirectional bulk SOA model is used [202].
This model solves the wave equations for the counter propagating multimode signal and ASE
field. The ASE is assumed to have a white Gaussian noise property. The CWGR has a Gaussian
bandpass filter shape. A more in-depth report on the simulation procedure will be given
elsewhere due to space constraints in this submission.
An important aspect is the spectral width of the multimode FCL. Figure 3-14(a) plots the spectral
width versus cavity round trip count for both SS and ASS-FCLs with a 1 m single mode fiber (SMF)
cavity. Initially, when the laser is turned on, the spectral width is determined by the CWGR
bandwidth. After a few round trips, the RSOA saturates and suppress the fast power fluctuations
of the seeding ASE, and the laser settles with a relatively constant power and a redshifted narrow
spectrum as shown in the insets of Figure 3-14(a). For the ASS-FCL, the second RSOA at the OLT
contributes additional gain and noise suppression.
Next, we consider the impact of the fiber cavity CD on the spectral and temporal power of both
SS and ASS-FCLs. Figure 3-14(b) plots the spectral width of SS and ASS-FCLs with SMF and
dispersion shifted fibers (DSF) cavities of different length. For both SS and ASS-FCLs, the spectral
width broadens when increasing the SMF cavity length. However, the spectral width shows
negligible change for DSF cavities. These results are in qualitative agreement with the
experiment. Figure 3-14(c) top shows that the temporal powers for SS-FCLs with 10 m SMF and
DSF cavities are relatively constant, and that they are similar to the 1 m SMF result in the inset
of Figure 3-14(a). If the cavity length is increased to 1 km, the SMF cavity shows strong power

67
Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

fluctuation while the DSF cavity power stays relatively constant, see Figure 3-14(c). The power
fluctuation or mode partition noise of the long SMF cavity is due to the accumulated CD that
alters the phase of the multiple modes, which are added to give the total power. The saturated
RSOA can no more suppress such fluctuations.

Figure 3-14. (a) Spectral width at every 10th round trip for SS and ASS-FCLs with 1 m SMF cavity; the insets show the temporal and
spectral power at the 100th roundtrip. (b) Simulated and measured Spectral width for SS and ASS-FCLs with SMF and DSF of
different length (c) Temporal power for SS-FCL with 10 m (top) and 1 km (bottom) SMF and DSF cavities

To assess the transmitter performance, we directly modulate the SS and ASS-FCLs with a non-
return to zero (NRZ) signal. Figure 3-15 shows simulated and measured eye diagrams for both SS
and ASS-FCLs. The SS-FCL with 420 m SMF modulated at 2.5 Gbit/s results in open eye diagrams
as shown in Figure 3-15(a). This proves that the RSOA is simultaneously and efficiently amplifying,
cancelling residual modulation and encoding new data onto the recirculating optical field.
However, at 10 Gbit/s, a 1 km SMF cavity results in completely closed eyes due to both
accumulated CD and limited modulation cancellation bandwidth of the RSOA Figure 3-15(b).
However, when using a 9.9 km DSF the ASS-FCL supports transmission at 10 Gbit/s, as can be
seen from the open eye diagrams in Figure 3-15(c). The improved performance is due to the
additional modulation cancellation provided by the second RSOA and the modest cavity CD.

68
 Reflective Semiconductor Optical Amplifier Fiber Cavity Laser (RSOA-FCL)

Figure 3-15. Simulated (red) and measured (colorgrade) eye diagrams for 2.5 and 10 Gbit/s directly modulated SS and ASS FCLs

Conclusion
We showed that mode partition noise due to CD within the fiber cavity and bandwidth limitations
of the RSOA impair the performance of self-seeded fiber cavity lasers. By placing a second RSOA
in an extended DSF cavity, as in the ASS-FCL transmitter, 10 Gbit/s upstream transmission is
possible for next generation access networks.

69
4 Comparison of RSOA-FCL with Alternative RSOA
Based Colorless Transmitters
This chapter answers qualitatively how the self-seeded RSOA-FCL compares with alternative
RSOA based colorless transmitters. In chapter 1, simulations have illustrated how placing an
RSOA at the ONU and a simple mirror at the RN of a WDM-PON form a Fabrey-Perot laser. At the
appropriate operating condition of the RSOA-FCL, the output is characterized by a narrower
spectrum and constant output power. This has enabled the RSOA-FCL to achieve 2.5 GBit/s NRZ
ASK modulation.

It is also necessary to specify quantitatively the value added by the RSOA-FCL in comparison with
alternative RSOA based colorless transmitters. Therefore, this chapter presents simulation and
experimental comparisons of RSOA-FCL with two main alternative sources namely an ASE
spectrum-sliced source and an externally seeded RSOA. The considered comparison parameters
are output power, signal-to-noise ratio (SNR), relative intensity noise (RIN), frequency response,
and transmission performances.

This chapter is organized as follows. After an introduction in section 4.1, the three transmitter
schemes are proposed for a WDM-PON deployment in a mobile fronthaul (MFH) scenario as
discussed in section 4.2. The operating principle of an RSOA-FCL and the modelling approach are
revised in section 4.3 and 4.4, respectively. Section 4.5 is dedicated to a comparison of the RSOA-
FCL with an ASE spectrum-sliced and externally seeded sources in continuous wave (CW)
operation mode. High-speed data transmission capacities of the three schemes are analyzed in
section 4.6.
This chapter has been published in the Applied Sciences Journal in December 2015, and it is
directly copied here. Some variables have been adapted to keep the same nomenclature in this
document.

71
Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

Self-Seeded RSOA-Fiber Cavity Lasers vs ASE Spectrum

Sliced or Externally-Seeded Transmitters – A Comparative
Study
Simon A. Gebrewold, Romain Bonjour, Sophie Barbet,
Anaelle Maho, Romain Brenot, Philippe Chanclou,
Marco Brunero, Lucia Marazzi, Paola Parolari,
Angelina Totovic, Dejan Gvozdic, David Hillerkuss,
Christian Hafner, Juerg Leuthold

Applied Sciences, vol. 5; Issue 4; 2015 [212].

4.1 Introduction
The development of low-cost colorless sources has become a major research topic. They are
needed to make the widespread deployment of wavelength division multiplexed passive optical
networks (WDM-PONs) practical [213, 214]. Relatively new alternatives are self-seeded reflective
semiconductor optical amplifier fiber cavity lasers (RSOA-FCLs) [194]. The RSOA-FCLs’ colorless
feature emanates from the automatic and passive self-tuning of their emission wavelength by
only a WDM filter positioned at a remote node (RN). The colorless feature allows for the mass
production and installation of identical upstream devices at all optical network units (ONUs) of a
WDM-PON. Having only one common source at the ONUs will lower the cost of WDM-PONs,
which has been a burden to a widespread deployment [215]. In addition, a passive wavelength
allocation by the PON infrastructure simplifies the wavelength management process. Thus, the
availability of low-cost colorless sources, such as RSOA-FCLs allows for the implementation of
WDM-PONs that offer high bandwidth. These could for instance be used for mobile fronthaul
networks (MFHs), which will be discussed in the next subsection [46, 213].

Quite a few other colorless transmitter schemes are currently under discussion. The two key
contenders are:
 Amplified spontaneous emission (ASE) spectrum sliced schemes: These are the
simplest low cost solutions available. Directly modulated broadband light sources such
as light emitting diodes (LEDs) [69, 70, 72, 73] or RSOAs [74, 216] can be used for
upstream (US) transmission. As an example, Figure 4-1(a) shows an RSOA based ASE
spectrum sliced WDM-PON. The upstream data is directly encoded onto the
broadband ASE of the RSOAs, which are placed at the ONUs. The waveguide grating
router (WGR, WDM MuxDeMux) at the RN slices the modulated ASE spectrum and
sends a sliced band upstream. The US spectral band from each ONU is determined by
the WGR port to which the individual ONUs are connected. Even if the low cost of LEDs
and RSOAs make this scheme attractive, their modulation speed are limited to 1 Gbit/s
over less than 10 km of transmission reach. The ASE amplitude noise and thus the high
relative intensity noise (RIN) are a limiting factor. There is also a high filtering power
loss during spectral slicing. In addition, the broad bandwidth of each spectrum slice
makes the scheme highly sensitive to chromatic dispersion.

72
 Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

 Externally-seeded RSOA schemes: These offer higher reach and speed at a higher
investment. The idea is to use a bank of lasers at the central office (CO), one for each
ONU, which send CW light downstream (DS) to the ONUs, as presented in Figure 4-1(c)
[97, 172, 217-219]. The WGR splits and distributes the different wavelengths to the
respective ONUs. The RSOAs at the ONUs amplify and directly encode data onto the
seeding CW light and send them back upstream. The amplification of the DS CW light
by the RSOAs compensates the transmission power loss by offering higher link power
budget. The chromatic dispersion penalty for this scheme is also more relaxed than
that of the spectrum sliced scheme. Of course, all the improvement comes at the price
of placing relatively expensive lasers at the CO.
The question is then if there is a colorless transmitter that is more economical than an externally
seeded RSOA scheme and still offers a better performance over the ASE spectrum sliced source.
We believe this question is answered with what is called the self-seeded RSOA-FCL transmitter
shown in Figure 4-1(b) [195, 198, 208, 220, 221]. In contrast to the externally-seeded scheme,
the RSOA-FCL does not need extra laser sources. As the name indicates, the seeding light is the
ASE emitted by the RSOA itself. A simple mirror at the RN reflects part of the emitted ASE back
to the RSOA. This cuts the cost significantly. The RSOA at the ONU and a mirror at the RN form
the resonant cavity that is imbedded in the distribution fiber trunk of the PON. The emission
wavelength of the laser is automatically tuned by the port of the WGR to which the ONU is
connected. The laser characteristics leads to a narrower spectrum, lower RIN, and a higher power
compared to the ASE spectrum sliced scheme, thus improving the transmission performance.
In this paper, we compare the self-seeded RSOA-FCL with the ASE spectrum sliced and the
externally seeded RSOA schemes, shown in Figure 4-1. Both simulation and experimental results
evidence that the RSOA-FCL outperforms the ASE spectrum slice scheme. With a simple
architecture and low cost, the RSOA-FCL offers more power, lower RIN, narrower spectrum and
~5 dB more signal to noise ratio (SNR). The externally seeded scheme delivers an additional ~5 dB
SNR over the RSOA-FCL. Yet, this is paid for through the expensive extra lasers needed. Finally,
we perform 5 and 10 Gbit/s transmission experiments using these schemes. The results show
that the RSOA-FCL is a cost-efficient solution with a performance in between of the ASE spectrum
sliced and the externally seeded source. Thus, the RSOA-FCL is a practical colorless source for
short-reach WDM-PONs with transmissions up to 10 Gbit/s.
The paper is organized as follows. First, we use a mathematical model [193] to describe and
compare the ASE spectrum sliced and RSOA-FCL. Next, we experimentally characterize the three
schemes by measuring output power, spectrum and RIN. Finally, the performance of the three
are compared by measuring their SNR and data transmission up to 10 Gbit/s.

73
Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

Figure 4-1. Scenarios showing how the three sources would be integrated in a WDM-PON. (a) RSOA ASE spectrum sliced scheme:
An upstream (US) signal can be directly modulated onto the ASE of the RSOAs at the optical network units (ONUs). The waveguide
grating router (WGR) slices the ASE into bands and sends them upstream. (b) Self-seeded RSOA-FCL: A lasing resonant cavity,
shown by the yellow shading, is formed between the RSOA at the ONU and a mirror at the remote node (RN). The US signal is
encoded onto this multimode laser signal. (c) Externally seeded RSOA: A bank of lasers at the central office (CO) send cw light
downstream (DS). The WGR distributes the seeding light to the respective ONUs. The RSOAs at the ONU encode data, amplify the
light, and send them upstream. The corresponding measured output spectra for each scheme are plotted as insets on the left side.
FF: feeder fiber, DF: distribution fiber.

4.2 Mobile Fronthaul WDM-PON with RSOA-FCL as a Colorless

Source
As an example of where low-cost, high-speed PONs are expected to emerge in masses, we discuss
the MFH application scenario. The MFH is the fiber network interconnecting the base station
antenna with the CO. The MFH is an integral part of the cloud radio access network (C-RAN)
architecture [222, 223]. In the C-RAN the signals from various base stations are processed at one
location. The remote radio head (RRH) is placed close to the antennae.

74
 Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

Figure 4-2. Arrangement of the self-seeded RSOA-FCL in a WDM-PON based mobile fronthaul (MFH) network as an upstream
transmitter. CO: central office, BBU: base band unit, US: upstream, DS: downstream, ONU: optical network unit, RRH: remote
radio head.

The base band unit (BBUs), where the digital signal processing takes place, are then located at
the central office (CO). Indeed, this scheme requires high data rate links - a MFH - between the
RRH and the BBU. As a protocol, C-RAN uses the common public radio interface (CPRI) to transmit
high bitrate digital signal over the MFH [46, 224]. The CPRI line rates range from 614.4 Mbit/s for
the CPRI option 1 to 10.14 Gbit/s for CPRI option 8 [224]. These line rate requirements of the
CPRI are very high compared to the transmitted radio signal bandwidth. This is because in CPRI
the analog radio signals are sampled at high-speed and transmitted as digital signals from the
RRH to the BBU. For example an up to 10 Gbit/s link is needed to transmit a 20 MHz radio signal
[225]. Therefore, a simpler solution would be to transmit signals by WDM-PONs that rely on cost
efficient transmitters, such as self-seeded RSOA-FCLs.

Figure 4-2 shows a schematic of a MFH link. The CO consists of a cluster of BBUs. The multiple
RRHs may be offset by up to 5 kilometers from the CO [213]. A WDM-PON may now be used to
build the MFH that links the multiple RRHs to the BBUs at the CO. The ONUs at the RRHs consist
of a downstream (DS) receiver (Rx) and an upstream (US) data transmitter. The transmitter
comprises of a RSOA which acts as a directly modulated colorless source. Lasing is obtained
through the cavity which extends from one highly reflective facet of the SOA up to another mirror
behind a WGR filter. This way the RSOA source within the mirrors forms a long Fabry-Perot laser
cavity [46, 213, 226, 227]. The color dashed lines indicate the lasing cavities.

4.3 Operating Principle of the RSOA-FCL

The operating principle of the RSOA-FCL as an US transmitter in the MFH is depicted in
Figure 4-3(a). First, for the RSOA-FCL to be used as a transmitter, the RSOA performs three
functionalities simultaneously:
1. It is used as a modulator to encode the US data onto the optical field;
2. It acts as a gain section of the Fabry-Perot fiber cavity and thus has to compensate for
the cavity losses in order to achieve lasing;
3. It nonlinearly suppresses the residual reflected modulation on the incoming reflected
light.
In the following we are going to explain in more detail the nonlinear suppression of the incoming
signal.

75
Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

Figure 4-3. Operation principle of the RSOA FCL. (a) RSOA-FCL with the three functions of the RSOA: 1) direct modulation, 2)
amplification and 3) modulation suppression. (b) Output power and (c) gain vs. input power for RSOA. The RSOA static
characteristics has two regions, a linear region where the input and output powers are linearly related (the gain is flat) and a
nonlinear region where the output power does not change for increasing input power (the gain is suppressed for higher input
power). Driving the RSOA in the nonlinear region allows operation such that modulation is suppressed.

We start the discussion with Figure 4-3(b) and (c), which show plots of the output power and the
gain, respectively, of a typical RSOA for varying input power. The shaded regions emphasize the
linear and nonlinear operation regions based on the input power. In the linear region, the RSOA
provides a relatively constant gain such that input power and output power have a linear
dependence. In the nonlinear region, the gain is decreasing for larger input power. Figure 4-3(b)
also shows how an incoming signal is either linearly amplified if its input power is weak, or
nonlinearly suppressed if its input power is strong [203]. Thus for the RSOA-FCL to suppress an
input signal, the reflected incoming signal needs to be operated in the nonlinear region of the
RSOA.

4.4 Theoretical Modelling of RSOA-FCL

In this section, we use the theoretical model of the RSOA-FCL developed in [193] to understand
the improvements brought by an RSOA-FCL compared to an ASE spectrum sliced source. A
simplified simulation schematic is given in Figure 4-4. The RSOA is modelled as a bidirectional
i  i 
SOA as described in Refs. [158, 202]. Eout,R SOA ( ) and Ein,RSOA ( ) represent the frequency
domain optical field leaving and entering the RSOA, respectively, on the i th roundtrip in the long
cavity. H fbr   is the fiber cavity frequency response that includes both the effects of chromatic
dispersion and attenuation, see Ref. [193]. The optical bandpass filter (OBPF) mimics the WGR
filter and has a super Gaussian shape.

76
 Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

Figure 4-4. Simplified RSOA-FCL simulation schematics: the RSOA mirror and a remote mirror with reflectivity RRN form the
fiber Fabry-Perot resonator cavity. The optical bandpass filter (OBPF) and the variable optical attenuator (VOA) represent the
WDM filter and the cavity losses, respectively.

It is given by H OBPF   . RRN and cpr represent the remote node (RN) mirror reflectivity and
the coupler output ratio, respectively. The overall cavity loss  VOA is determined by the variable
optical attenuator (VOA).
i 
The optical field output Eout ( ) of a signal emitted from the RSOA on the i th cavity roundtrip is
given by
i 
Eout ( )  cpr   VOA  H OBPF    H fbr    Eout,RSOA
i 
( ) . (4.1)

The reflected seeding optical field is given by

i 
( )  RRN (1  cpr )   VOA  H OBPF    H fbr     Eout,RSOA
 i 1
2
Ein,RSOA ( ) . (4.2)

The details of the model have been presented in [193].

Now we use this model in Figure 4-4 to show how placing a simple mirror behind the OBPF
provides a higher power for the RSOA-FCL compared to a simpler ASE spectrum slice scheme. In
the initial model, we consider the simplest case of a 15 m fiber cavity. We simulated the output
power of the RSOA-FCL by varying the RN mirror reflectivity RRN from -60 dB to 0 dB. A RRN of
-60 dB reflectivity is very low and thus represents the case of the ASE spectrum slicing (no mirror).
Figure 4-5 plots the output optical power for different values of RRN when the VOA is swept
from 5 to 32 dB to mimic different cavity losses.

Figure 4-5. Impact of RN mirror reflectivity RRN and cavity loss on the optical output power of the RSOA-FCL. Depending on the
RRN and the losses, the source could be an ASE spectrum sliced source (dashed blue line) or a RSOA-FCL laser (dashed red line).

It is clearly visible that the output power follows either of two lines, the dashed red or blue line.
These two dashed lines represent either of two states of operation; the lasing source state (red),

77
Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

and ASE spectrum sliced source state (blue), respectively. For RRN  60 dB (blue solid line) the
output power stays close to the blue-dashed line. As RRN increases to 0 dB, the optical power
increases and reaches the red dashed line that shows lasing. The transition from the ASE source
operation to the lasing operation occurs when the RSOA gain is able to compensate for the cavity
losses. Therefore, if the source is designed as a RSOA-FCL it will have a power advantage of ~8 dB
over an ASE sliced source.
Another difference between the RSOA-FCL is in the optical spectrum, which is narrower
compared to that of an ASE spectrum sliced source. To understand this, we followed (in
simulation) the output optical field from the time it is launched to the 80 th round trip. One
roundtrip time is given by 2nL c , where L , c and n are the cavity length, the speed of light
in vacuum and the refractive index of SMF. For this initial exemplary case with 15 m cavity, the
roundtrip time is ~0.145 ns. We then plotted the 10 dB spectral width shown in Figure 4-6(a)
(which is derived from [193]). Initially, the spectral width of the RSOA-FCL is determined by the
optical bandpass filter, which has a 3 dB bandwidth of 200 GHz. Gradually, the spectrum narrows
down to less than 75 GHz bandwidth. As lasing sets in, a large number of longitudinal modes
compete and only a fraction of the modes dominates the lasing process. The resulting narrow
spectrum makes the RSOA-FCL more robust against dispersion compared to an ASE spectral
sliced scheme. Figure 4-6(b) and (c) show sample spectra before and after the spectrum settles.
Additional simulated and measured spectral width for longer cavities are discussed in [193].

Figure 4-6. Spectral width evolution of an RSOA-FCL. The insets show the simulated spectra before and after the spectrum settles.

In the same manner we have simulated and plotted the optical power evolution as a function of
the round trips in Figure 4-7(a). It is clearly visible that the power requires 3 to 5 round trips to
reach a steady state for different bias currents [228, 229]. Initially, only filtered ASE is emitted
when the RSOA bias current is turned on. On the second roundtrip, the RSOA amplifies and re-
emits the reflected light (after passing through the cavity). This step repeats continuously and
the power builds up until it reaches a steady state when the RSOA gain saturates.
We also followed the change in the RIN as a function of the cavity round trip times, Figure 4-7(b).
For practical reasons, we do not plot the RIN as a function of the frequency but give the RIN as
an average RIN within the spectral window up to 2.5 GHz. Figure 4-7(b) indicates the RIN
evolution with the round trip, and shows that the RIN follows a similar time scale as the output
power to reach steady state. Initially the RIN is very high as the cavity just emits spectrum sliced
ASE. The dashed lines depict the initial high RIN for 50, 100 and 200 GHz OBPFs. As the power in
the cavity builds up, the RIN also goes down.

78
 Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

Figure 4-7. Evolution of (a) output power and (b) RIN with the number of cavity roundtrips. The output power builds up within the
first few roundtrips before saturating. The RIN also follows a similar evolution in time before it reach its minimum value.

This can be explained by the same concept used in the nonlinear suppression of amplitude
modulation in Figure 4-3(b) and (c). In the same manner, as the optical power of the RSOA-FCL
builds up, the RSOA is driven into the nonlinear region and amplitude noise of the seeding ASE is
suppressed. This results in a reduction of RIN with each round trip, as is shown in Figure 4-7(b).
Figure 4-7(b) also shows that for wider filter bandwidths we get lower RINs [230]. This could be
accredited to two effects. One, wider bandwidth means lower filtering loss of power and thus
the RSOA is driven into the nonlinear region more quickly and thus stronger noise suppression.
Another explanation is related to the mode partition noise of multimode lasers, In Ref. [128], it
is discussed that in case of multimode lasers, increasing the number of modes will spread the
noise over a broader spectrum thus lowering the mode partition noise and the noise spectral
density thus the RIN. In the same manner, for larger bandwidth filter in the RSOA-FCL, the
number of modes increases and in turn lowers the RIN.

4.5 Characterization
In this section, we experimentally compare the ASE spectrum sliced, self-seeded RSOA-FCL, and
the externally seeded RSOA colorless sources. We measure their optical output power, spectrum
and RIN.

Experimental Setup
Figure 4-8 shows the experimental setups for the three schemes under investigation when used
as upstream transmitters in the WDM-PON networks shown in Figure 4-1. A DC bias current I bias
drives the RSOA at the ONU. The distribution fiber (DF) links the ONU to the optical bandpass
filter (OBPF). An OBPF mimics the WGR WDM filter at the RN. For the case of the RSOA-FCL we
use a Faraday rotator mirror at the RN and a Faraday rotator (FR) in the ONU. The laser cavity is
then formed by FRM and the RSOA, Figure 4-8(b). The FRM and FRs are needed, since RSOAs
often have a polarization dependent gain (PDG). With this arrangement the forward and
backward propagating signals maintain the polarization state at the RSOA [125]. A 90/10 power
splitter sends 10 % of the power through the feeder fiber (FF) to the CO. At the CO, the signal is
pre-amplified and directly detected with a photodiode. An optical spectrum analyzer (OSA) is
used to measure the power and the optical spectrum. For the externally seeded RSOA scheme
shown in Figure 4-8(c), the seeding laser is also positioned at the CO. A circulator is used to split

79
Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

the DS cw seed and the US signal. We use a variable optical attenuator (VOA) to control the
seeding power.

Figure 4-8. Experimental setups to characterize and compare the three schemes; (a) RSOA ASE spectrum sliced scheme, (b) self-
seeded RSOA-FCL, (c) externally seeded RSOA. In all setups, the transmitter is placed at the optical network unit (ONU) and the
receiver is placed at the central office (CO). An optical bandpass filter (OBPF) mimics the WGR WDM filter at the remote node
(RN). In (b), a Faraday rotator (FR) and mirror (FRM) are placed at the ONU and the RN, respectively, for polarization control in
the RSOA-FCL shown by the yellow shading [125]. The ONU is connected to the RN and the CO through the distribution fiber (DF)
and the feeder fiber (FF) respectively.

CW Results and Discussion

First, we measured the output optical power and spectrum of the three sources in Figure 4-8.
Figure 4-9(a) plots the powers when the RSOA bias current is swept from 20 to 200 mA for a
OBPF bandwidth of 2 nm. The blue and red curves show the measured powers for the ASE
spectrum sliced and the self-seeded RSOA-FCL sources, respectively. The RSOA-FCL delivers
power, which is up to 9 dB higher than the power of the ASE spectrum sliced source. At 100 mA
bias current, the spectrum sliced power and the RSOA-FCL powers are -13 and -3.8 dBm,
respectively.The ASE spectrum sliced source suffers from the high filtering loss of the OBPF. The
green curves show the measured power for the externally seeded RSOA. At 100 mA bias current
and seeding input powers of -25 and -6 dBm, the measured output powers are -0.9 and 4 dBm,
respectively. The corresponding 100 mA spectra for the three cases are plotted in Figure 4-9(b).
For comparison, the total RSOA ASE is depicted by the black curve. The ASE sliced spectrum (blue
curve) is just a filtered version of the total ASE. However, the RSOA-FCL output shown by the red
curve has a narrower spectrum with a stronger peak, see Figure 4-9(b) inset. The spectral shape
qualitatively agrees with the simulated spectra in Figure 4-6(c). The peaks are due to the RSOA
ripple. The strong mode competition in the RSOA-FCL cavity results in a few modes with the least

80
 Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

losses that dominate the spectrum. The green curve shows the single mode spectral output of
the externally seeded RSOA.

Figure 4-9. Experimental characterization: (a) shows measured optical power versus bias current for the ASE spectrum sliced, the
self-seeded RSOA-FCL and the externally seeded RSOA colorless transmitters. (b) shows the corresponding spectrum at a bias
current of 100 mA. The inset shows a zoomed in plot for the three spectra.

In a next step we have measured the RIN of the three sources. The measurement was done at
60, 100 and 140 mA RSOA bias currents. Figure 4-10(a) shows the RIN for the ASE spectrum sliced
source, and one can see that the RIN reduces with higher current. This is related to the increase
in optical power with bias current, as shown in Figure 4-9(a). The ASE-ASE beat noise results in
higher RIN at lower frequencies [231]. The RIN for the RSOA-FCL is plotted in Figure 4-10(b). The
RIN level is lower by more than 10 dB/Hz compared to that of the spectrum sliced case, especially
at lower frequencies. This is attributed to the reduction of the amplitude fluctuation noise of the
seeding ASE by the saturated RSOA [160]. As the bias current is increased, the power in the RSOA-
FCL cavity increases, thus strongly saturating the RSOA and improving the noise suppression
efficiency. However, this noise suppression efficiency is not constant across the whole spectrum.
Due to the slow RSOA gain recovery during amplitude fluctuation suppression, low frequency
fluctuations are more efficiently suppressed. Another important point to note is that one can
resolve the multiple longitudinal modes of the RSOA-FCL.
Figure 4-10(d) shows the zoomed section of the RIN in Figure 4-10(b), where modes spaced by
free spectral range (FSR) of 6.4 MHz are visible. This relates to a L = 15 m long SMF cavity with
FSR= c 2nL , where c and n are the speed of light in vacuum and the refractive index of SMF.
Finally, the RIN of the externally seeded RSOA case is presented in Figure 4-10(c). At 100 mA, its
RIN is around 25 and 15 dB/Hz lower than the RIN of the ASE spectrum sliced and the RSOA-FCL
source, respectively.
Further, we investigated the impact of the cavity parameters on the RIN of the RSOA-FCL. First,
the RIN is measured by adding 0, 2, and 4 dB more loss in the cavity and is plotted in
Figure 4-11(a). Initially the total cavity loss due to the insertion losses of the different
components is 7 dB. The RIN increases for higher losses. It is discussed in [160] that the RSOA
needs to be in the saturation region to achieve efficient amplitude noise suppression, and low
RIN. Thus high loss means the RSOA moves to the linear region and the noise is no more
suppressed, resulting in high RIN. Next, Figure 4-11(b) shows the measured RIN when using a 0.6,

81
Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

2, and 3 nm bandwidth OBPFs in the cavity. The 0.6 nm gives relatively higher RIN. This can be
related to the higher power loss with narrow band filtering. The 2 and 3 nm have similar RIN.

Figure 4-10. Measured RIN at 60, 100 and 140 mA RSOA bias currents; (a) For the RSOA ASE spectrum sliced, (b) self-seeded RSOA-
FCL, and (c) externally seeded RSOA colorless transmitters. For all cases, the RIN increases for lower bias current. The externally
seeded RSOA shows the lower RIN followed by the RSOA-FCL. (d) shows the zoomed in section of (b) where the RIN shows peaks
corresponding to the longitudinal modes of a 15 m SMF cavity RSOA-FCL

Figure 4-11. RIN characterization of the RSOA-FCL; (a) The RIN increases when introducing additional loss in the cavity. (b) The
impact of the optical bandpass filter (OBPF) bandwidth on the RIN is shown.

Another interesting point about the RSOA-FCL is the large number of longitudinal modes due to
the extra-long cavity. The length of the cavity is determined by the distribution fiber of the PON,
which ranges from tens of meters to kilometers. We can resolve these modes in the electrical

82
 Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

domain, and the results are plotted in Figure 4-12. The plots show exemplary section of an
electrical spectrum for 100 m and 1 km SMF cavity RSOA-FCLs. As previously stated, the FSR of
the modes is c 2nL . The FSR for the three cavities are 1 and 0.1 MHz, respectively. For a cavity
of 1 km length with a FSR of 0.1 MHz and an OBPF filter bandwidth of 2 nm millions of modes
could exist in this source. This highly multimode nature makes it a low coherence source that is
quite susceptible to chromatic dispersion and mode partition noise [193].

Figure 4-12. Resolved longitudinal modes of the self-seeded RSOA-FCL for; (a) 100 m and (b) 1 km SMF cavities. FSR= c 2nL ,
where c n , L are the speed of light in vacuum, the refractive index and the length of the SMF cavity.

4.6 Data Transmission Comparison

In this section, we discuss the data transmission performances of the ASE spectrum sliced, the
RSOA-FCL, and externally seeded RSOA schemes in the experimental setups of Figure 4-8.

Figure 4-13. Measured SNR and small signal frequency response: (a) Measured signal to noise ratio (SNR) vs. frequency for the
RSOA ASE spectrum sliced, self-seeded RSOA-FCL and the externally seeded RSOA-FCL colorless transmitters driven at 100 mA. (b)
Small signal modulation frequency response of an RSOA at different bias currents.

First, we are interested in the frequency response of the respective sources. Instead of measuring
only the small signal modulation frequency response, we measure the frequency dependence of
their SNR. The SNR frequency response is interesting, because it directly relates to the bit error
ratio (BER) and includes effects from both a degradation of the frequency response and also a
frequency dependence of the RIN. The frequency dependent SNR has been measured by
transmitting a QPSK encoded discrete multi-tone (DMT) signal with a PRBS sequence of 211-1,

83
Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

[97]. In our case, 256 subcarrier DMT signal within a total bandwidth of 10 GHz have been
generated by an arbitrary waveform generator (AWG) to drive the RSOA. The RSOA bias current
has been optimized separately for each transmitter. In the case of the externally seeded RSOA
scheme, the input seeding power was -4 dBm. After direct detection, we measured the SNR of
each subcarrier. The measured SNR for the three schemes are plotted in Figure 4-13(a). One can
see that the RSOA-FCL sits between the spectrum sliced and the externally seeded RSOA
schemes. It has ~5 dB higher SNR than the former and ~5 dB lower SNR than the later. For all
cases, the SNR reduction at higher frequency is due to the bandwidth limitation of the RSOA. The
small signal modulation frequency of the RSOA for bias currents from 50 to 120 mA are plotted
in Figure 4-13(b). The RSOA has a 3 dB bandwidth of 2.4 GHz at 100 mA.
Next, we transmitted 5 and 10 Gbit/s amplitude shift keying (ASK) non-return-to-zero (NRZ)
signals with the three schemes. Figure 4-14 shows plots with measured quality factors Q2 [dB]
for a BtB (back-to-back), a 8 km SMF and 25 km NZDSF transmission experiment with 5 and
10 Gbit/s NRZ signals. Since the use of forward error correction (FEC) in MFH is common option,
we used the standard BER thresholds of 10-2 and 10-3 as a minimum requirement [232, 233]. Then
we calculated the Q2 values for each transmitter scheme at various received input powers. We
have been using Q2 values rather than BER as a quality metric, because our dominant noise
source is the RSOA ASE noise that acts as an almost ideal white Gaussian noise source for not too
high input powers [234]. The black dashed lines in Figure 4-14 indicate the Q2 values 7.3 and
9.8 dB which correspond to a BER of 10-2 and 10-3, respectively.

Figure 4-14. Measured quality factor Q2 [dB] for transmission of (a) 5 Gbit/s and (b) 10 Gbit/s using the ASE spectrum sliced
scheme (blue), the self-seeded RSOA-FCL (red) and the externally seeded RSOA (green) transmitters.

In Figure 4-14(a), it is shown how all three transmitter schemes can achieve Q 2 values above
9.8 dB for a BtB transmission. Of the three schemes, the externally seeded RSOA delivers the
highest Q2 with a value of up to 17 dB. The self-seeded RSOA-FCL follows with up a Q2 value of
15 dB. The RSOA ASE spectrum sliced scheme has the lowest BtB Q2 with a value just above the
9.8 dB limit. For the 8 km SMF transmission, shown by the dashed lines, only the RSOA-FCL and
the externally seeded RSOA can achieve Q2 above 9.8 dB. Chromatic dispersion and chirp
interplay reduce the performances in all three schemes. Of course, the penalty is far stronger for
the spectrum sliced source. We also transmitted the 5 Gbit/s signal over 25 km NZDSF with
chromatic dispersion of 5 ps/nm∙km, depicted by dotted lines in Figure 4-14(a). The transmission
over NZDSF was carried out to reduce the impact of chromatic dispersion in order to extend the

84
 Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

reach. The ASE spectrum sliced source can reach Q2 values of 7.3 dB at a -31 dBm received power.
The externally seeded RSOA and the RSOA-FCL reach Q2 values of 17 and 12 dB, respectively. The
impact of the dispersion of the NZDSF on the externally seeded RSOA scheme is negligible.

Figure 4-15. Eye diagrams: (a) Measured eye diagram for 5 Gbit/s transmission and (b) 10 Gbit/s transmission with the three
transmitter schemes. The measurements were performed at a received power of -12 dBm. The transmission distances are
indicated in the figure.

However, due to the relatively broader spectra of the ASE spectrum sliced and RSOA FSL schemes
the overall dispersion accumulates and the signal quality degrades for 25 km.
Eye diagrams of the 5 Gbit/s transmission experiments for the three schemes at various distances
are plotted in Figure 4-15(a). Depending on the transmitter scheme fiber dispersion impacts the
eye diagram opening to a different degree. For the externally seeded RSOA, the 25 km NZDSF
eye diagrams are as open as the BtB case. This is due to the relatively narrow optical spectrum
of this scheme. Thus, when transmitted through the NZDSF of only ~5 ps/nm∙km dispersion, it
does not suffer much inter-symbol interference (ISI) at 5 Gbit/s (symbol period of 200 ps).
However, due to the broader spectra of the ASE spectrum sliced and the RSOA-FCL sources,
25 km NZDSF does not offer similar performance as for the BtB case.
We also performed 10 Gbit/s transmission experiments. Figure 4-14(b) shows that only the
externally seeded RSOA and the RSOA-FCL can achieve 10 Gbit/s with Q2 above 9.8 and 7.3 dB,
respectively. The performance limitation at 10 Gbit/s emanates from the limited modulation
bandwidth of the RSOA, which has been presented in Figure 4-13 and that causes ISI. This is
illustrated by the closing of the 10 Gbit/s eye diagrams plotted in Figure 4-15(b) compared to
that of the 5 Gbit/s eye diagrams in Figure 4-15(a). In addition, At 10 Gbit/s the transmission is
limited to a 2 km SMF due to the interplay between chirp and chromatic dispersion.
To summarize, it can be seen how the RSOA-FCL in any of the transmission experiments performs
midway between the ASE spectrum sliced and the externally seeded sources. It is thus, an
alternative source for many instances, where neither ultimate speed nor ultimate transmission
distance is needed. Our experiments indicate that the RSOA-FCL is valid source for transmission
of up to 10 Gbit/s in SMF fiber over up to 2 km and up to 5 Gbit/s over 8 km SMF and 25 km of
NZDSF. The need for a NZDSF to achieve longer reach may challenge the cost efficiency of the
RSOA. However, such dispersion management can either be obtained by deploying NZDSF or by
deploying O-band RSOAs [235, 236]. In central offices, electronic chromatic dispersion could also
be an option.

85
Comparison of RSOA-FCL with Alternative RSOA Based Colorless Transmitters

4.7 Conclusion
In this paper, we have demonstrated that self-seeded RSOA-FCL is a viable colorless source for
short-reach applications up to 10 Gbit/s. It shares the simplicity and cost-efficiency of an ASE
spectrum sliced colorless source but offers a 5 dB SNR advantage over an ASE spectrum sliced
source. This SNR is only 5 dB below the SNR of a more expensive externally seeded RSOA source.
Similarly, the frequency response is halfway in-between those of an ASE and an externally seeded
sources. Here, we have shown transmission up to 5 and 10 Gbit/s over distances of 25 km NZDSF
and 2 km SMF, respectively. Therefore, the RSOA-FCL is an economical alternative as an
upstream source for e.g. WDM-PON based mobile fronthaul networks.

86
5 Bit- and Power-Loaded DMT for Maximum Bit Rate

This chapter is devoted to the implementation of advanced signal processing to overcome the
limited modulation bandwidth of RSOAs. In chapter 2, the measured small signal modulation
response of the RSOA and SNR of the three RSOA based colorless sources have proved that the
narrow bandwidth is one main cause for the reduced bit rate. When using standard NRZ ASK
modulation format, 10 GBit/s is already challanging to achieve with the RSOA-FCL and the
externally seeded RSOA. For the ASE spectrum-sliced source, 5 GBit/s is barely feasible only in
BtB. Thus, one resorts to advanced modulation formats with higher spectral efficiency to squeeze
in more data within the limited bandwidth.
This chapter explores adaptive multicarrier signal format, specifically discrete multi-tone (DMT).
The limited modulation bandwidth of the RSOA is sliced into multiple orthogonal subcarriers.
These individual subcarriers are then optimally bit- and power-loaded depending on the available
SNR at that frequency, thus maximizing the total bit rate. Thus, this chapter presents a
comparative study of the capacity increase brought by implementing bit- and power-loaded DMT
modulation for low-cost colorless transmitters.
The flow of this chapter is as follows. The general DMT signal generation and reception flowchart
is presented in 5.2. The Experimental setups in section 5.3 are used to characterize the three
schemes in section 5.4 where SNR is used as performance metric. Section 5.5 presents the results
of bit- and power-loaded DMT transmission to boost the transmitted bit rate.

This chapter is submitted to Optics Express Journal in April 2016. Here the submission is
presented word by word.

87
Bit- and Power-Loaded DMT for Maximum Bit Rate

Bit- and Power-Loading – A Comparative Study on

Maximizing the Capacity of Colorless DMT Transmitters
Simon A. Gebrewold, Romain Bonjour, Romain Brenot,
David Hillerkuss, Juerg Leuthold

Submitted to Optics Express Journal, April 2016

5.1 Introduction
Maximizing the bit-rate of colorless, inexpensive and “poorly” performing transmitters has
become a research topic on its own. This is of particular importance in wavelength division
multiplexed (WDM) passive optical networks (PONs) that are considered a possible solution to
cope with the capacity surge in access networks [28, 30].
WDM-PONs are being considered for applications requiring bidirectional secure and high
capacity links. WDM-PONs are therefore ideal solutions to serve fiber to the home and building
(FTTH/B) users with high capacity demands and also ideal to cover the needs of mobile fronthaul
(fiber to the antenna, FTTA) [213, 232]. The latter is characterized by small cell sizes with dense
base stations. If WDM-PON are to be mass deployed, large quantity of wavelength specific optical
network units (ONUs) would be required, which highly complicates the inventory. Thus, the
challenge is to develop ONU transmitters that allow the installation of identical devices across all
subscribers. The solution are colorless transmitters that operate across a broad wavelength
range, while their specific operation wavelength is controlled by an external factor [75, 81, 215,
237]. Installing identical colorless transmitters across all ONUs consequently, reduces the
number of specialized components - thus simplifying the inventory. In addition, relatively low-
quality mass producible colorless sources will make WDM-PON more economical.
For low-cost colorless transmitters, the specific operating wavelength can be controlled by two
main methods. A filter at a remote location can automatically choose the transmission
wavelength of an ONU [210, 238]. Alternatively, colorless operation can also be implemented by
using an external seed laser placed at the central office (CO) to determine the upstream
wavelength [96, 107, 115, 150, 218].
One promising colorless source is based on reflective semiconductor optical amplifiers (RSOAs).
RSOAs are ideal sources as they offer a broad optical bandwidth and low cost.
There are three main architectures to integrate RSOAs in a WDM-PON as upstream (US)
transmitters at the ONU. A theoretical and experimental comparison of the three architectures
has been presented in [212].
 The simplest and most economical scheme is using RSOAs as amplified spontaneous
emitters (ASE) in a spectrum-sliced scheme, as shown in Figure 5-1(a). The US signal is
directly modulated onto the ASE of the RSOA. At the remote node (RN), a WDM filter
slices the modulated ASE, passing only the assigned sub bands from each ONU to the
central office (CO) [74, 216]. The scheme is low-cost and simple, but so far only bit
rates below 1 Gbit/s have been achieved with this scheme. The limitations of this
scheme are high filtering losses and susceptibility to chromatic dispersion due to a
broad optical spectrum.

88
 Bit- and Power-Loaded DMT for Maximum Bit Rate

 Another cost-efficient source – however with improved performance – is the self-

seeded RSOA fiber cavity laser (RSOA-FCLs) as suggested in [194, 195] and depicted in
Figure 5-1(b). In this case, a mirror is placed at the RN behind the WDM filter. This
establishes a feedback mechanism to the RSOA to form a Fabry-Perot laser resonator.
The operating wavelength of each ONU’s RSOA-FCL is determined by the WDM filter
port it is connected to. This passive assignment of the emission wavelength simplifies
the wavelength control. Thus far, RSOA-FCL have shown to transmit up to 10 Gbit/s
[239, 240].
 The highest performance - however, a relatively expensive scheme - is the externally
seeded RSOA, shown in Figure 5-1(c).Here, a bank of lasers, one for each ONU, is
installed at the CO. The WDM filter at the RN redistributes the individual cw
(continuous wave) laser lines to each ONU. At the ONUs the RSOAs amplify the seeding
light and encode the US signal. Externally seeded RSOAs have already shown
transmission beyond 20 Gbit/s with pulse amplitude modulation (PAM) [92] and more
than 25 Gbit/s using discrete multi-tone (DMT) [96, 97]. However, in both cases,
controlled offset filtering by a WDM filter or an additional delay line interferometer
was required, which makes the system sensitive to any wavelength drift. In addition,
the extra lasers at the CO increases cost and energy consumption.
In these three schemes, the modulation bandwidth limitation of RSOAs is a challenge that has to
be addressed to fulfill the high data rate requirements of WDM-PONs. There have been different
approaches to alleviate this issue. One solution is to improve the speed of the RSOAs by
optimizing the physical design of the device [88]. Initial implementations of RSOA, mainly in the
externally seeded scheme, employed non-return-to-zero (NRZ) amplitude-shift keying (ASK)
modulation with either electronic [241, 242] or optical signal processing [242, 243] or a
combination of electronic and photonic signal processing [95] to demonstrate 10 Gbit/s. Return-
to-zero (RZ) and NRZ are compared in [90]. Advanced modulation formats have also been
demonstrated for higher spectral efficiency with electronic and optical equalization in [218].
Polar RZ PAM4 with delay interferometer optical signal processing has achieved 20 Gbit/s [91].
Recently, multicarrier schemes such as adaptive orthogonal frequency division multiplexing
(Adaptive-OFDM) [219] have been implemented for improved performance of the externally
seeded scheme. Adaptive OFDM [97] and bit- and power-loaded DMT [96], both with optical
equalization have already demonstrated beyond 25 GBit/s transmission.
In general, DMT has been used to maximize capacities of bandwidth limited intensity modulated
direct detected (IM-DD) systems [97, 244-246]. DMT is a type of OFDM with real-valued output
signals that can be easily detected with inexpensive direct detection receivers. As a multicarrier
format, DMT also brings the flexibility of adapting the modulation format and power of any
individual subcarrier to cope with the frequency dependence of the channel signal-to-noise ratio
(SNR).
In this paper, we employ bit- and power-loaded DMT to show that the ASE spectrum-sliced
scheme, the self-seeded RSOA-FCL, and the externally seeded RSOA configurations are viable US
transmitters. Actually, the SNR offered by each sources has been measured and compared. It is
found that the SNR improves by more than 5 dB when switching from an ASE spectrum-sliced
scheme to a self-seeded RSOA-FCL, and it improves by another ~5 dB when switching from a self-
seeded RSOA-FCL to an externally seeded RSOA. Finally, transmission experiments have been
performed. We demonstrate 6.25, 20.1, and 30.7 GBit/s in back-to-back (BtB) and 4.17, 10.1, and
24.5 Gbit/s over 25 km NZDSF (non-zero dispersion shifted fiber) for the ASE spectrum-sliced,

89
Bit- and Power-Loaded DMT for Maximum Bit Rate

RSOA-FCL and externally seeded RSOA sources, respectively. No offset filtering was needed in
any of these experiments [96, 97].

Figure 5-1. WDM-PON topologies with RSOA based colorless upstream (US) transmitters: (a) ASE spectrum-sliced transmitter: the
US signal is directly modulated on to the ASE of the RSOA. The WDM filter at the remote node (RN) passes only the assigned sub
band from each RSOA at the optical network units (ONUs). (b) Self-seeded RSOA-FCL: a Fabry-Perot cavity indicated by the red
dashed lines is formed between the RSOA and a mirror at the RN. The cavity is embedded in the distribution fiber (DF) of the PON.
(c) Externally seeded RSOA: a laser at the central office (CO) seeds the RSOA at the ONU. The RSOA encodes the US signal onto
the seed laser line. In all case, the ONUs are connected by the DF to the RN. The RN consists of the WDM filter and it is connected
to the CO through the feeder fiber (FF).

The organization of the paper is as follows. First, we discuss the steps for generating and
analyzing a DMT signal. In section 3, we present the experimental setups. In section 4, we
measure the SNR of the RSOA-FCL by varying WDM filter bandwidth, cavity length and bias
current. We compare these results with the SNR of the ASE spectrum-sliced and externally
seeded RSOA schemes. Finally we present the results of bit- and power-loaded DMT transmission
experiments.

5.2 DMT Signal Generation and Reception

DMT is a multi-carrier transmission technique with real-valued orthogonal subcarriers. The signal
processing flow charts for a DMT transmitter and receiver are given in Figure 5-2. A
pseudorandom bit sequence (PRBS) is generated to test the transmitter. The bits are then
rearranged from serial into parallel according to the number of subcarriers and the information
content in the respective subcarriers. Each parallel bit stream is then mapped onto its individual
constellation (e.g. quadrature amplitude modulation – QAM). DMT gives us the flexibility to
encode a different modulation format onto every subcarrier. Therefore, depending on the

90
 Bit- and Power-Loaded DMT for Maximum Bit Rate

available SNR at each subcarrier the number of bits per symbol can be adapted to achieve the
highest possible throughput, a feature called bit-loading. In our experiments, the modulation
schemes ranged from binary phase shift keying (BPSK with 1 bit/symbol) to 64 quadrature
amplitude modulation (64QAM with 6 bit/symbol). The constellation diagrams of the modulation
formats are shown in Figure 5-3. After bit mapping, the power of each subcarrier is optimized in
a power-loading stage depending on the SNR profile, see Figure 5-2. This is done such that all
subcarriers will end up with a relatively similar bit error ratio (BER) performance even if they
have different modulation formats. The adaptive bit-and power-loading algorithm is based on
Chow’s method discussed in [247]. In a next step, an inverse fast Fourier transform (IFFT) with
Hermitian symmetry is performed to obtain the real valued DMT symbols. A cyclic prefix is added
to each DMT symbol before being serialized and loaded into an arbitrary waveform generator
(AWG). The AWG converts the DMT signal to the analog domain before it is fed to an RF amplifier
and encoded to the optical domain by means of an RSOA.
In the receiver, a direct detection scheme is used to detect the optical signals. The electrical
signals are then fed into a real-time oscilloscope (RTO). The RTO effectively digitizes the signal
with a resolution of ~5.8 bits. In a first step, timing synchronization is performed using the non-
data aided DMT timing estimation technique discussed in [248].

Figure 5-2. DMT signal generation and reception flowchart. In a first step the pseudorandom bit sequence (PRBS) is generated,
which is then rearranged into a parallel bit-stream. Subsequently, adaptive bit mapping and power-loading are implemented
depending the system SNR profile. Inverse fast Fourier transform (IFFT) is then used to encode the QAM symbols on the DMT
subcarriers. Cyclic prefix (CP) is added on the DMT symbols before the signal is converted back to serial. The serial data is loaded
onto an arbitrary waveform generator (AWG) which converts the digital signal to analog using an inbuilt digital to analog
converter (DAC). On the receiver side the signal is detected by a photodiode and sampled by a real-time oscilloscope (RTO). The
finite resolution of the RTO effectively converts the analog signal into a digital signal. Next, timing synchronization is performed
to recover the DMT clock. Then the serial signal is converted into a parallel data stream. After CP removal, FFT is used to decode
the DMT subcarriers. Frequency and phase equalization of the received QAM symbols is performed. Here, we measure the error
vector magnitude (EVM) which is used to measure the signal-to-noise ratio (SNR). Then, the symbols on each subcarrier are
demapped to recover the transmitted PRBS for bit error ratio (BER) measurements.

Subsequently, the signal is parallelized and the cyclic prefix is removed. The subcarriers within
the electrical signal are then derived by the fast Fourier transform (FFT). To properly decode the
bits within a subcarrier blind phase and frequency offset estimation methods are applied on
every subcarrier [249]. The error vector magnitude (EVM) of the QAM symbols on each subcarrier

91
Bit- and Power-Loaded DMT for Maximum Bit Rate

is calculated and converted into SNR [250]. Then, the QAM symbols are de-mapped to extract
the received bit pattern. The BER is measured by comparing the transmitted and received bit
patterns on each subcarrier. The overall BER is also measured by comparing the total transmitted
and received bits.

5.3 Experimental Setups

In our experiments, we investigated the three colorless upstream transmitters in WDM-PON
scenarios, see Figure 5-4. For all experiments, the transmitted DMT signal is generated offline
with MATLAB and converted to an analog waveform by an AWG (M8195A – 60 GSa/s). The analog
signal is amplified and a DC bias current is added before being fed to an RSOA. In all cases, a
1 mm long InGaAsP quantum well C-band RSOA (provided by III-V labs, 10G-RSOA-11-S) is placed
at the ONU. The distribution fiber (DF) links the ONU to the RN. At the RN, a flat top optical band-
pass filter (OBPF) emulates the WDM filter. RN and CO are connected by the feeder fiber (FF).

Figure 5-3. Constellation diagrams of modulation formats used for the bit- and power-loaded DMT signal generation. Binary phase
shift keying (BPSK) encodes 1 bit/symbol. Quadrature phase shift keying (QPSK) encodes 2 bit/symbol. The quadrature amplitude
modulation (QAM) formats: 8QAM, 16QAM, 32QAM, and 64QAM encode 3, 4, 5, and 6 bit/symbol, respectively.

The three implementations of the RSOA based transmission links are shown in Figure 5-4 and
described in more detailed now.
Figure 5-4(a) shows the setup for the ASE spectrum-sliced scheme. In this scheme, the current of
an ROSA in the ONU is modulated with the data. This encodes the information onto the ASE
output power of the RSOA. At the remote note, an OBPF band limits the modulated ASE-signal
to the assigned wavelength channel. The filtered signal is then sent to the CO. The filter
bandwidth has been chosen to be identical to the filter bandwidth of the other schemes.
Figure 5-4(b) shows the setup for the self-seeded RSOA-FCL. Here, a Faraday rotator (FR) is
placed at ONU after the RSOA. In addition, a Faraday rotator mirror (FRM) is placed at the RN
after a coupler behind the WDM filter. The FRM and the RSOA reflective facet form the Fabry-
Perot resonator cavity embedded in the DF. In case of RSOA with high polarization dependent
gain (HPDG), the FR and FRM cancel out the birefringence effect in the cavity and maintain a
constant polarization at the RSOA input [125]. The coupler at the RN has a coupling ratio of 90/10,
where 90 % of the power is kept in the cavity and 10 % is transmitted to the CO. The large amount
of the power kept in the cavity is required to overcome cavity losses and thereby to achieve a
net gain and with this lasing. In addition, the high power in the cavity maintains the RSOA in the
nonlinear operation region for better noise and residual modulation suppression [160, 212].
Figure 5-4(c) shows the experimental setup for the externally seeded RSOA scheme. Here, a laser
source at the CO seeds the RSOA at the ONU through the fiber link. The seeding power entering
the RSOA has been optimized for the highest SNR. The RSOA now only serves as a modulator and
amplifier for the respective laser wavelength.

92
 Bit- and Power-Loaded DMT for Maximum Bit Rate

In all cases, the power budget between the ONU and the CO is ~12 dB. At the CO, a direct
detection receiver detects the signals from the three transmitters. Here, a 90/10 coupler sends
10 % of the power to an optical spectrum analyzer (OSA) to monitor the signal. A variable optical
attenuator (VOA) is used to control the power entering the preamplifier Erbium doped fiber
amplifier (EDFA). A pre-amplified direct detection receiver is used here to guarantee that the
signal quality measured here is as much as possible independent of the receiver scheme. A 5 nm
flat top OBPF suppresses the out of band ASE noise from the EDFA. The optical signal is finally
detected by a photodiode (PIN, responsivity 0.6 A/W), which is directly connected to a RTO
(DSOX96204Q – 20 GHz bandwidth and 40 GSa/s). The oscilloscope records the signal for offline
post-processing with MATLAB as outlined in Figure 5-2.

Figure 5-4. Experimental setups of three RSOA based colorless transmitter schemes in a WDM-PON scenario: (a) ASE spectrum-
sliced, (b) self-seeded RSOA-FCL, and (c) externally seeded RSOA. At the optical network units (ONUs), the upstream signal from
the arbitrary waveform generator (AWG) is amplified and added to a bias current to drive the RSOA. The ONU and the remote
node (RN) are connected by the distribution fiber (DF). At the RN, an optical bandpass filter (OBPF) mimics the WDM filter. In the
case of the RSOA-FCL in (b), the Faraday rotator mirror (FRM) is placed behind the filter and a coupler to form the Fabrey-Perot
cavity, shown by the red dashed line. For the externally-seeded RSOA in (c), the seeding cw laser is place at the central office (CO).
On the receiver side at the CO, pre-amplification and direct detection are performed. A real-time oscilloscope (RTO) is used to
record the electrical signal after the photodiode. An optical spectrum analyzer (OSA) is used to monitor the optical signal at the
CO.

5.4 SNR Characterization

In this section, we present the measured SNRs versus subcarrier frequency offered by the three
transmitter schemes. Before presenting the results, we study the performance of the RSOA itself
by measuring the electrical modulation bandwidth. In a second step, we then focus on the self-
seeded RSOA-FCL. We study the impact of the WDM filter bandwidth in the RN, the fiber cavity

93
Bit- and Power-Loaded DMT for Maximum Bit Rate

length and the bias current on the SNR offered by the RSOA-FCL. Once the optimum operation
parameters and limitations of the RSOA-FCL have been understood, we adjust the operation
points of the ASE spectrum-sliced and the externally seeded RSOA. So for instance we set the
WDM filter bandwidth of the ASE spectrum-sliced source to be identical with the RSOA-FCL. The
input power for the externally seeded transmitters is also set for maximum performance. Here
we only give the operation points that have been selected while one can find more details on
finding the ideal operation points of an ASE spectrum-sliced transmitter in [251, 252] and the
externally seeded transmitter [97, 253].

Electrical Bandwidth of the RSOA

We start our investigation by measuring the small signal modulation frequency response of the
RSOA itself. We chose the externally seeded configuration to exclude the impact of dispersion
and the external cavity. The frequency response for 50 – 120 mA bias current and at an input
power of -3 dBm are plotted in Figure 5-5, which shows that the modulation response improves
with higher bias current. At 100 mA, the RSOA exhibits -3 and -6 dB bandwidths of 2.4 and
4.55 GHz, respectively.

Figure 5-5. Measured small signal modulation frequency response of an RSOA at 50 – 120 mA bias currents. The modulation
bandwidth increases with bias current. At 100 mA, the RSOA has -3 and -6 dB bandwidths of 2.4 and 4.55 GHz.

RSOA-FCL Performance
Here, we perform three studies to analyze the performance of the RSOA-FCL: First, we investigate
the impact of the filter bandwidth in the cavity. Second, we study the SNR dependence on cavity
length and bias current. Last, we explore the transmission performance for different cavity
lengths at optimized the operation conditions. In all cases, we use the SNR over the modulation
frequency as a measure for the link performance and for comparing parameters.
The SNR over the modulation frequency is measured according to the following procedure. We
transmit DMT signals and evaluate the SNR on each subcarrier at the receiver. Here, the
modulation format and power of each subcarrier is identical. We generate a DMT signal having
256 subcarriers with QPSK as modulation format and equal electrical power for all subcarriers. A
PRBS 11 is used as the transmitted bit sequence. The total symbol rate is 10 GBd. The IFFT size is
512 and a 2.5 % cyclic prefix is added to each DMT symbol.
The signal is generated offline with MATLAB. An AWG generates the analog signal with 0.5 peak-
to-peak voltage. The output of the RF amplifier was optimized to ~3.5 peak-to-peak voltages. A
bias tee combines the amplified signal with the specific bias current to drive the RSOA. On the

94
 Bit- and Power-Loaded DMT for Maximum Bit Rate

receiver side, an RTO records the signal detected by a photodetector. Subsequently, the signal is
demodulated in an offline digital signal processing step. The measured EVM of the subcarriers is
used to calculate the SNR of each subcarrier. This way, we can determine the frequency resolved
SNR of the transmission link across the whole modulation bandwidth. This allows us to determine
the usable bandwidth of the transmission link. The useable bandwidth is the frequency band,
within which the SNR is high enough to still allow for a BER below 10-3 with BPSK modulation.
Theoretically, a minimum SNR of 6.7 dB is required.
First, we investigated the impact of the WDM filter bandwidth in the RN. We compared OBPF
bandwidths of 0.6 nm and 2 nm for three cavity lengths of 15 m, 115 m, and 1 km standard single
mode fiber (SSMF). The bias current was maintained at 100 mA.

Figure 5-6. Measured SNR comparison of self-seeded RSOA-FCL transmitter with WDM filter bandwidths of 0.6 and 2 nm in RN.
The dotted and solid lines show the measured SNR for the 0.6 and 2 nm filters, respectively. For both cases, three different cavity
lengths, 15 m, 115 m and 1 km were used. 0.6 nm filter cavity has peak SNR which is ~5 dB lower than that of the 2 nm one.

Figure 5-6 shows the SNR over the modulation frequency for both cases. The SNR is higher for
the filter with wider bandwidth for all cavity lengths. For a 0.6 nm filter, the 15 and 115 m cavities
have a peak SNRs of ~14 dB. The maximum SNR obtained for the 1 km cavity with 0.6 nm is only
~12 dB. The usable bandwidth of the 15, 115 m and 1 km cavities with 0.6 nm filter was ~4.6, 4.5
and 2.7 GHz. When a 2 nm filter has been used in the receiver one finds SNRs that are 5 dB higher.
The usable bandwidth then increases by ~40 % to ~6.7, ~6.4, and ~4.5 GHz. The results are
summarized in Table 5-1.
Table 5-1. Summary of results to compare the peak SNR and usable SNR bandwidth for the RSOA-FCL with 0.6 nm or 2 nm filter
bandwidths. For both filters, the fiber cavity length varied between 15 m, 115 m and 1 km.

RSOA-FCL Cavity OBPF -3 dB Usable SNR

Peak SNR [dB]
length Bandwidth bandwidth [GHz]

0.6 nm 14.3 4.6

15 m
2 nm 19.45 6.7

0.6 nm 14.1 4.5

115 m
2 nm 19.5 6.4

0.6 nm 11.9 2.7

1 km
2 nm 17.7 4.5

95
Bit- and Power-Loaded DMT for Maximum Bit Rate

One explanation for the SNR degradation with narrower filters would be that a narrower filter
leads to higher losses [198]. For example, at 100 mA, the reflected power into the RSOA was ~-
6 dBm and ~-2 dBm for the 0.6 and 2 nm filters, respectively. Thus, higher losses means that the
RSOA is less saturated. Therefore, the RSOA will be unable to efficiently suppress the residual
modulation and noise. As a consequence the RIN will be higher and the SNR is degraded [198].
The second explanation based on [128] is that in case of highly multimode sources, increasing
the number of modes (in this case wider filter) will distribute the noise over the multiple modes
thus lowering the noise spectral density and RIN. Subsequently, we continue the investigation
with the 2 nm filter only.

Figure 5-7. Measured SNR versus subcarrier frequency of self-seeded RSOA-FCLs operated at 80, 100, 110 and 120 mA bias
currents for cavity lengths of (a) 15 m, (b) 115 m and (c) 1 km. The plots indicate that the SNR bandwidth generally increases with
the bias current. The 1 km cavity shown in (c) has a reduced SNR peak and bandwidth when compared to the shorter cavities. The
dashed line indicates the minimum SNR of 6.7 dB required to transmit BPSK, thus, it determines the usable bandwidth. In each
plot, the insets show a sample electrical domain DMT spectra. The black and the red curves show the transmitted and received
DMT electrical spectrum, respectively, at 110 mA. The received DMT spectrum for the 1 km cavity RSOA-FCL shown in plot (c) inset
has a higher noise level compared to that of the 15 and 115 m cavities spectra shown in insets of (a) and (b).

Next, we further investigated the impact of the distribution fiber cavity length and the bias
current on the SNR of the self-seeded RSOA-FCL. We considered cavity lengths with 15 m, 115 m
and 1 km SSMF operated with RSOA bias currents of 80, 100, 110 and 120 mA.
Figure 5-7(a) and (b) show similar performance for cavities with 15 m and 115 m length. The
insets in Figure 5-7(a) and (b) which are the transmitted (black curve) and the received (red
curve) DMT electrical spectrums also show similar characteristics. The similar performance
originates from negligible chromatic dispersion for the two lengths. In both cases, the frequency

96
 Bit- and Power-Loaded DMT for Maximum Bit Rate

dependent SNR increases with the bias current, which is related to the improved RSOA frequency
response for higher bias currents as shown in Figure 5-5. The SNR degradation for higher
modulation frequencies is due to the bandwidth limitations of the RSOA as discussed in
Figure 5-5. At 110 mA, both achieve maximum SNR of ~19 dB. The usable bandwidth where the
SNR is still sufficiently high to receive a BPSK signal is ~6.7 and ~6.4 GHz for cavity lengths of 15 m
and 115 m, respectively.
In Figure 5-7(c) we have plotted the SNR for the longer 1 km SSMF cavity. It is way lower than
those in Figure 5-7(a) and (b). This can be understood by the fact that total accumulated
chromatic dispersion is larger. Since the RSOA-FCL is highly multi-moded, chromatic dispersion
leads to considerable mode partition noise. This noise increases the RIN [193, 208], causing a
SNR degradation. A bit higher loss of the 1 km fiber cavity makes only a minor contribution to
the SNR increase. At 110 mA, the maximum SNR is ~17 dB only and the usable bandwidth is now
~4.5 GHz. In contrast to the RSOA-FCL with shorter cavity lengths, the SNR in the 1 km cavity is
not increasing with bias current. We expect that this originates from the strong chirp during
direct modulation at higher bias currents. The chirp causes spectral broadening which increases
chromatic dispersion penalty. This induces an additional SNR degradation for higher bias currents
[254]. This is supported by the large noise level in the received electrical DMT spectrum of the
1 km RSOA-FCL that can be seen in the inset of Figure 5-7(c) (red curve). However, the negative
impact of the dispersive cavity on the SNR can be avoided by using O-band RSOAs. In the O-band
where chromatic dispersion is minimum, RSOA-FCL cavity could be extended beyond 1 km [235,
239].

Figure 5-8. Measured SNR to compare the impact of transmission fiber on the performance of an RSOA-FCL: The solid and dotted
lines show the measured SNR for BtB (back-to-back) and after 25 km non-zero dispersion shifted fiber (NZDSF) transmission,
respectively. In both cases, the results are for 15 m, 115 m and 1 km fiber cavity lengths. The figure shows that the SNR bandwidth
narrows by up to 50 % after transmission over 25 km fiber.

Finally, we investigated the SNR after transmission over 25 km NZDSF (chromatic dispersion:
~5ps/nm∙km). This fiber is used to emulate a low dispersion PON fiber – similar to what one might
expect if the system were operated in the O-Band. The probed SNR for the three cavity lengths
for both a BtB and a 25 km transmission are plotted in Figure 5-8. When comparing the BtB and
25 km SNRs, the maximum SNR values for the respective cavity lengths did not show major
change. Rather, the usable SNR bandwidth was reduced to ~3.3, 3.2, and 2.8 GHz for the 15 m,
115 m and 1 km cavities, respectively. The results are summarized in Table 5-2. This shows 40-
50 % bandwidth reduction compared to the BtB case. The finite signal bandwidth of the directly
modulated RSOA results in chirp after propagation that induces power fading and bandwidth

97
Bit- and Power-Loaded DMT for Maximum Bit Rate

reduction. In addition, such highly multi-mode source suffers from mode partition noise induced
by the NZDSF dispersion which increases the RIN.
Table 5-2. Summary of results for the SNR performance of the RSOA-FCL with 15 m, 115 m and 1 km cavities for back-to-back
(BtB) and 25 km NZDSF transmission.

RSOA-FCL Cavity Usable SNR

Tx Peak SNR [dB]
length bandwidth [GHz]

15 m BtB 19.45 6.7

25 km NZDSF 19.01 3.2

115 m BtB 19.5 6.4

25 km NZDSF 19.2 3.2

1 km BtB 17.7 4.5

25 km NZDSF 17.1 2.8

Performance Comparison of the Three Transmitter Schemes

Here, we compared the SNR performance of the ASE spectrum-sliced with the self-seeded RSOA-
FCL and the externally seeded RSOA sources. For this experiment, the ASE spectrum-sliced source
was equipped with the 2 nm filter, similar to that of the RSOA-FCL. For the RSOA-FCL, we took
the 15 m and 1 km cavities, which offered the best and the worst performance, respectively.
Both RSOA-FCLs were equipped with a 2 nm filter. The RSOA input power for the externally
seeded scheme was -3 dBm. The bias current for all experiments was set to 100 mA. Figure 5-9(a)
and (b) show the measured SNR values for BtB and transmission over 25 km NZDSF, respectively.
It can be seen that the externally seeded RSOA and the self-seeded RSOA roughly offer a 10 dB
and a 6 to 8 dB higher SNR over the ASE spectrum-sliced source across the whole spectral band.
The self-seeded RSOA-FCL with the shorter cavity outperforms the longer cavity FCL by about 1
– 2 dB across the spectral band. And while the externally seeded RSOA exhibited the highest SNR
bandwidth of ~9.6 GHz for BtB, the 15 m and 1 km long RSOA-FCLs offer useable bandwidths of
6.7 and 4.5 GHz and the ASE spectrum-sliced scheme features a usable bandwidth of only
2.5 GHz.
When transmitting the signals over 25 km NZDSF fiber, the usable SNR bandwidth reduced to
~1.6 GHz for the ASE spectrum-sliced source. The RSOA-FCL declines to a usable bandwidths of
3 and 2.8 GHz for the 15 m and 1 km long cavities. And the externally seeded RSOA still offers a
6.7 GHz bandwidth with sufficient SNR for transmission. The source of the SNR degradation with
transmission is the interplay of the chirp (induced by the direct modulation of the RSOA) and the
dispersion of the transmission fiber. The narrower usable bandwidth recorded after transmission
will influence the maximum achievable bit rate.
We attribute the higher performance observed for the RSOA-FCL when compared to the ASE
spectrum-sliced source to three main factors. First, the RSOA-FCL has a significantly narrower
spectrum, which significantly reduces the impact of chromatic dispersion [193, 212]. Second, the
RSOA-FCL has a larger output power. For example, at 100 mA, the RSOA-FCL has an output power
of -3.4 dBm while the ASE spectrum sliced scheme emits -11.8 dBm. Third, the saturated RSOA

98
 Bit- and Power-Loaded DMT for Maximum Bit Rate

in the cavity nonlinearly suppress the amplitude noise in the lasing cavity, which lowers the RIN.
These three contribute to the improve SNR of the RSOA-FCL compared to the ASE spectrum sliced
source [212].

Figure 5-9. Comparison of the measured SNR for the three colorless transmitter schemes: RSOA ASE spectrum-sliced (SS), self-
seeded RSOA-FCL and externally seeded (ES) RSOA. (a) Shows the BtB (back-to-back) SNR measurement. The externally seeded
RSOA (green line) outperforms the RSOA-FCL (red and brown) and the ASE spectrum-sliced scheme (blue) by ~10 and ~5 dB SNR
across the whole frequency range, respectively. (b) Shows the SNR after 25 km NZDSF transmission. The SNR bandwidth is reduced
due to dispersion of the transmission fiber. In general, the RSOA-FCL offers performance midway between an ASE spectrum sliced
and an externally seeded RSOA.

5.5 Bit- and Power-Loaded DMT for Highest Bit Rates

In this section, bit- and power-loaded DMT transmission is used to maximize the data throughput
for any of the three bandwidth limited sources. We proceed in two steps. We again first study
the impact of the cavity length on the RSOA-FCL and the achievable data throughput in more
detail. Only when we know the optimum operation conditions of the RSOA-FCL, we investigate
and compare the maximum capacity of all three schemes.

Maximizing the Capacity of the RSOA-FCL

In our experiments, we performed bit- and power-loading based on Chows algorithm [247].
Chow’s algorithm can briefly be described as follows. It takes in the SNR characterization of a
transmitter at the subcarrier frequencies of the DMT signal. It also uses the SNR requirement of
all advanced modulation formats (see Figure 5-3) to achieve a predetermined BER. Then, it
disregards subcarriers with the lowest SNR that cannot support transmission at the
predetermined BER for even the simplest format. The power of such unusable subcarriers is then
transferred to usable subcarriers. For subcarrier with non-integer bit per symbol, the bit number
is rounded off to the nearest integer, which requires the power to be re-adapted accordingly.
This power adaptation step (power-loading) helps to achieve similar BER for all subcarriers.
First, we performed bit- and power-loaded DMT transmission for the RSOA-FCL with 15 m, 115 m
and 1 km cavities. The signal was optimized for a BtB experiment and for transmission over 25 km
NZDSF. Figure 5-10(a), (b) and (c) show the SNR, and the bit- as well as the power-loading
patterns of the DMT and the measured per-subcarrier BER for 15 m, 115 m and 1 km long RSOA-
FCLs, respectively. The 15 and 115 m cavities can support up to 5 bit/symbol (32QAM) for low
frequency subcarriers. In case of the 1 km RSOA-FCL, the highest order modulation format is
16QAM, with 4 bits/symbol.

99
Bit- and Power-Loaded DMT for Maximum Bit Rate

Figure 5-10. Measured SNR, bit- and power-loading pattern, and bit error ratio (BER) versus sub-carrier frequencies and the
achieved total line rate for different RSOA-FCL cavity lengths: Plots (a), (b) and (c) show the results for 15, 115 m and 1 km long
cavities, respectively. The red and orange colors indicate BtB (back-to-back) and 25 km NZDSF transmission measurements. In all
cases, the overall BER (solid lines in the BER plots) are kept below the FEC limit of 9.29∙10 -3. (d) Plots the measured total line rate
for 15, 115 m and 1 km long cavities, both for BtB and after 25 km transmission. In a BtB measurement, the RSOA-FCL offers up
to 20 Gbit/s line rates. The line-rate is still 10 Gbit/s when transmitting signals over 25 km.

The power-loading pattern is arranged to guarantee a minimum SNR difference for all subcarriers
with the same modulation format, so that a similar BER performance can be obtained. The solid
lines in the plots indicate the overall BER, i.e. the total BER across all subcarriers. It can be seen
that the overall BER is below 9.29∙10-3 in all instances, which is the limit for an FEC with 12.5 %
overhead [255].
Figure 5-10(d) shows the maximum line rates obtained with RSOA-FCL cavity lengths of 15 m,
115 m and 1 km length for a BtB and a transmission over 25 km NZDSF. Maximum capacities are
obtained for short FCL cavities of 15 and 115 m length. The capacities are between 20 and
10 Gbit/s depending on the transmission distance. For longer cavity lengths the capacity dropped
to 12.4 Gbit/s due to a reduced SNR and a limited bandwidth for the BtB experiment and 7 Gbit/s
could at best be transmitted over 25 km.

100
 Bit- and Power-Loaded DMT for Maximum Bit Rate

Comparison of Maximizing Capacity

Second, we compared the maximum capacity achievable with bit-loaded and power-loaded ASE
spectrum-sliced, two types of self-seeded FCL RSOAs and externally seeded RSOAs.

Figure 5-11. Measured SNR, bit- and power-loading pattern and per-subcarrier BER for the three transmitter schemes: RSOA ASE
spectrum-sliced (SS) (blue), self-seeded RSOA-FCL (red and brown) and externally seeded (ES) RSOA (green). The ASE spectrum
sliced scheme can support modulation formats with up to 3 bits/symbol while RSOA-FCL and externally seeded RSOA can encode
up to 6 bits/symbol for low subcarrier frequencies. The solid line in the bottom plot indicates the overall BER that can be achieved
with either of the three transmission schemes. All BERs are within the FEC limit of 7.5 10-3.

Figure 5-11(a) and (b) show BtB and 25 km DMT transmission results, respectively. For low
subcarrier frequencies, the ASE spectrum-sliced source supported up to 3 bit/symbol (8QAM),
the RSOA-FCL up to 5 bit/symbol (32QAM), and the externally seeded RSOA up to 6 bit/symbol
(64QAM). The power-loading pattern shown in Figure 5-11(a) and (b) led to relatively similar
SNRs for subcarriers with the same modulation format. As a result, the BER for different
subcarriers was relatively constant over the modulation frequency. The solid lines show the
overall BER. For all schemes, the overall BERs were kept below 9.29∙10-3 (12.5 % overhead FEC)
[255].
Table 5-3 summarizes the achieved maximum line rates for the three transmitter schemes and
the corresponding overall BER. In a BtB experiment the ASE spectrum-sliced, the RSOA-FCL and
the externally seeded sources were delivering capacities up to 6.25, 20.1 and 30.7 Gbit/s,
respectively. For a link with 25 km the sources could still deliver maximum capacities of 4.17,
10.1 and 24.5 Gbit/s. It is interesting to note that the chirp-dispersion induced usable bandwidth
reduction results in considerable capacity reduction, especially in the case of the ASE spectrum-
sliced and RSOA-FCL transmitters. This can be explained by the relatively wider linewidth of the

101
Bit- and Power-Loaded DMT for Maximum Bit Rate

two sources, thus incurring higher dispersion penalty. However, O-band RSOAs, [235], can be
used to lower dispersion penalty and avoid the need for NZDSF. O-band has already been used
for upstream transmission in legacy PONs to relax the requirement on inexpensive poor sources
[30, 256].
Table 5-3. Summary of achieved line rate and the respective BER for the three transmitter schemes. The results are for a BtB (back-
to-back) and after 25 km NZDSF transmission.

ASE Spectrum- Externally Seeded

RSOA-FCL (15 m) RSOA-FCL (1 km)
Sliced RSOA
Line Line Line Line
Rate BER Rate BER Rate BER Rate BER
[Gb/ [Gb/s] [Gb/s] [Gb/s]
s]
6.25 5.8∙10-3 20 5.8∙10-3 12.4 5.8∙10-3 30 5.7∙10-3
25 km BtB
NZDSF

4.2 2.3∙10-3 10 4∙10-3 7 5.8∙10-3 25 7.1∙10-3

5.6 Conclusion
We show how transmission capacities of low-cost sources can be optimized for maximum
capacities by means of bit- and power-loaded DMT transmission. Record high transmission data
rates for an ASE spectrum-sliced, self-seeded RSOA-FCL and externally seeded RSOA of 6.25, 20,
and 30.7 Git/s in BtB and 4.2, 12.4, and 24.5 Gbit/s over 25 km NZDSF transmission are reported.
To the best of our knowledge, these are the highest capacities reported for such inexpensive and
colorless sources. Our experiments show that even lowest cost sources such as spectral sliced
ASE sources or RSOA-FCLs can offer quite some capacity if properly operated, for example, for
O-band upstream transmission. More precisely, ASE sources can provide capacities in the range
of 6 to 4 Gbit/s depending on the distance. Fabry-Perot sources such as the RSOA-FCL may offer
capacities in the order of 20 to 10 Gbit/s for distances up to 25 km. Finally, more costly externally
seeded colorless sources may provide capacities in the range of 30 to 25 Gbit/s if operated with
an optimized bit- and power loaded DMT transmission format without offset filtering.

102
6 Summary and Outlook

6.1 Summary
This thesis presents a comprehensive investigation of colorless upstream transmitters, which are
crucial for the widespread and economical deployment of WDM-PONs. Specifically, self-seeded
RSOA-FCL is studied through mathematical modeling, experimental characterization and high-
speed data transmission tests. The results obtained are summarized as follows.
 Modelling of self-seeded RSOA-FCL

Theoretical treatment of a self-seeded RSOA-FCL has been used to reveal the particulars of its
operation principle, advantages, and limitations. It provides thorough insights on the utilization
of such extra-long cavity and highly multimode source as a directly modulated transmitter. One
of the key outcomes is that despite the ultra large number of longitudinal modes that are excited
in such a long fiber cavity, the output settles with relatively fewer surviving modes. As example,
in a WDM-PON with filter bandwidth of 100 GHz, the RSOA-FCL output power spectral width is
below 20 GHz. Further, thanks to the strong gain saturation of the RSOA to suppress the
amplitude noise of the seeding ASE, the output power is relatively constant, when the cavity
dispersion is low. It has been demonstrated that the resulting constant output power can then
be used for transmitting NRZ ASK signal, which was proved by the resulting open eye diagrams
at 2.5 GBit/s.
 RSOA-FCL experimental characterization and comparison with alternative schemes

The self-seeded RSOA-FCL is quantitatively compared with two alternative RSOA based colorless
upstream transmitters, namely an RSOA ASE spectrum-sliced source and an externally seeded
RSOA scheme. It is measured that there is more than 5 dB SNR advantage when switching from
ASE spectrum-sliced source to a self-seeded RSOA-FCL. Further, the relatively expensive
externally seeded RSOA offers another 5 dB in SNR compared to the RSOA-FCL. This proves that
an RSOA-FCL is a balanced compromise between the lowest cost and simple ASE spectrum-sliced
source and the expensive externally seeded RSOA.
The reason for the 5 dB SNR advantage of the self-seeded RSOA-FCL over the ASE spectrum-sliced
source is credited to the lasing process, which alters the characteristics of the source. Lasing is
achieved when the RSOA gain compensates for the cavity loss. However, since the RSOA has high
PDG, a combination of FR and FRM were utilized to cancel out any polarization rotation due to
the birefringence of the fiber cavity. At such appropriate operating condition, the RSOA-FCL
output power is more than 8 dBm higher for the same bias current. Further, for low enough
cavity loss, a high optical power in the cavity saturates the RSOA, which then nonlinearly
suppresses amplitude noise on the filtered ASE. Thus, the RIN of the RSOA-FCL is measured to be
~10 dB/Hz lower than that of the ASE spectrum-sliced source.
The potentials of the RSOA-FCL was demonstrated by transmitting 5 GBit/s and 10 GBit/s NRZ
ASK signal over 8 and 2 km SSMF, respectively. The narrow modulation bandwidth of the RSOA,
which is around 2.5 GHz, has limited the transmission to only 10 GBit/s. The power fading
emanating from the interplay between the chirp induced by direct modulation and fiber
dispersion further degrade the modulation speed. Therefore, NZDSF was used to minimize the
dispersion penalty and the 5 GBit/s transmission was extended to 25 km.

103
Summary and Outlook

 High-speed transmission with bit- and power-loaded DMT for up to 20 GBit/s

A highly spectral efficient and adaptive bit- and power-loaded DMT modulation is implemented
to squeeze in maximum information within the limited modulation bandwidth of the RSOA – for
all three colorless transmitter schemes. Through a bit- and power-loaded DMT modulation,
record high bit rates of 6, 20 and 30 GBit/s have been measured in back-to-back for an ASE
spectrum-sliced, an RSOA-FCL and an externally seeded RSOA, respectively. Further, 4, 10 and
25 GBit/s have been transmitted over 25 km NZDSF. These are a factor of 2 to 3 capacity boost
for such low-cost and low quality transmitters.
The dependence of RSOA-FCL SNR performance was also explored in depth. The SNR is enhanced
by almost 5 dB and the usable SNR bandwidth is broadened by a factor of 1.5 when switching
from 0.6 nm to 2 nm bandwidth WDM filter. However, extending the cavity length from 115 m
to 1 km SSMF induces ~2 dB SNR penalty and 30 % reduction of the usable SNR bandwidth from
6.7 GHz to 4.5 GHz.

6.2 Outlook
The results presented in this work exhibit the promising aspects of self-seeded RSOA-FCLs and
even the simplest ASE spectrum-sliced sources. They offer a practical and economically feasible
alternative approaches for widespread deployment of WDM-PONs. However, there is still space
for more study and improvement for various parameters.
 Maximizing bit rate
Particularly, in a RSOA-FCL, the RSOA is assigned with three simultaneous tasks: amplification,
noise and residual modulation suppression, and direct modulation. Consequently, the RSOA
design has to fulfill three individually challenging tasks. Thus, distributing these tasks to other
devices may allow for better overall performance. For instance, since it is challenging to gain
significant bandwidth improvement in RSOAs, this task could be assigned to a faster external
modulator, for example an EAM, which can be integrated in the RSOA-FCL. This enables to
concentrate on designing an RSOA with higher gain and low saturation power.

When discussing about bandwidth limitation of an RSOA in the context of RSOA-FCL, it is not only
the electrical modulation bandwidth (see Appendix A), rather also the optical modulation
suppression bandwidth. An ideal RSOA would erase any residual modulation or amplitude noise
on the incident optical power. However, RSOAs have a high pass optical-optical (O/O) response
due to their slow gain recovery. Thus, high-speed modulation or noise will survive and degrade
the performance at high speed, for example at 10 GBit/s. One proposed solution is to insert a
second highly saturated SOA or RSOA in the cavity for efficient noise and residual modulation
suppression [211, 257]. The other scheme was to use a modulation averaging reflector (MAR)
instead of a standard mirror. The MAR is a cascade of reflectors, similar to FBG, where the
reflector spacing is designed in a way that it changes the amplitude probability density
distribution of the modulation into a near-normal distribution. This broadens the noise to the
lower frequencies where the RSOAs can easily erase it. This improved the reach and power
budget [258, 259]. One can also use electrical domain adaptive equalization to remove the
recirculating modulation as presented in [260, 261]. Studying the feasibility of these approaches
individually or in combination will add value.

104
 Summary and Outlook

Thanks to the recent progress in economical implementation of DSP, advanced modulation

formats can help in alleviating bandwidth limitation issues. This works has already demonstrated
how adaptive DMT modulation can boost the capacities of low-quality transmitters by up to 3
fold. This calls for further investigation of alternative techniques such as Nyquist frequency
division multiplexing (NFDM) with Higher order QAM. The benefit of pulse shaping techniques
and pre- and post-signal processing could help push beyond the limitations.
 Extending reach
The RSOA-FCL cavity length, i.e. the drop fiber of the PON, plays a major role in the performances
of the systems through its dispersion and attenuation. The dispersion induces strong mode
partition noise and increases the RIN. The attenuation of the fiber and the insertion losses in the
cavity push the RSOA out of the nonlinear region thus lowering the noise suppression efficiency.
High losses also mean low output power. The dispersion issue is solved by operating the RSOA in
the O-band or including a dispersion compensation fiber. To overcome the cavity losses and
extend the length, high gain RSOAs, > 30 dB gain will be needed. Low saturation power, <-10 dBm
should also be investigated to accommodate higher power budget.

105
Appendix A: Improved Euler’s Method
The carrier density rate equation is solved by using the improved Euler’s method for differential
equations. The differential equation has the form of

N  t 
 f t, N t  , (7.1)
t

where N is the carrier density and t is time. If i is the time index and the calculation has a
time step of t  ti1  ti , the approximation of the carrier density at i  1 is given by

N i1  N i  tf  ti , N i 
t (7.2)
N i1  N i 
2
 
f  ti , N i   f  ti 1 , N i 1  .

107
Appendix B: Additional RSOA Simulation
Here, SOA and RSOA device simulation results are summarized. A bidirectional travelling wave
RSOA model was presented in chapter 3, which was the integral part of the RSOA-FCL. However,
the chapter dealt only with the RSOA-FCL modelling results. Therefore, this appendix uses this
model to explain some characteristics of RSOAs and compare them with SOAs. The simulations
are carried out for static and dynamic operations.

B.1. Static Modelling

This section focuses on static modelling, where the bias is a direct current. The spatial distribution
of carrier densities, ASE and signal powers are modelled. The static characterization of the gain
saturation is also discussed for both SOAs and RSOAs. In the simulations, the same parameters
are used for both devices, except of course the reflective facet of the RSOA.

B.1.1. Spatial Distribution of Carrier Density, ASE and Signal Powers

Here, we compare the spatial distribution of the carrier density, ASE and signal powers for SOA
and RSOA. The simulation was carried out for DC bias current from 10 to 160 mA. The simulation
results for input power of -30 dBm and 0 dBm are shown in Figure B-1(a) and (b), respectively. In
all cases, the solid and dotted lines show results for RSOAs and SOAs, respectively. In case of the
ASE and signal power plots, the back dashed lines shows the summed total power of the forward
and backward propagating fields.
Figure B-1(a.1) shows the carrier density distribution of SOAs and RSOAs for input power of -
30 dBm. In SOAs, the carrier density has bell shape with the peak at the center of the device for
higher bias currents. At lower bias currents where the ASE is negligible, the carrier density
distribution is uniform along the device. At such low input power, ASE dominates the output
power. The ASE spatial distribution in an SOA is shown in Figure B-1(a.3), where the total ASE
power at the end facets is strong. Thus, carrier density is depleted at the facet edges, and this
non-uniform distribution of the carrier density is referred to as longitudinal spatial hole burning
(LSHB) [262-264]. Thus the forward and backward propagating ASE and signals experience the
higher gain at the center of the SOA, which is visualized by a steeper gain in power in
Figure B-1(a.3) and (a.5).
On the contrary, the peak of the carrier distribution in an RSOA is closer to the reflective facet.
The forward, backward and total ASE power for the ASE are shown in Figure B-1(a.2). The ASE
power at the output facet is much stronger since it has been amplified as it propagates twice in
the device (ones forward and second backward). Thus, the carriers are depleted at the
input/output facet. Therefore, the forward and backward propagating ASE and signals
experience the higher gain at the position close to where the carrier density has its peak, see
Figure B-1(a.2) and (a.4). These results are in agreement with similar investigations [265].

109
Additional RSOA Simulation

Figure B-1. SOA and RSOA simulation results of spatial distribution of carrier density, amplified spontaneous emission (ASE) and
signal: In all plots, solid and dotted lines indicate results for RSOA and SOA, respectively. The black dashed lines depict the
summation of the forward or backward propagating ASE or signal. (a) and (b) present results for input power of -30 dBm and
0 dBm, respectively, for bias current from 20 – 160 mA. (a.1) and (b.1) are the carrier density spatial distribution for -30 dBm and
0 dBm input powers. In (a.1), at low input power, the output is dominated by ASE, which is symmetric in case of SOA. Therefore,
the SOA carrier density has a symmetric bell shape. On the contrary, in an RSOA, the ASE power is stronger at the input/output
facet, thus the carriers are depleted at this port. (a.2) and (b.2) are RSOA ASE spatial distribution, where the ASE is amplified as it
propagates back and forth in the RSOA. (a.3) and (b.3) are the SOA ASE spatial distribution, which show symmetric characteristics.
(a.4) and (b.4) are the signal power spatial distribution for an RSOA. (a.5) and (b.5) are the signal power distribution in an SOA.

In case of higher input power, 0 dBm results shown in see Figure B-1(b), the carrier distribution
of the SOA also becomes asymmetric. This is because now the input signal power dominants and
it is stronger at the output facet as it is amplified during the propagation, see Figure B-1(b.5).
Therefore, the carrier density has a peak at the input facet, see Figure B-1(b.1). The signal and
ASE experience the highest gain at the input facet due to the larger carrier density. In RSOAs, the

110
 Additional RSOA Simulation

carrier densities are depleted strongly across the device due to the double propagation of the
signal. The peak of the signal power is at the output facet of the RSOA thus depleting the carrier
at this facet.

B.1.2. Gain Saturation

In this work, gain saturation in RSOAs has been exploited mainly for ASE amplitude noise and
residual modulation suppression. Here the gain saturation is simulated and result are presented
by ploting the output power and the Gain in Figure B-2(a) and (b), respectively, for different input
powers. The solid and dotter lines shows the results for an RSAO and SOA, respectively. For both
devices, as the input power increases, the output power also increases linearly, see Figure B-2(a).
However, at higher input powers, the output powers saturates.

Figure B-2. Gain saturation in SOA and RSOA: The solid and dotted lines show simulation results for RSOAs and SOAs, respectively.
(a) shows the output power vs. input power for bias current from 20 – 160 mA. (b) plots the gain vs. input power (c) shows the
change in gain with the output power when input power increase. It is interesting to see that the output power starts to decrease
at relatively higher input powers, >0 dBm only for an RSOA.

The RSOA saturates early at a lower input saturation power compared to that of SAOs. The early
Gain drop due to saturation of the RSOA is visible in Figure B-2(b). Interestingly, at higher input
power, >0 dBm and higher bias current, the output power of the RSOA reduces as shown in
Figure B-2(c), which plots the gain vs. the output power. On the contrary for the SOA, the output
power for strong input power just stays saturated, and does not reduce. The 3 dB saturation
output power for the simulated SOA and RSOA at different bias current are plotted in
Figure B-2(d). It shows that the SOA has better saturation output power compared to the RSOA.

111
Additional RSOA Simulation

The stronger gain saturation seen in RSOAs compared to SOA is mainly due to the bidirectional
propagation of both signal and ASE[266]. The double propagation also leads to the stronger
contribution of ASE power for the saturation, when using RSOA than in case of SOA. This can be
understood by studying the carrier distribution and ASE of SOAs and RSOAs at different input
power. Figure B-3(a) and (b) shows the carrier density distribution and ASE power, respectively,
at 120 mA bias current for -30, -10 and 0 dBm input power. The solid and the dotted lines shows
the results for the RSOA and SOA, respectively. For the RSOA, the carrier density remains flat at
the input/output facet for all input powers. At higher input power, the peak at the reflective fact
reduces significantly due to the depletion of the carriers. In case of the SOAs, the peak moves to
the input facet. The signal, being amplified during propagation, is strong at the output facet, thus
depleting the carrier at the output facet. The ASE spatial distribution in SOAs and RSOAs in
Figure B-3(b) can be directly related to the carrier density. The peak of the carrier density
coincides with the deep of the ASE. In general, one can see that the ASE power is larger for the
RSOAs (solid lines) compared to the SOA (dotted line).

Figure B-3. Carrier density and ASE power spatial distribution in RSOA (solid lines) and SOA (dotted lines) for -30, -10 and 0 dBm
input power. (a) shows the carrier distribution. For SOA, the peak moves to the input facet at higher input power as the carrier is
depleted at the output due to a stronger signal. For RSOA, the peak stays at the reflective facet, but highly reduces with increasing
input power, demonstrating the stronger gain saturation. (b) shows the total (summation of forward and backward propagating)
amplified spontaneous emission (ASE) spatial distribution. The ASE deep always corresponds to the peak of the carrier density. as
signal power increases, the ASE power is reduced. In general, the ASE in the RSOA is larger than the ASE in the SOAs.

The gain dependence on bias current is also simulated for input power of -30, and 0 dBm and
bias current from 20 – 160 mA. In Figure B-4(a), solid and dotted lines represent the gain in an
RSOA and SOA, respectively. For lower bias current, the gain increases exponentially. However,
the gain saturates as the current increases beyond 60 mA. There are few reasons suggested for
the gain saturation at higher bias current. One, the strong ASE generation depletes the carrier,
thus clamping the gain. The other causes are non-radiative recombination, Shockley-Read-Hall
and Auger recombination [263, 267]. Especially, the Auger recombination becomes considerably
large at higher current as temperature of the system raises [263]. The carrier density change with
bias current and input power is plotted in Figure B-4(b), and it clearly shows that the carriers
density is degraded at higher input power, resulting in reduced gain.

112
 Additional RSOA Simulation

Figure B-4. Gain and carrier density vs. bias current for different input power for RSOAs (solid lines) and SOAs (dotted lines). (a)
plots the gain as bias current is increases. Initially, the gain increases exponentially with the bias current. However, above 60 mA,
the gain saturates. (b) shows the carrier density with bias current, which has the same characteristics as the gain. Higher input
power also means less carries and consequently lower gain.

Another parameters affecting the gain saturation in RSOAs is the confinement factor,  . The
impact of  is discussed by simulating the gain vs. input power for low and high confinement
factors.

Figure B-5. Impact of confinement factor  on gain and carrier density: (a) shows the gain change with input power when for
  0.2 and   0.8 , and bias current of 80, 100 and 120 mA. The Higher confinement factor lowers the gain due to
stronger carrier depletion. In (b), the carrier density distribution is plotted for the two confinement factors and low input power (-
30 dBm). It is clearly visible that the carrier density is lower for higher confinement factor. At low input power it is the amplified
spontaneous emission (ASE) that depletes the carriers. In (c) the input power is increased to 0 dBm, and similarly the higher
confinement of the signal power depletes the carriers consequently lowering the gain at higher confinement factor.

113
Additional RSOA Simulation

Figure B-5(a) plots the gain versus input power for RSOAs with   0.2 and   0.8 operated
with 80, 100 and 120 mA. The plot shows that the higher confinement offers lower gain. The
reason is a strong carrier depletion when major part of the ASE and signal power is now confined
within the active medium. To make it clearer, the spatial carrier distributions are plotted in
Figure B-5(b) and (c) for low (-30 dBm) and high (0 dBm) input power respectively. At low power,
the spatial carrier distribution is determined by the ASE profile, refer to Figure B-1. And at higher
confinement factor, the strong ASE in the active region depletes the carriers, thus lowering the
gain [172, 267]. In case of larger input power, as in Figure B-5(c), now the highly confined signal
power depletes the carriers and lowers that gain.

B.2. Dynamic Modelling

This section investigates the dynamic behavior of RSOAs. The dynamic investigation takes into
account two responses, electro-optical (E/O) and optical-optical (O/O). E/O measures the
response for an electrical modulation when the input optical power is constant. The O/O
response measures the response for a modulated optical input power while keeping the bias
current constant.
Both E/O and O/O responses are important for the proper functioning of the RSOA-FCL. The E/O
response measures how fast the RSOA can be modulated to achieve higher bit rates, whereas
the O/O response determines the optical noise and residual modulation suppression efficiency
of the RSOA.
Initially, the simulation concentrates on small signal modulation frequency response of the RSOA
at varying bias current and input optical power. On a next step, the E/O and O/O influences are
studies for a NRZ ASK signal at bit rates from 2.5 to 10 Gbit/s.

B.2.1. Small Signal Modulation Response

In principle, the modulation bandwidth is determined by how fast the gain can be switched [268-
271]. The gain is directly related to the carrier density, thus the carrier life time determines the
gain switching speed and thus the frequency response. The carrier life time  C is in turn
influenced by the carrier injection and the photon rate, both ASE and signal. These parameters
are also influenced by device physical parameters such as device length, confinement factor, and
reflectivity of the facets.
The E/O and O/O responses are simulated by injecting sinusoidal modulated bias current or
optical input power, respectively, as given by

I (t )  I   I cos  2 ft 
(8.1)
P(t )  P   P cos  2 ft 

I and P represent the DC bias current and the CW average optical power, respectively.  I
and  P are the modulation amplitude of the current and optical power. The amplitude of the
modulation is kept to the minimum in order to be able to ignore any higher order effects. The
E/O and O/O results for varying bias current and injected optical power are plotted in
Figure B-6(a) and (b), respectively. In Figure B-6(a), the RSOA has a low pass characteristics and
the response improves with higher bias current, which is related to an increase in the ASE with
current. The higher ASE depletes the carrier through stimulated recombination and consequently

114
 Additional RSOA Simulation

minimizing the carrier lifetime [272, 273]. The increase with bias current is even stronger at
higher input power, in the saturation region of the RSOA [274]. This is because strong input
power (in the saturation region) mean that the stimulated emission is high and further shortens
the carrier lifetime.

Figure B-6. Electro-optical (E/O) and optical-optical (O/O) small-signal modulation response simulation results: (a) shows the E/O
low pass response of an RSOA with increasing bias current and input power. AS expected the bandwidth broadens with increasing
bias current. (b) shows the O/O response for increasing bias current and input power. In general, the RSOA behaves as a high pass
filter by suppressing low frequency modulations. At -5 dBm input power, the RSOA is saturated and nonlinearly suppress low
frequency modulation, however at -20 dBm, the suppression is no more efficient since the RSOA is not yet saturated. At 10 dBm,
the RSOA has a flat response.

The O/O results are also plotted in Figure B-6(b) for increasing bias current and injected optical
power. The RSOA behaves as a high pass filter, where low frequency modulations are suppressed
whereas the RSOA is transparent at higher frequencies. The suppression efficiency of an optical
small-signal modulation depends on the input power, which determines the operation region of
the RSOA.
Figure B-7 explains how an optical modulation could be amplified, suppressed or transmitted
through an RSOA depending on the average optical input power and the modulation frequency.
At lower input power where the RSOA is linear, the modulation is amplified. In the nonlinear
region, the modulation either could be suppressed or transmitted depending the frequency. In
Figure B-7, the low frequency sinusoids (black curve) is suppressed. This is because the lower and
higher peak powers experience higher and lower gain respectively, thus flattening the output.
However, at higher frequencies, red curve, the gain is not fast enough to switch between the
high and the lows, thus the sinusoid is simply transmitted. However, at higher input power,
~10 dBm, RSOA becomes all pass. This is due to the dual propagation [86]. The small modulation
on the forward propagation will be cancelled due to the high saturation. However, the backward
propagation will be modulated due to the residual strong carrier modulation.

One general observation is that the 3 dB bandwidth of RSOAs (in the range of 1 – 3 GHz) is much
narrower than that of similar semiconductor FP-LDs (now days >10 GHz possible). In case of FP-
LDs, after the threshold current is reached, the carrier density is clamped while the photon
density increases. This reduces the carrier lifetime thus broadening the frequency response. The
relaxation oscillation due to the resonator cavity also contributes to the relatively speed of the
FPLDs. On the contrary, due to the increased carrier density in RSOAs, the Therefore, the speed
of the RSOAs is reduced due to the large carrier lifetime [269].

115
Additional RSOA Simulation

Figure B-7. Gain Saturation and its impact on the optical-optical (O/O) modulation response of an RSOA. At lower input power
where the RSOA is in the linear region, any optical signal modulation is linearly amplified. When the input power is large enough
to drive the RSOA in to nonlinear region, the outcome depends on the frequency of the optical modulation. For lower frequencies,
the RSOA suppress the modulation. This is because the high and low peaks of the modulation experience low and high gain
respectively, thus having similar output power. However, this is not the case for high frequency modulation, as the RSOA gain is
not fast enough to switch between high and low levels at high speed.

B.2.2. Direct Modulation and Modulation Cancellation

 Direct Modulation

Direct modulation of the RSOA is achieved by encoding the non-return-to-zero (NRZ) signal on to
the bias current. The bias current directly modulates the carrier density and in turn the gain of
the RSOA. The modulation of the gain is then imprinted on to the intensity of the amplified
output of the RSOA. The modulation speed of the RSOA is highly limited as demonstrated by the
small signal modulation response presented in the previous subsection.

Figure B-8. RSOA direct modulation eye-diagrams at 2.5, 5 and 10 GBit/s. Widely open eye diagrams are observed at 2.5 and
5 GBit/s. However, inter symbol interference starts to kick in at 10 GBIt/s.

The simulation setup and the resulting eye-diagrams are presented in Figure B-8. The bias current
swings between 50 and 150 mA and the RSOA input power is set at -6 dBm. A PRBS 11 is
generated at 2.5, 5 and 10 Gbit/s in a NRZ format and it is encoded onto the bias current. The
eye diagrams are wide open for 2.5 Gbit/s and then deteriorate as the bit rate increase to
10 Gbit/s, see Figure B-8. A higher inter symbol interference (ISI) is observed at 10 GBit/s due to

116
 Additional RSOA Simulation

the E/O bandwidth limitation of the RSOA, refer to Figure B-6(a). The carrier density, and thus
the gain is not fast enough to follow the bias current modulation switching at 10 Gbit/s.
 Modulation Cancellation
Modulation cancellation is a special application of RSOAs needed for the proper functionality of
an RSOA-FCL. Modulation cancellation exploits the O/O response of RSOAs presented in
Figure B-6(b), which is explained with Figure B-7. To realize modulation cancellation, an ASK
modulated optical power is injected into the RSOA, and the output optical signal is observed, as
shown in Figure B-9(a).
Figure B-9 shows simulation results when the RSOA is injected with 2.5 and 10 Gbit/s ASK
modulated optical signal. The cancellation is studied by varying the modulation index (MI) and
average power of the injected optical ASK signal. Figure B-9(b) shows the eye diagrams for the
outputs at different bit rate, MI and average input power. For the 2.5 Gbit/s, the modulation is
highly suppressed when the input power is at -5 dBm. The gain can really switch back to steady
state within a time much shorter than the bit period. On the contrary, for the 10 Gbit/s signal,
the suppression is not efficient as the gain is not as fast to return to steady state within the bit
period.
In general an important parameter that quantifies the modulation cancellation efficiency of an
RSOA is the modulation cancellation dynamic range (MCDR), see Figure B-9(c). Within this range,
dPout dPin  0 , which means any power modulation will be suppressed [158].

117
Additional RSOA Simulation

Figure B-9. Modulation cancellation characterization of RSOA: (a) shows the setup for modulation characterization. The RSOA is
injected with an amplitude shift keying (ASK) modulated optical signal. The modulation index (MI) and the average input power
are controlled. (b) presents the modulation cancellation results for 2.5 and 10 GBit/s modulations in the blue and red boxes,
respectively. The x-axis is for the average power increasing from -20 to 10 dBm and x-axis shows MI increasing from 3 to 6 dB. (c)
shows the output power vs. input power plot where the average input power positions are located. For the input power in the
modulation cancellation dynamic range (MCDR), the suppression is efficiently high.

118
Appendix C: DMT Signal Processing
Here, a deeper discussion on the generation, reception and processing of a bit- and power-
loaded DMT signal is presented. First, Chow’s bit- and power-loading algorithm is presented
based on [247]. Second, the calculation steps to generate a real valued output using an IFFT with
a Hermitian symmetry is described. Third, the DSP steps of the receiver are discussed. These
include DMT symbol timing synchronization and frequency error estimation.

A DMT transceiver signal-processing flowchart is repeated in Figure C-1 from chapter 5. On the
transmitter side, a serial PRBS is parallelized based on the number of subcarriers and their
modulation formats. The bit per symbol value and electrical power of each subcarrier is
individually adapted depending on the SNR at that frequency by using Chow’s algorithm
discussed in D.1. and [247]. Afterwards, IFFT with Hermitian symmetry results in real valued DMT
symbols. Cyclic prefix is added on to each DMT symbol before they are parallelized.

Figure C-1. DMT Transceiver flowchart: First, a pseudorandom bit sequence (PRBS) is generated. The PRBS is fed to the DMT
transmitters, which starts by parallelizing the bit sequence. The bit are mapped to QAM symbols, and the order of QAM symbols
and the power on each subcarrier depends on the signal-to-noise ratio (SNR) on that carrier. Inverse fast Fourier transform (IFFT)
is used to modulate the QAM symbols on the subcarriers. Cyclic prefix (CP) is added to the DMT symbols before serialization and
digital to analog conversion (DAC). On the receiver side, the analog signal from a photodiode is digitized by analog to digit al
convertor (ADC). The DMT symbols are parallelized, aligned in time and the CP is removed. FFT is carrier out to derive the QAM
symbols from the subcarriers. Before extracting the bit sequence from the QAM symbols, any phase and frequency offset is
equalized. Error vector magnitude (EVM) is also measured. The bit error ratio (BER) is calculated by comparing the transmitted
and received bit sequence.

On the receiver side, an ADC digitizes the analog electrical signal from the photodiode. Next, a
non-data aided timing estimation is done to align the DMT symbols in time by a technique
presented in D.3.1. and [248]. After cyclic prefix removal, FFT is used to separate subcarrier and
to extract the symbols. After frequency and phase correction (discussed in D.3.2.), the symbols
can be demapped to a bit sequence. An EVM or BER can be used for performance quality metrics.

119
DMT Signal Processing

C.1. Bit- and Power-Loading: Chow’s Algorithm

In general, bit- and power-loading has two purposes [246, 275].
 Rate adaptive: The aim is to maximize the bit rate that can be transmitted through a
channel. This is done by enabling transmission even through the worst channel even
at a low bit rate at a specified BER.
 Margin adaptive: The idea is to minimize the BER for a specified bit rate through a
given channel.
Rate adaptive algorithm is the most common one. It is mainly interesting for communication links
that are approximately time invariant, for example wired communication links. In wired links,
due to the channel time invariance, bit- and power-loading can be done once at the system
installation and no frequent modification or adaptation is needed. Consequently, rate adaptive
bit- and power-loading with DMT is widely used in DSL systems.
Chow’s algorithm is a relatively simple bit- and power-loading procedure, which well
approximates the highest bit rate. It entails three steps: transmitter SNR characterization, bit-
loading and power-loading. The transmitter SNR characterization is general to any adaptive
transmission system.

C.1.1. Transmitter SNR Characterization

A standard DMT signal, i.e. with out any bit- or power-loading, is used to measure the available
equivalent SNR in the limited modulation bandwidth of a transmitter. This is accomplished by
transmitting DMT signal with identically powered and identically QPSK modulated subcarriers,
see Figure C-2(a). After transmission, the degraded DMT signal is received as depicted by
Figure C-2(b). The signal degradation is due to the frequency dependent response of the
transmitter and the transmission system. After an FFT and equalization (refer to DMT receiver
flowchart in Figure C-1), the EVM of each subcarrier frequency is extracted and converted into
an equivalent SNR as shown in Figure C-2(c). Since the transmitted signal is known, one can
measure the data aided average EVM, EVM avg , which is converted to equivalent SNR using,

1
SNR  2
(9.1)
EVM avg

Figure C-2. Channel signal to noise ratio (SNR) characterization: (a) A discrete multi-tone (DMT) signal with identically modulated
and equally power subcarriers is transmitted. Here a QPSK modulation is used. (b) After transmission, a degraded DMT signal i s
received. The DMT electrical spectrum is mainly a blue print of the frequency response of the transmitter. (c) after DMT receiver
signal processing, the SNR on each subcarrier frequency is calculated.

120
 DMT Signal Processing

C.1.2. Bit-Loading
Once the available transmitter SNR is known, the next step is to adapt the modulation format,
i.e. the bit per symbol of each subcarrier. Chow’s algorithm performs the following procedures.
It finds the subcarriers, which do not allow any transmission at the predetermined BER. The
power of the unusable subcarriers is transferred to usable ones. For subcarrier with non-integer
bit per symbol, the value is approximated to the nearest integer. This approximation then leads
to an SNR margin that can be further iteratively optimized. The ideal result is to achieve similar
BER on for all the modulated subcarriers.
Chow’s bit-loading algorithm [247] flow chart is depicted in Figure C-3. Few important
parameters need to be specified first:
 SNR of the transmitter or channel, SNR , which was measured in Figure C-2
 The expected BER on each subcarrier
 The SNR gap  gap : The  gap is the difference between the SNR required to achieve
the maximum capacity (based on Shannon) and the SNR required to achieve this
capacity at a predefined BER [275-277]. For Gaussian noise characterized channel and
QAM signal transmission,  gap can be calculated for a specific symbol error rate (SER)
or BER for cases with Gray or other codding schemes. The results have been published
in a lot of papers, and in this work these values are borrowed from [278].
 The number of subcarriers N sc
 The iteration number I iter,max : Chow’s algorithms iterates to find the optimum
combination of bit per symbol and power for the highest bit rate.
Additional variables in the flow chart are:
 The SNR margin  margin quantifies the extra noise that can be tolerated while
maintaining the same expected BER at that subcarrier [279].
 The vector matrix Btarget is a matrix consisting of the maximum bit per symbol that can
be transmitted on the individual subcarriers in the ideal case without any SNR margin.
 The indexes i and iter represent the loops though the subcarriers and the overall
optimization iteration, respectively.
 The usable subcarrier number N sc,used : Initially all subcarrier are assumed usable, thus
N sc,used  N sc
 The steps of the flowchart in Figure C-3 are summarized as follows.
(1) Parameter initialization is the first step.
(2) SNR i The bit per symbol capacity of each subcarrier is calculated using Shannon’s
theory and taking into account the SNR gap and the margin. These values are rounded
off to the next integer and the fractions are recorded as bdiff (i ) .
(3) Subcarriers, which do not allow transmission, are discarded as unusable.
(4) Step (2) to (3) are repeated for all the subcarriers.
(5) The total capacity of the channel is calculated by summing the entire bit per symbols of
all the usable subcarriers.
(6) If this summation in (5) equals zero, the channel is deemed bad, or unusable. Else, in
(7) If there are usable subcarriers, then the SNR margin is updated by comparing the
obtained channel capacity Btot and Btarget within the usable subcarriers.
(8) The steps from (2) to (7) are repeated if the maximum iteration number is not reached
and Btot  Btarget . Here the number of used subcarriers is re initialized to N sc,used  N sc

121
DMT Signal Processing

. This is because the new calculated SNR margin can be added onto the previously
unusable subcarriers to increase their SNR.
(9) Else, if the achieved capacity is smaller than the target capacity, Btot  Btarget , one can
increment the capacity of the subcarrier with largest bdiff (i ) .
(10) On the contrary, if Btot  Btarget , the capacity of the subcarrier with the smallest .. is
decremented by one. (9) and (10) are repeated until Btot  Btarget .
(11) Now that the bit per symbol capacity of each subcarrier is optimized, the next step is
power loading

122
 DMT Signal Processing

Figure C-3. Chow’s bit- and power-loading algorithm flowchart: the parameters, the variables and the overall procedure is
explained in section B.1.2.

Figure C-4 presents the impact of transmitting a DMT signal, which is only bit-loaded, however
without power loading. The different colors of the subcarriers in Figure C-4(a) indicate that
different modulation formats are encoded on the subcarriers. The received bit-loaded DMT
signal in Figure C-4(b) and the resulting per subcarrier SNR plotted in Figure C-4(c) show that
subcarriers with identical modulation formats have different amplitude and SNR. This will
degrade the final overall performance. Practically, when measuring the overall BER, it is always
the worst performing subcarrier that dominantly determine the final overall performance.
However, power loading helps to obtain similar SNR on identically modulated subcarriers.

123
DMT Signal Processing

Figure C-4. Transmission of only the bit-loaded DMT signal: (a) shows the electrical spectrum of a bit-loaded DMT. The different
colors correspond to different modulation format. (b) On the receiver side, the degraded DMT spectrum is plotted. This shows
that identically modulated subcarriers have different amplitude, which results in different performance. (c) shows the extracted
signal-to-noise ratio (SNR). The variation of the SNR between identically modulated subcarriers will result in degraded
performance, which can be avoided by power loading.

C.1.3. Power-Loading
The last step in Chow’s algorithm is to adapt the power of the individual subcarriers. Let the
measured SNR at subcarrier with index i be , also shown in Figure C-2(c). After the following bit-
loading step, each subcarrier has a capacity of bi . However, bi is an optimum approximate
capacity and the exact SNR required for this capacity may differ from the measured SNR at the
subcarrier SNR i . Therefore, one can calculate back this exact SNR denoted by SNR i ,b , which is
required for capacity bi bits per symbol to achieve the required BER. Then the power of
subcarrier i is adapted. The power of the individual subcarrier Pi can be calculated as

Pi  10
SNR i ,b SNR n  /10
(9.2)

where the SNR values are given in dB. Here, if SNR i ,b  SNR i , the measured and the required
SNR are equal thus Pi  1 . If SNR i ,b >SNR i , then Pi  1 since the required SNR is higher than
the available measured SNR and the power of this subcarrier is increased to compensate for SNR
difference. If SNR i ,b <SNR i , then the measured SNR is more than what is required for achieve
capacity bi , thus the power of this subcarrier is reduced, Pi  1 . The extra power is transferred
to other subcarriers. Here, the total power must be kept constant to a given value of Ptot by
fulfilling
N sc

P  P
i 1
i tot . (9.3)

The objective of the power loading is to enable identically modulated subcarriers have the same
BER or SNR performance. The output of a bit- and power-loaded DMT is presented in Figure C-5.
The transmitted adapted DMT signal is depicted in Figure C-5(a), whereas Figure C-5(b) and (c)
show the received electrical spectrum of the received DMT signal and the extracted, respectively.
Ideally, identically modulated subcarriers will have equal power and SNR which produces a step
like results in the electrical spectrum and SNR as shown in Figure C-5(b) and (c).

124
 DMT Signal Processing

Figure C-5. Transmission of a bit- and power-loaded DMT: (a) shows the bit- and power-loaded DMT where the different colors
represent different modulation format. (b) The received DMT spectrum displays equal power for identically modulated subcarriers
resulting in a step like ideal case signal-to-noise ratio (SNR) plotted in (c).

C.2. Real-Valued DMT Signal Generation

One advantage of DMT is that it has a real-valued output signal, which can easily be directly
modulated and directly received with a low-cost transceivers. It has similar transmitter signal
processing steps as an OFDM, however is uses Hermitian symmetry with double the IFFT size (i.e.
2N sc ). This section details how Hermitian symmetry is applied to produce the real-valued signal.
Let the transmitted symbols be given as X   X 0 , X 1 , X 2 ,..., X N 1 , which are modulated on to
 
the N sc subcarriers, see Figure C-6. Let x  x0 , x1 , x2 ,..., x2 Nsc 1 represent the output of an IFFT
of size 2N sc which are given as

1 2 Nsc 1
 k 
x(k )   X n exp  j2 n  (9.4)
2 Nsc n 0  2 Nsc 

Where k 0,1, 2,..., 2 N sc  1 . In the IFFT input, the inputs from index 0 to N sc  1 are the symbol
inputs X whereas the remaining inputs are the complex conjugates of the original symbols as
shown in Figure C-6. Thus, one can split Eq. (9.4) into two summations as

1 Nsc 1
  k  
x( k )    X n exp  j2 n  
2 Nsc n 0   2 N sc  

(9.5)
1  *
Nsc 1
 k  

n 0 
 X n exp  j2  2 Nsc  n   ,
2 N sc   2 N sc  


where the first and second summations represent the IFFTs over the original and complex
conjugate symbols, respectively. The second part can be further simplified as:

  k   
*
1 
Nsc 1
 k   1 Nsc 1

x(k )  
n 0 
 X n exp  j2 n  
2 N sc  
  n
n 0  
X exp  j2 n   .
2 N sc   
(9.6)
2 N sc   2 N sc
  

Finally, the summation of the two results in a real valued signal calculated as

125
DMT Signal Processing

1 
Nsc 1
 k    k   
*

x(k )    n
X exp  j2 n 
  n
2 N sc  
X exp  j2 n  
2 N sc   
n 0 
2 N sc
    (9.7)
1 Nsc 1   k  
  2 * Re  X n exp  j2 n 2 N 
2 N sc n0   sc  


Figure C-6. Real-valued IFFT output of DMT signal processing: a block of symbols X  X 0 , X 1 ,..., X N 1 is fed into an
sc

inverse fast Fourier transform (IFFT) with size twice the number of subcarriers, 2N sc . The second half of the IFFT is filled with
 
the complex conjugate of the original symbol block. The resulting DMT output x  x0 , x1 , x2 ,..., x2 N 1 are real valued.
sc

C.3. DMT Receiver DSP

There are two DSP procedures required at the receiver of a DMT system to recover the
transmitted bit sequence accurately. The DMT symbols timing synchronization and subcarrier
frequency offset correction are discussed here.

C.3.1. DMT Timing Synchronization

The DMT symbol timing synchronization is based on a simple non-data aided timing estimation
and correction method discussed in [248]. Let the received DMT signal be given by r (k ) , where
k represents the time index. In addition, let N CP and m represent the cyclic prefix sample
number and the DMT symbol index as shown in Figure C-7(a). One DMT symbol has 2N sc  N CP
samples including the cyclic prefix. Then, the idea is to maximize the likelihood function:
N CP 1
L( )  
m
 r   n  m(2 N
n 0
sc  N CP )    Ts   r *   n  m(2 N sc  N CP )  2 N sc    Ts  (9.8)

where N sc , 2N sc and Ts are the subcarrier number, the IFFT size and the sampling time,
respectively. Figure C-7(b) shows the plot for L() which has peaks at the start of each DMT
symbol.

126
 DMT Signal Processing

Figure C-7. Non-data aided DMT timing synchronization: (a) shows the schematics of the algorithm where the received signal is
taken in blocks of the cyclic prefix size N CP and is convolved with a shifted version of itself. The resulting likelihood function
L() is plotted in (b). one can clearly see the peaks at the start of each DMT symbol.

C.3.2. Frequency Synchronization

After the FFT step of the DMT receiver, residual subcarrier frequency offset needs to be
corrected. A blind subcarrier-wise frequency offset estimation and correction is carried out using
M-th power algorithm [249, 280]. For QAM modulation formats with order M  4 , taking the
4th power erases the modulation signal and only a single point exists on the constellation for the
ideal case that there is no frequency offset. Any deviation from this can be used to detect
frequency offset.
Figure C-8(a) and (b) show exemplary constellation of QPSK before and after taking the 4th power,
respectively. Figure C-8(a) shows a constellation of QPSK with strong frequency offset (black
circle), whereas the green points depict where the constellation points should have been.
Figure C-8(b) shows the constellation for 4th power of an ideal QPSK signal. The time average of
this signal is non-zero. On the contrary, the time average of the 4th power of the signal with
frequency offset, black plot in Figure C-8(a), is almost zero. Therefore, the idea of the 4th power
algorithm is to sweep through frequencies f which will maximize the time average. Here,
 Rs 2  f  Rs 2 with Rs being the symbol rate. Thus, for y(n) being the received symbols,
the frequency offset error fˆ is then given by

1  1 M 1
  j2 f  
2

fˆ  arg  max
 f M
 y 4
( n ) exp   . (9.9)
4
 n 0  Rs  

M is the number of symbols considered. In Eq. (9.9), the summation part is a discrete Fourier
transform (DFT). This means the frequency offset can simply be the frequency at which the DFT
of y (n)4 has a peak divided by 4.

Figure C-8. QPSK Constellation: (a) shows two plots: black plot shows the constellation of a QPSK signal with a frequency offset.
The green dots are the ideal position of a correct QPSK signal. (b) shows the 4 th Power constellation of an correct QPSK signal.

127
Appendix D: Polarization Retracing in FCLs
This appendix discusses how the state of polarization (SOP) of a light entering the RSOA is always
maintained constant despite the birefringence of the fiber cavity in an RSOA-FCL. This is critical
because specially designed faster RSOAs have strong HPDG up to 20 dB. Unless HPDG is
circumvented, the RSOA-FCL cannot lase due to the high cavity loss. An interesting technique by
utilizing Faraday rotator (FR) and Faraday rotator mirror (FRM) has been presented in [125],
which is summarized here.
Figure D-1(a) show the schematics of an RSOA-FCL in a WDM-PON scenario. In [125], it is
proposed to place an FR at the input/output facet of the RSOA and an FRM is positioned at the
RN. The combination of the FR and FRM cancels out any SOP rotation that a light experiences as
it propagates in the birefringent DF or other components. This enables to maintain constant SOP
at the input of the RSOA. The blue shaded region indicates the section between the FR and FRM
that will contribute to the SOP rotation.

Figure D-1. Faraday mirror (FR) and Faraday rotator mirror (FRM) based method to enable lasing in RSOA fiber cavity laser (RSOA-
FCL) by canceling polarization rotation due to fiber cavity birefringence when using RSOAs with high polarization dependent gain
(HPDG). (a) shows an RSOA-FCL in a WDM-PON. An FR is placed at the optical network unit (ONU) in front of the RSOA. An FRM
at the remote node forms the lasing Fabrey-Perot cavity. However, in case of RSOA with HPDG, the state of polarization (SOP) at
must be maintained constant for the reflected light. Based on [125] the FR-FRM combination cancels out any SOP rotation in
the blue shaded area. To analyze this concept, the whole system can be replaced by an equivalent circuit as shown in (b). The
birefringent part of the FCL can be represented by a retarder wave plate (RWP). The FRM can be split into two FRs and one mirror.
The equivalent circuit can be further simplified by converting it into a single propagation direction.

To understand how the FR-FRM combination makes the cavity immune to birefringence, it is
useful to discretize the whole system as shown in Figure D-1(b). Such equivalent circuit allows
for the consideration of Jone’s SOP transfer function of each component. The equivalent circuit
can even be simplified into a single path propagation to simplify the calculation, see
Figure D-1(c). Therefore, the task is to demonstrate that the SOP of the light emitted by the right
side RSOA is maintained as the light reach the RSOA on the left side.

129
Polarization Retracing in FCLs

The FRM is split to a normal mirror and FR for simplicity, which is highlighted by the green shade.
As emphasized by the blue shaded area, the transfer functions of the fiber, the WDM filter and
coupler can be combined into one retarder wave plate (RWP) with transfer function w(, )
given by [281],

w(,  )  R( ) L() RT ( )

 cos   sin  
R( )    (10.1)
 sin  cos  


L( )  

 exp  j
2  0 

.


0 exp j 
 
2



R( ) quantizes a rotation of the SOP by and angle  , whereas L() is the retardation by  .
Here, one particularly needs a 45° FR. The transfer function of a FRM can be calculated by
splitting it into an FR and a mirror, and consider that the light propagates in the FR twice, i.e.
before and after reflection. The transfer function is then M FRM given as,

M FRM  R(45) L(180) R(45)

0 1 (10.2)
 j 
1 0

Based on Figure D-1(c), the over RSOA-FCL transfer function is written from left to right as

M FCL  M FR w(,  ) M FRM w(, ) M FR

 R(45) w(,  ) M FRM w(, ) R(45)
 R(45) R( ) L() RT ( ) M FRM R( ) L() RT ( ) R(45) (10.3)
0 1
 R(45) R( ) L() RT ( ) j   R( ) L() R ( ) R(45)
T

1 0

However, one can use

RT ( )  R( ) (10.4)

and get

0 1
M FCL  R(45) R( ) L() R( ) j   R( ) L() R( ) R(45) (10.5)
1 0

This can be further simplified by using,

130
 Polarization Retracing in FCLs

0 1
R( ) M FRM R ( )  R( ) j   R( )
1 0
0 1
 j 
1 0
(10.6)
0 1
L() M FRM L()  L() j   L()
1 0
0 1
 j 
1 0

Substituting Eq. (10.6) results step by step in Eq. (10.5) follows as

0 1
M FCL  R(45) R ( ) L() R( ) j   R( ) L() R( ) R(45)
1 0
0 1
 R(45) R ( ) L() j   L() R ( ) R(45)
1 0
0 1
 R(45) R ( ) j   R( ) R(45)
1 0
0 1 (10.7)
 R(45) j   R(45)
1 0
 2 2   2  2 

 2 2  j 0 1   2 2
   
 2 2  1 0 2 2 
 2 2  2 2 
1 0
 j 
0 1

The result of Eq. (10.7) shows that the overall transfer function of the fiber cavity is an identity
matrix. This proves that the light that is injected back to the RSOA has the same SOP as the one
emitted.

131
 Glossary

Greek Symbols
 dB/km Optical fiber attenuation in dB/km
H Linewidth enhancement factor, alpha factor, Henry factor

 VOA Variable optical attenuation

 Confinement factor
 gap SNR gap used in Chow’s bit- and power-loading algorithm

 TM,TE Confinement factor for the TM and TE polarizations

 x, y Confinement factor in the x and y axis
 int Internal scattering loss of an active medium
 margin SNR margin in Chow’s bit- and power-loading algorithm
  ,
j Forward and backward propagating ASE field for the j th mode
 NLGC Nonlinear gain compression factor
 out,j Output of j th ASE mode
 r,NL Nonlinear relative permittivity
r Relative permittivity
cpr Remote node coupler ratio
RSOA RSOA-fiber coupling loss
0 Center wavelength
k , j Wavelength of a k th and j th mode for an optical signal and ASE,
respectively
 Sample index in non-data aided timing estimation of DMT signal

t , z Time and space step sizes

 Linewidth of spreading function due to intra-band relaxation
FSR Free spectral range

 Photon density
 CV Optical joint density of state
C Effective carrier lifetime
 in,relax Intra-band relaxation time

133
Glossary

 Susceptibility

Latin Symbols
a0 Differential gain
Axs Cross-section area of active device

A21 Einstein-coefficient for spontaneous emission

A Non-radiative recombination rate

bi Bit per symbol on subcarrier i

B Spontaneous recombination coefficient

B12 Einstein-coefficient for stimulated absorption
B21 Einstein-coefficient for stimulated emission
Btarger Target bit per symbol per subcarrier
Btot Total bit per symbol per subcarrier after bit- and power-loading
c Speed of light in vacuum

C Auger recombination coefficient

d RSOA active layer thickness
d LC Light confinement layer thickness of SCH

d CC Carrier confinement layer (intrinsic layer) thickness of SCH

D Chromatic dispersion
q Electron charge

E Complex electric field

E1,2 Energy of valence and conduction band for a two level systems
Eg Bandgap energy
Eg ,1,2,3 Bandgap energy of layers in heterostructures and SCH
EVM avg Average error vector magnitude
1
E CB Lowest energy level of a semiconductor conduction band
N
E CB Nth energy level of a semiconductor conduction band
EFN,FV Fermi level of n-layer and p-layer
1
E VB Highest energy level of a semiconductor conduction band

134
 Glossary

N
EVB Nth energy level of a semiconductor valence band

Ein,out, k Input and output optical field of the k th mode

Ek , Forward and backward propagating optical signal field for the k th mode
f Electrical RF frequency

f C,V Fermi functions for the conduction and valence bands

F  Fourier transform .
F ( x, y) Lateral distribution function of electric field

gm Material gain
h Plank constant
H fbr Optical fiber transfer function

H OBPF Optical bandpass filter transfer function

I Bias current
I iter,max Chow’s algorithms iteration number

kB Boltzmann Constant
k0 Wave number in vacuum

l RSOA active layer length

l   Broadening function for

L Fiber cavity length
L  Maximize the likelihood function for timing synchronization of DMT signal

L() Retardation of SOP by 

m Number of spatial slices of an RSOA active layer
M Number of longitudinal modes bundled together in RSOA-FCL modelling

M FRM,FCL SOP transfer function of FRM or an FCL

n Order of a Gaussian shaped OBPF
n Complex refractive index
nfbr Fiber core refractive index
ng Group refractive index of RSOA active layer
nim imaginary part of a refractive index

nNL Nonlinear complex refractive index

nrl Real part of a refractive index

135
Glossary

nsp Inversion factor

N Carrier density
N0 Carrier density at transparency
N1,2 Carrier densities of valence and conduction band for a two level systems

N CP Cyclic prefix size

Nm Number of longitudinal modes considered
N sc Number of subcarriers

N sc,used Number of used subcarriers

N syms Number of DMT symbols

PDG Polarization dependent gain

Pi Power on subcarrier i
Psc,tot Total power on all subcarriers
Ptot ( z, t ) Total power at ( z, t ) in RSOA active region
r12,stim Stimulated absorption rate
r21,spon Spontaneous emission rate
r21,stim Stimulated emission rate
rase ( z, t ) Carrier depletion rate due to ASE
rs ( z, t ) Carrier depletion rate due to signal
R   Rotation of an SOP by and angle  ,

Rh,l RSOA reflective and input/output facet reflectivity

RRN Remote node mirror reflectivity
Rspon Carrier depletion due to spontaneous emission
Rsp Spontaneous emission rate
Rstim Carrier depletion due to stimulated emission
Rs Symbol rate
dRVBCB Transition rate from the conduction band to valence band
SNR i Signal-to-noise ratio on subcarrier i
SNR i ,b Signal-to-noise ratio required for subcarrier i to achieve capacity bi

ts Sampling time
T Temperature

136
 Glossary

 Optical field frequency

3dB 3 dB bandwidth of OBPF

e Emitted photon frequency
vg Group velocity
 Optical angular frequency
x1 , x2 Random numbers with Gaussian distribution

V Applied voltage
w Active layer width

137
 Acronyms
10G-PON ................................................................................. 10 Gigabit Passive Optical Network
AOM ........................................................................................................... Active Optical Network
APD ............................................................................................................. Avalanche Photodiode
ASE .............................................................................................. Amplified Spontaneous Emission
ASIC .....................................................................................Application-Specific Integrated Circuit
ASK ............................................................................................................. Amplitude Shift Keying
ASS ............................................................................................................... Amplified Self-Seeded
ATM ................................................................................................. Asynchronous Transfer Mode
AWG ....................................................................................................... Array Waveguide Grating
BBU ......................................................................................................................... Baseband Unit
BER ........................................................................................................................... Bit Error Ratio
BLS ............................................................................................................Broadband Light Source
BPON ......................................................................................Broadband Passive Optical Network
BPSK ....................................................................................................... Binary Phase Shift Keying
BtB .............................................................................................................................Back-to-Back
CAPEX.............................................................................................................. Capital Expenditure
CDMA ......................................................................................... Code Division Multiplexed Access
CO ............................................................................................................................. Central Office
CoMP ....................................................................................................... Coordinated Multi-Point
CPE ................................................................................................. Customer Premises Equipment
CPRI ............................................................................................... Common Public Radio Interface
CW ........................................................................................ Continuous Wave, Continuous Wave
DBA ................................................................................................Dynamic Bandwidth Allocation
DBR ..................................................................................................... Distributed Bragg Reflector
DF ....................................................................................................................... Distribution Fiber
DFB................................................................................................................ Distributed Feedback
DML.........................................................................................................Directly Modulated Laser
DMT ................................................................................................................ Discrete Multi-Tone
DOS ...................................................................................................................... Density of States
DrF ................................................................................................................................. Drop FIber
DRoF.......................................................................................................... Digital Radio over Fiber
DS .............................................................................................................................. Downstream
DSP ........................................................................................................... Digital Signal Processing
DWDM ........................................................................... Dense Wavelength Division Multiplexing
EAM ................................................................................................ Electro Absorption Modulator
ECL ................................................................................................................ External Cavity Laser
EDC......................................................................................... Electronic Dispersion Compensation
EFM ......................................................................................................... Ethernet in the First Mile
EML ..................................................................................................... Externally Modulated Laser
EPON ......................................................................................... Ethernet Passive Optical Network
FBG.................................................................................................................. Fiber Bragg Grating
FCL ..................................................................................................................... FIber Cavity Laser
FEC .......................................................................................................... Forward Error Correction
FF ............................................................................................................................... Feeder FIber
FMC ....................................................................................................... Fixed Mobile Convergence
FPGA ............................................................................................... Feild Programable Gate Array

139
Acronyms

FP-LD ..................................................................................................... Fabrey-Perot Laser Diodes

FR .......................................................................................................................... Fraday Rotator
FRM ............................................................................................................. Fraday Rotator Mirror
FSAN.................................................................................................... Full Service Access Network
FTTA ............................................................................................................. Fiber to the Antennae
FTTB ............................................................................................................... Fiber to the Building
FTTC ..................................................................................................................... Fiber to the Curb
FTTCab ............................................................................................................ Fiber to the Cabinet
FTTD ..................................................................................................................... Fiber to the Desk
FTTH ................................................................................................................... Fiber to the Home
FTTN ........................................................................ Fiber to the Neighborhood, Fiber to the Node
FTTO ................................................................................................................... Fiber to the Office
FTTx ........................................................................................................................... Fiber to the x
GEPON .......................................................................... Gigabit Ethernet Passive Optical Network
GPON .......................................................................................... Gigabit Passive Optical Network
GRIN .......................................................................................................................... Graded Index
HFC..................................................................................................................... Hybrid Fiber Coax
HPDG.........................................................................................High Polarization Dependent Gain
IEEE ...................................................................... Institute of Electrical and Electronics Engineers
IFFT ................................................................................................ Inverse Fast Fourier Transform
IP ........................................................................................................................Internet Protocol
IPDR .................................................................................................. Input Power Dynamic Range
ISI .......................................................................................................... Inter Symbol Interference
LAN.................................................................................................................. Local Area Network
LED ................................................................................................................. Light Emitting Diode
MAC .......................................................................................................... Multiple Access Control
MAN ...............................................................................................................Metro Area Network
MAR .............................................................................................Modulation Averaging Reflector
MCDR ............................................................................. Modulation Cancellation Dynamic Range
MDU .................................................................................................................Multi Dwelling Unit
MFH .................................................................................................................... Mobile Fronthaul
MQW..............................................................................................................Multi-Quantum Well
MSO ........................................................................................................... Multi Service Providers
MTU ..................................................................................................................... Multitenant Unit
MZM ...................................................................................................... Mach Zehnder Modulator
NF ............................................................................................................................... Noise Figure
NG-PON......................................................................... Next Generation Passive Optical Network
NIU ............................................................................................................. Network Interface Unit
NRZ.................................................................................................................. Non-Return-to-Zero
NZDSF........................................................................................ Non-Zero Dispersion Shifted Fiber
ODN .................................................................................................. Optical Distribution Network
OFDMA...........................................................Orthogonal Frequency Division Multiplexing Access
OLT ................................................................................................................ Optical Line Terminal
ONT ........................................................................................................Optical Network Terminal
ONU .............................................................................................................. Optical Network Unit
OPEX ........................................................................................................Operational Expenditure
PAM .................................................................................................. Pulse Amplitude Modulation
PDG ................................................................................................... Polarization Dependent Gain
PON .......................................................................................................... Passive Optical Network

140
 Acronyms

PRBS .................................................................................................. Pseudorandom Bit Sequence

PtMP ................................................................................................................ Point-to-Multipoint
PtP............................................................................................................................ Point-to-Point
QAM ........................................................................................ Quadrature Amplitude Modulation
QPSK .............................................................................................. Quadrature Phase Shift Keying
QW .......................................................................................................................... Quantum Well
REAM .............................................................................. Reflective Electro Absorption Modulator
RIN ............................................................................................................ Relative Intensity Noise
RM................................................................................................................. Reflective Modulator
RN ............................................................................................................................. Remote Node
RRH ................................................................................................................. Remote Radio Head
RSOA ........................................................................... Reflective Semiconductor Optical Amplifier
SBU..................................................................................................................Single-Business Unit
SCH.................................................................................... Separate Confinement Heterostructure
SCM ........................................................................................................... Subcarrier Multiplexing
SDM ................................................................................................... Spatial Division Multiplexed
SFU ..................................................................................................................... Single Family Unit
SLED ................................................................................... Superluminecent Light Emitting Diode
SNR................................................................................................................ Signal-to-Noise Ratio
SOA ............................................................................................. Semiconductor Optical Amplifier
SS ................................................................................................................................ Self-Seeded
SSMF ...................................................................................................Standard Single Mode Fiber
TDM ......................................................................................................Time Division Multiplexing
TE ..................................................................................................................... Transverse Electric
TL ............................................................................................................................ Tunable Laser
TM ................................................................................................................. Transverse Magnetic
US ................................................................................................................................... Upstream
VCSEL .................................................................................. Vertical Cavity Surface Emitting Laser
WDM ......................................................................................... Wavelength Division Multiplexing
WGR ....................................................................................................... waveguide grating router

141
142
 References

References
[1] C. V. N. I. Cisco, "Global Mobile Data Traffic Forecast Update, 2015–2020," White paper,
2016.
[2] L. G. Kazovsky, N. Cheng, W.-T. Shaw, D. Gutierrez, and S.-W. Wong, Broadband Optical
Access Networks: John Wiley & Sons, 2011.
[3] A. N. Pilipetskii, "High Capacity Submarine Transmission Systems," in Optical Fiber
Communication Conferencen (OFC), Los Angeles, California, 2015, p. W3G.5.
[4] The COMSOC Guide To Passive Optical Networks: IEEE Press, Wiley, 2012.
[5] A. Gumaste and T. Antony, "First Mile Access Networks and Enabling Technologies,"
Indianapolis, IN, vol. 46240.
[6] A. Stavdas, Core and metro networks: John Wiley & Sons, 2010.
[7] C. V. N. I. Cisco, "The Zettabyte Era: Trends and Analysis," White paper, 2015.
[8] C. V. N. I. Cisco, "Global mobile data traffic forecast update, 2014–2019," white paper,
2014.
[9] S. Koenig, D. Lopez-Diaz, J. Antes, F. Boes, R. Henneberger, A. Leuther, et al., "Wireless
sub-THz communication system with high data rate enabled by RF photonics and active
MMIC technology," in IEEE Photonics Conference, 2014, pp. 414-415.
[10] F. T. T. H. C. Europe, "FTTH Handbook " vol. 6, 2014.
[11] C. Lin, Broadband optical access networks and fiber-to-the-home: systems technologies
and deployment strategies: John Wiley & Sons, 2006.
[12] ITU-T. 40-Gigabit-capable passive optical networks 2 (NG-PON2): Physical media
dependent (PMD) layer specification [Online].
[13] ITU-T. G.989.1: 40-Gigabit-capable passive optical networks (NG-PON2): General
requirements [Online].
[14] ITU-T. G.987: 10-Gigabit-capable passive optical network (XG-PON) systems: Definitions,
abbreviations and acronyms [Online].
[15] ITU-T. G.987.1: 10-Gigabit-capable passive optical networks (XG-PON): General
requirements [Online].
[16] Europe 2020 Intiative, "National Broadband Infrastructure Plan for Next Generation
Access," in Digital Agenda for Europe, A. E. Initiative, Ed., ed. Sofia, Bulgaria, 2014.
[17] S. Chamberland, "Global Access Network Evolution," IEEE/ACM Transactions on
Networking, vol. 18, pp. 136-149, 2010.
[18] I. Radovanovik, W. v. Etten, and H. Freriks, "Ethernet-Based Passive Optical Local-Area
Networks for Fiber-to-the-Desk Application," Journal of Lightwave Technology, vol. 21, p.
2534, 2003/11/01 2003.
[19] S. Azodolmolky and I. Tomkos, "A Techno-economic study for active Ethernet FTTH
deployments," Journal of Telecommunications Management, vol. 1, pp. 294-310, 2008.
[20] S. Kulkarni, M. El-Sayed, P. Gagen, and B. Polonsky, "FTTH network economics: Key
parameters impacting technology decisions," in International Telecommunications
Network Strategy and Planning Symposium (Networks), 2008, pp. 1-9.
[21] F. Effenberger, D. Clearly, O. Haran, G. Kramer, L. Ruo Ding, M. Oron, et al., "An
introduction to PON technologies [Topics in Optical Communications]," IEEE
Communications Magazine, vol. 45, pp. S17-S25, 2007.
[22] W. Stephen and T. Wang, "The Comsoc Guide to Passive Optical Networks Enhancing the
Last Mile Access," ed: John Wiley & Sons, Chichester, 2012.
[23] J. R. Vacca, Optical Networking Best Practices Handbook: John Wiley & Sons, 2006.
[24] D. Hood and E. Trojer, Gigabit-capable passive optical networks: John Wiley & Sons, 2012.

143
References

[25] J. i. Kani, F. Bourgart, A. Cui, A. Rafel, M. Campbell, R. Davey, et al., "Next-generation PON-
part I: Technology roadmap and general requirements," IEEE Communications Magazine,
vol. 47, pp. 43-49, 2009.
[26] P. Chanclou, F. Bourgart, B. Landousies, S. Gosselin, B. Charbonnier, N. Genay, et al.,
"Technical options for NGPON2 beyond 10G PON," in European Conference on Optical
Communication (ECOC), Amsterdam, The Netherlands, 2011, pp. 1-3.
[27] P. Chanclou, A. Cui, F. Geilhardt, H. Nakamura, and D. Nesset, "Network operator
requirements for the next generation of optical access networks," IEEE Network, vol. 26,
pp. 8-14, 2012.
[28] P. Vetter, "Next Generation Optical Access Technologies," in European Conference on
Optical Communication (ECOC), Amsterdam, The Netherlands, 2012, p. Tu.3.G.1.
[29] J.-i. Kani, "Advanced Optical Access Beyond NG-PON2," in Advanced Photonics for
Communications, San Diego, CA, USA, 2014, p. PT5B.4.
[30] F. J. Effenberger, H. Mukai, P. Soojin, and T. Pfeiffer, "Next-generation PON-part II:
Candidate systems for next-generation PON," IEEE Communications Magazine, vol. 47, pp.
50-57, 2009.
[31] ITU-T. G.987.2: 10-Gigabit-capable passive optical networks (XG-PON): Physical media
dependent (PMD) layer specification [Online].
[32] D. Nesset, "NG-PON2 Technology and Standards," Lightwave Technology, Journal of, vol.
PP, pp. 1-1, 2015.
[33] F. Saliou, P. Chanclou, F. Laurent, N. Genay, J. A. Lazaro, F. Bonada, et al., "Reach Extension
Strategies for Passive Optical Networks [Invited]," IEEE/OSA Journal of Optical
Communications and Networking, vol. 1, pp. C51-C60, 2009.
[34] R. Brenot, "Potential Benefits and Limitations of SOA in Access Networks," in Advanced
Photonics & Renewable Energy, Karlsruhe, Germany, 2010, p. AThC1.
[35] K. Grobe, M. H. Eiselt, S. Pachnicke, and J.-P. Elbers, "Access Networks Based on Tunable
Lasers," Journal of Lightwave Technology, vol. 32, pp. 2815-2823, 2014/08/15 2014.
[36] D. Nesset, "NG-PON2 Technology and Standards," Journal of Lightwave Technology, vol.
33, pp. 1136-1143, 2015.
[37] S. Gosselin, A. Pizzinat, X. Grall, D. Breuer, E. Bogenfeld, J. Torrijos Gijón, et al., "Fixed and
Mobile Convergence: Which Role for Optical Networks?," in Optical Fiber Communication
Conference (OFC), Los Angeles, CA, USA, 2015, p. Th3H.2.
[38] H. Lin, D. Liu, Y. Luo, F. Effenberger, and N. Cheng, "TWDM-PON: System, Standards, and
Key Technologies," in Asia Communications and Photonics Conference (ACP), Shanghai,
2014, p. ATh2F.1.
[39] L. Yuanqiu, Z. Xiaoping, F. Effenberger, Y. Xuejin, P. Guikai, Q. Yinbo, et al., "Time- and
Wavelength-Division Multiplexed Passive Optical Network (TWDM-PON) for Next-
Generation PON Stage 2 (NG-PON2)," Journal of Lightwave Technology, vol. 31, pp. 587-
593, 2013.
[40] N. Cheng, J. Gao, C. Xu, B. Gao, D. Liu, L. Wang, et al., "Flexible TWDM PON system with
pluggable optical transceiver modules," Optics Express, vol. 22, pp. 2078-2091,
2014/01/27 2014.
[41] S. Pachnicke, M. H. Eiselt, K. Grobe, and J. P. Elbers, "The frontiers of optical access
networks," in International Conference on Optical Network Design and Modeling (ONDM),
Pisa, Italy, 2015, pp. 12-15.
[42] Z. Li, L. Yi, and W. Hu, "Symmetric 40-Gb/s TWDM-PON with 51-dB loss budget by using a
single SOA as preamplifier, booster and format converter in ONU," Optics Express, vol. 22,
pp. 24398-24404, 2014/10/06 2014.

144
 References

[43] Y. Lilin, L. Zhengxuan, B. Meihua, W. Wei, and H. Weisheng, "Symmetric 40-Gb/s TWDM-
PON With 39-dB Power Budget," IEEE Photonics Technology Letters, vol. 25, pp. 644-647,
2013.
[44] R. Bonk, H. Schmuck, W. Poehlmann, and T. Pfeiffer, "Beneficial OLT Transmitter and
Receiver Concepts for NG-PON2 Using Semiconductor Optical Amplifiers," in Optical Fiber
Communication Conference (OFC), San Francisco, CA, USA, 2014, p. W1D.5.
[45] G.-K. Chang, A. Chowdhury, Z. Jia, H.-C. Chien, M.-F. Huang, J. Yu, et al., "Key Technologies
of WDM-PON for Future Converged Optical Broadband Access Networks [Invited],"
IEEE/OSA Journal of Optical Communications and Networking, vol. 1, pp. C35-C50,
2009/09/01 2009.
[46] P. Chanclou, A. Pizzinat, F. Le Clech, T. L. Reedeker, Y. Lagadec, F. Saliou, et al., "Optical
fiber solution for mobile fronthaul to achieve cloud radio access network," in Future
Network and Mobile Summit (FutureNetworkSummit), Lisbon, Portugal, 2013, pp. 1-11.
[47] A. Pizzinat, P. Chanclou, F. Saliou, and T. Diallo, "Things You Should Know About
Fronthaul," Journal of Lightwave Technology, vol. PP, pp. 1-1, 2015.
[48] P. D. Townsend, A. Clarke, P. Ossieur, D. W. Smith, A. Borghesani, D. G. Moodie, et al.,
"Towards colourless coolerless components for low power optical networks," in European
Conference on Optical Communication (ECOC), Geneva, Switzerland, 2011, pp. 1-3.
[49] IEEE 802.3 Industry Connections Feasibility Assessment for the Next Generation of EPON
[Online].
[50] C. Knittle, "IEEE 100G-EPON," in Optical Fiber Communication Conference (OFC), Anaheim,
CA, USA, 2016, p. Th1I.6.
[51] D. Nesset, "PON Roadmap," in Optical Fiber Communication Conference (OFC), Anaheim,
CA, USA, 2016, p. W4C.1.
[52] "Alcatel-Lucent Fires NG-PON2 Starting Gun," in http://www.lightreading.com/, ed, 2014.
[53] "Verizon plans NG-PON2 first office deployments early in 2016," in
http://www.lightwaveonline.com, ed: http://www.lightwaveonline.com, 2015.
[54] H. Salib, "FTTH : an MSO perspective," in Optical Fiber Communication Conference (OFC),
Anaheim, CA, USA, 2016, p. Th1I.1.
[55] D. van Veen and V. Houtsma, "High speed TDM PON beyond 10G," in Optical Fiber
Communication Conference (OFC), Anaheim, CA, USA, 2016, p. Tu3C.3.
[56] A. L. Teixeira, A. Shahpari, R. Ferreira, F. Guiomar, and J. D. Reis, "Coherent Access," in
Optical Fiber Communication Conference (OFC), Anaheim, CA, USA, 2016, p. M3C.5.
[57] I. Cano, J. C. Velásquez Micolta, and J. Prat, "7.5 Gb/s Direct DFB Phase Modulation with
8-DPSK for 6.25 GHz Spaced Coherent UDWDM-PONs," in Optical Fiber Communication
Conference (OFC), Anaheim, CA, USA, 2016, p. M3C.4.
[58] H. Hu, R. Asif, F. Ye, S. Gross, M. Withford, T. Morioka, et al., "Bidirectional 120 Gbps SDM-
WDM-PON with Colourless ONU using 10 Gbps Optical Components without DSP," in
Optical Fiber Communication Conference (OFC), Anaheim, California, 2016, p. M3C.1.
[59] Y. Chen, J. Li, P. Zhu, Z. Wu, P. Zhou, Y. Tian, et al., "A Novel WDM-MDM PON Scheme
Utilizing Self-homodyne Detection for High-speed/capacity Access Networks," in Optical
Fiber Communication Conference (OFC), Anaheim, CA, USA, 2016, p. M3C.3.
[60] J. D. Reis, A. Shahpari, R. Ferreira, D. M. Neves, M. Lima, and A. L. Teixeira, "Performance
Optimization of Nyquist Signaling for Spectrally Efficient Optical Access Networks
[Invited]," IEEE/OSA Journal of Optical Communications and Networking, vol. 7, pp. A200-
A208, 2015/02/01 2015.
[61] L. A. Coldren, G. A. Fish, Y. Akulova, J. S. Barton, L. Johansson, and C. W. Coldren, "Tunable
semiconductor lasers: a tutorial," Journal of Lightwave Technology, vol. 22, pp. 193-202,
2004.

145
References

[62] S. Pachnicke, S. Mayne, B. Quemeneur, D. Sayles, H. Schwuchow, J. Zhu, et al., "Field

Demonstration of a Tunable WDM-PON System with Novel SFP+ Modules and Centralized
Wavelength Control," in Optical Fiber Communication Conference (OFC), Los Angeles, CA,
USA, 2015, p. M2A.6.
[63] J. Zhu, S. Pachnicke, M. Lawin, S. Mayne, A. Wonfor, R. V. Penty, et al., "First
Demonstration of a WDM-PON System Using Full C-band Tunable SFP+ Transceiver
Modules [Invited]," IEEE/OSA Journal of Optical Communications and Networking, vol. 7,
pp. A28-A36, 2015/01/01 2015.
[64] S. Sekiguchi, K. Takabayashi, A. Hayakawa, S. Tomabechi, A. Uetake, M. Ekawa, et al., "10
Gb/s Wavelength-Tunable EML with Continuous Wavelength Tuning Covering 50 GHz × 8
Channels on ITU Grid," in Optical Fiber Communications Conference (OFC), Anaheim, CA,
USA, 2007, p. OMS4.
[65] D. McIntosh-Dorsey, R. Bikky, H. Zhang, K. A. Anselm, J. Lii, H.-C. Wang, et al.,
"Continuously Tunable Laser Based on Multiple-Section DFB Laser Technology for 1.25
Gbps WDM-PON Applications," in Conference on Lasers and Electro-Optics (CLEO), San
Jose, CA, USA, 2015, p. SF1I.3.
[66] C. J. Chang-Hasnain, "Tunable VCSEL," IEEE Journal of Selected Topics in Quantum
Electronics, vol. 6, pp. 978-987, 2000.
[67] F. Koyama, "Energy Efficient Tunable Light Sources for Next Generation Flexible Access
Networks," in Optical Fiber Communication Conference (OFC), Los Angeles, CA, USA, 2015,
p. W1J.4.
[68] H. Dalir, Y. Takahashi, and F. Koyama, "High-speed compact electro-absorption modulator
laterally integrated with VCSEL," in European Conference on Optical Communication
(ECOC), Cannes, France, 2014, pp. 1-3.
[69] M. Reeve, A. Hunwicks, W. Zhao, S. Methley, L. Bickers, and S. Hornung, "LED spectral
slicing for single-mode local loop applications," Electronics letters, vol. 24, pp. 389-390,
1988.
[70] M. Zirngibl, C. H. Joyner, L. W. Stulz, C. Dragone, H. M. Presby, and I. P. Kaminow, "LARnet,
a local access router network," IEEE Photonics Technology Letters, vol. 7, pp. 215-217,
1995.
[71] D. K. Jung, S. K. Shin, C. H. Lee, and Y. C. Chung, "Wavelength-division-multiplexed passive
optical network based on spectrum-slicing techniques," IEEE Photonics Technology Letters,
vol. 10, pp. 1334-1336, 1998.
[72] O. K. Jung, S. K. Shin, H. G. Woo, and Y. C. Chung, "Wavelength-tracking technique for
spectrum-sliced WDM passive optical network," IEEE Photonics Technology Letters, vol.
12, pp. 338-340, 2000.
[73] T. E. Chapuran, "Broadband high-density WDM transmission using superluminescent
diodes," Electronics Letters, vol. 26, pp. 696-697, 1990.
[74] P. Sung-Bum, J. Dae Kwang, S. Dong Jae, S. Hong Seok, I. K. Yun, L. Jeong Seok, et al.,
"Colorless Operation of WDM-PON Employing Uncooled Spectrum-Sliced RSOA," IEEE
Photonics Technology Letters, vol. 19, pp. 248-250, 2007.
[75] S. L. Woodward, P. P. Iannone, K. C. Reichmann, and N. J. Frigo, "A spectrally sliced PON
employing Fabry-Perot lasers," IEEE Photonics Technology Letters, vol. 10, pp. 1337-1339,
1998.
[76] J. H. Lee, C. H. Kim, Y. G. Han, and S. B. Lee, "Broadband, high power, erbium fibre ASE-
based CW supercontinuum source for spectrum-sliced WDM PON applications,"
Electronics Letters, vol. 42, pp. 549-550, 2006.

146
 References

[77] A. D. McCoy, P. Horak, B. C. Thomsen, M. Ibsen, and D. J. Richardson, "Noise Suppression

of Incoherent Light Using a Gain-Saturated SOA: Implications for Spectrum-Sliced WDM
Systems," Journal of Lightwave Technology, vol. 23, p. 2399, 2005/08/01 2005.
[78] K. Sun-Jong, H. Jung-Hee, L. Jae-Seung, and P. Chang-Soo, "Intensity noise suppression in
spectrum-sliced incoherent light communication systems using a gain-saturated
semiconductor optical amplifier," IEEE Photonics Technology Letters, vol. 11, pp. 1042-
1044, 1999.
[79] S.-R. Moon, S.-H. Yoo, and C.-H. Lee, "Effect of Noise Distribution in a WDM System Seeded
by a Spectrum-Sliced ASE," Journal of Lightwave Technology, vol. 32, pp. 2271-2276,
2014/06/15 2014.
[80] F. Payoux, P. Chanclou, and R. Brenot, "WDM PON with a single SLED seeding colorless
RSOA-based OLT and ONUs," in European Conference on Optical Communication (ECOC),
Cannes, France 2006, pp. 1-2.
[81] F. Payoux, P. Chanclou, and N. Genay, "WDM-PON with colorless ONUs," in Optical Fiber
Communication Conference (OFC), Anaheim, CA, USA, 2007, pp. 1-3.
[82] D. J. Shin, H. S. Shin, S. B. Park, D. K. Jung, and Y. Oh, "Design and performance of low-cost
WDM-PON access networks," in European Conference on Optical Communication (ECOC),
Berlin, Germany, 2007, pp. 1-3.
[83] J. Kim, S.-R. Moon, S.-H. Yoo, and C.-H. Lee, "800 Gb/s (80 × 10 Gb/s) capacity WDM-PON
based on ASE injection seeding," Optics Express, vol. 22, pp. 10359-10365, 2014/05/05
2014.
[84] Z. Al-Qazwini and K. Hoon, "Ultranarrow Spectrum-Sliced Incoherent Light Source for 10-
Gb/s WDM PON," Journal of Lightwave Technology, vol. 30, pp. 3157-3163, 2012.
[85] J.-Y. Kim, S.-H. Yoo, S.-R. Moon, D. C. Kim, and C.-H. Lee, "400 Gb/s (40 × 10 Gb/s) ASE
Injection Seeded WDM-PON based on SOA-REAM," in Optical Fiber Communication
Conference (OFC), Anaheim, CA, USA, 2013, p. OW4D.4.
[86] G. Talli, A. Naughton, S. Porto, C. Antony, P. Ossieur, and P. D. Townsend, "Advantageous
Effects of Gain Saturation in Semiconductor Optical Amplifier-Based Integrated Reflective
Modulators," Journal of Lightwave Technology, vol. 32, pp. 392-401, 2014.
[87] S. Dong Jae, Y. C. Keh, J. W. Kwon, E. H. Lee, J. K. Lee, M. K. Park, et al., "Low-cost WDM-
PON with colorless bidirectional transceivers," Journal of Lightwave Technology, vol. 24,
pp. 158-165, 2006.
[88] G. de Valicourt, D. Make, C. Fortin, A. Enard, F. Van-Dijk, and R. Brenot, "10 Gbit/s
modulation of RSOA without any electronic processing," in Optical Fiber Communication
Conference (OFC), Los Angeles, CA, USA, 2011, pp. 1-3.
[89] B. Schrenk, G. de Valicourt, M. Omella, J. A. Lazaro, R. Brenot, and J. Prat, "Direct 10-Gb/s
Modulation of a Single-Section RSOA in PONs With High Optical Budget," IEEE Photonics
Technology Letters, vol. 22, pp. 392-394, 2010.
[90] K. Y. Cho and Y. C. Chung, "10-Gb/s Operation of RSOA for WDM PON Using Return-to-
Zero Modulation Format," in Optical Fiber Communication Conference (OFC), Los Angeles,
California, 2012, p. OTh1F.2.
[91] S. Hyun Kyu, K. Hoon, and Y. C. Chung, "20-Gb/s Polar RZ 4-PAM Transmission Over 20-km
SSMF Using RSOA and Direct Detection," IEEE Photonics Technology Letters, vol. 27, pp.
1116-1119, 2015.
[92] H. Shim, H. Kim, and Y. C. Chung, "20-Gb/s Operation of RSOA using Polar Return-to-Zero
4-PAM Modulation Format and Direct Detection," in Optical Fiber Communication
Conference (OFC), Los Angeles, CA, USA, 2015, p. W1J.2.

147
References

[93] A. Gatto, P. Parolari, L. Marazzi, M. Brunero, S. Barbet, A. Maho, et al., "Off-set filtering for
enhanced transmission in RSOA based WDM-PON," in International Conference on
Transparent Optical Networks (ICTON), 2013, pp. 1-4.
[94] K. Hoon, "10-Gbps upstream transmission for WDM-PON using RSOA and delay
interferometer," in Optical Fiber Communication Conference (OFC), Los Angeles, CA, USA,
2011, pp. 1-3.
[95] I. Papagiannakis, M. Omella, D. Klonidis, A. N. Birbas, J. Kikidis, I. Tomkos, et al.,
"Investigation of 10-Gb/s RSOA-Based Upstream Transmission in WDM-PONs Utilizing
Optical Filtering and Electronic Equalization," IEEE Photonics Technology Letters, vol. 20,
pp. 2168-2170, 2008.
[96] M. Presi, R. Corsini, A. Chiuchiarelli, F. Bottoni, G. Cossu, L. Giorgi, et al., "25 Gb/s
Operation of 1-GHz Bandwidth R-SOA by using DMT and Optical Equalization," in Optical
Fiber Communication Conference (OFC), Anaheim, CA, USA, 2013, p. OW1A.7.
[97] G. Cossu, F. Bottoni, R. Corsini, M. Presi, and E. Ciaramella, "40 Gb/s Single R-SOA
Transmission by Optical Equalization and Adaptive OFDM," IEEE Photonics Technology
Letters, vol. 25, pp. 2119-2122, 2013.
[98] J. Tang, B. Cao, M. Deng, R. Giddings, X. Duan, Q. Zhang, et al., "RSOA Intensity Modulator
Frequency Chirp-Enabled 40Gb/s over 25km IMDD PON Systems," in Optical Fiber
Communication Conference (OFC), Los Angeles, CA, USA, 2015, p. W1J.3.
[99] G. Qi and T. An Vu, "Demonstration of 40-Gb/s WDM-PON System Using SOA-REAM and
Equalization," IEEE Photonics Technology Letters, vol. 24, pp. 951-953, 2012.
[100] Q. Guo and A. V. Tran, "40Gb/s Operation of SOA-REAM in Single-Feeder WDM-PON
[Invited]," IEEE/OSA Journal of Optical Communications and Networking, vol. 4, pp. B77-
B84, 2012/11/01 2012.
[101] S. Yu-Chuan, C. Yu-Chieh, L. Shih-Ying, L. Yi-Cheng, T. Cheng-Ting, W. Hai-Lin, et al., "Effect
of Injection Coherence on Noise and Bandwidth of Long-Cavity Colorless Laser Diode for
Digital Modulation and Transmission," IEEE Journal of Quantum Electronics, vol. 51, pp. 1-
14, 2015.
[102] M.-C. Cheng, Y.-C. Chi, C.-T. Tsai, C.-Y. Lin, and G.-R. Lin, "TO-56-can packaged colorless
WRC-FPLD for QAM OFDM transmission at 42 Gbit/s over 25-km SMF," Optics Express, vol.
23, pp. 22676-22690, 2015/08/24 2015.
[103] R. Brenot, F. Lelarge, A. Shen, F. Pommereau, J. Landreau, O. Legouezigou, et al., "A Low
Cost Solution for Sources and Receivers in WDM-PON," in Optical Fiber Communication
Conference (OFC), San Diego, California, 2009, p. JWA89.
[104] H. Takesue and T. Sugie, "Wavelength channel data rewrite using saturated SOA
modulator for WDM networks with centralized light sources," Journal of Lightwave
Technology, vol. 21, pp. 2546-2556, 2003.
[105] L. Wooram, P. Mahn-Yong, C. Seung-Hyun, L. Jihyun, K. Byoung-Whi, J. Geon, et al.,
"Bidirectional WDM-PON based on gain-saturated reflective semiconductor optical
amplifiers," IEEE Photonics Technology Letters, vol. 17, pp. 2460-2462, 2005.
[106] S. Yu-Chuan, C. Yu-Chieh, C. Hsiang-Yu, and L. Gong-Ru, "Data Erasing and Rewriting
Capabilities of a Colorless FPLD Based Carrier-Reusing Transmitter," IEEE Photonics
Journal, vol. 7, pp. 1-12, 2015.
[107] C. Arellano, V. Polo, C. Bock, and J. Prat, "Bidirectional Single Fiber Transmission Based on
a RSOA ONU for FTTH Using FSK-IM Modulation Formats," in Optical Fiber Communication
Conference (OFC), Anaheim, CA, USA, 2005, p. JWA46.
[108] Q. T. Le, A. Emsia, D. Briggmann, and F. Küppers, "Direct DPSK modulation of chirp-
managed laser as cost-effective downstream transmitter for symmetrical 10-Gbit/s WDM
PONs," Optics Express, vol. 20, pp. B470-B478, 2012/12/10 2012.

148
 References

[109] A. Zeghuzi, H. Schmeckebier, M. Stubenrauch, C. Meuer, C. Schubert, C.-A. Bunge, et al.,

"25 Gbit/s differential phase-shift-keying signal generation using directly modulated
quantum-dot semiconductor optical amplifiers," Applied Physics Letters, vol. 106, p.
213501, 2015.
[110] X. Zhaowen, W. Yang Jing, Z. Wen-De, M. Attygalle, C. Xiao-Fei, W. Yixin, et al., "Carrier-
Reuse WDM-PON Using a Shared Delay Interferometer for Separating Carriers and
Subcarriers," IEEE Photonics Technology Letters, vol. 19, pp. 837-839, 2007.
[111] X. Zhaowen, W. Yang Jing, Z. Wen-De, M. Attygalle, C. Xiaofei, W. Yixin, et al., "WDM-PON
Architectures With a Single Shared Interferometric Filter for Carrier-Reuse Upstream
Transmission," Journal of Lightwave Technology, vol. 25, pp. 3669-3677, 2007.
[112] Z. Vujicic, R. S. Luis, J. M. Delgado Mendinueta, A. Shahpari, N. B. Pavlovic, B. J. Puttnam,
et al., "Demonstration of Wavelength-Shared Coherent PON Using RSOA and Simplified
DSP," IEEE Photonics Technology Letters, vol. 26, pp. 2142-2145, 2014.
[113] Z. Vujicic, R. S. Luis, J. M. D. Mendinueta, A. Shahpari, N. B. Pavlovic, B. J. Puttnam, et al.,
"Self-Homodyne Detection-Based Fully Coherent Reflective PON Using RSOA and
Simplified DSP," IEEE Photonics Technology Letters, vol. 27, pp. 2226-2229, 2015.
[114] J. Prat, "Technologies for a Cost-effective Coherent udWDM-PON," in Optical Fiber
Communication Conference (OFC), Los Angeles, CA, USA, 2015, p. Th3I.3.
[115] J. L. Wei, E. Hugues-Salas, R. P. Giddings, X. Q. Jin, X. Zheng, S. Mansoor, et al., "Wavelength
reused bidirectional transmission of adaptively modulated optical OFDM signals in WDM-
PONs incorporating SOA and RSOA intensity modulators," Optics Express, vol. 18, pp. 9791-
9808, 2010/05/10 2010.
[116] Q. Guo and A. V. Tran, "Demonstration of a 40Gb/s Wavelength-Reused WDM-PON Using
Coding and Equalization [Invited]," IEEE/OSA Journal of Optical Communications and
Networking, vol. 5, pp. A119-A126, 2013/10/01 2013.
[117] N. Nadarajah, L. Ka Lun, and A. Nirmalathas, "Upstream Access and Local Area Networking
in Passive Optical Networks Using Self-Seeded Reflective Semiconductor Optical
Amplifier," IEEE Photonics Technology Letters, vol. 19, pp. 1559-1561, 2007.
[118] L. Quang Trung, Q. Deniel, F. Saliou, P. Chanclou, S. Tsyier, G. de Valicourt, et al., "RSOA-
based external cavity laser as cost-effective upstream transmitter for WDM passive optical
network," in Conference on Lasers and Electro-Optics (CLEO), Baltimore, MD, USA, 2011,
pp. 1-2.
[119] H.-C. Kwon, W. Yong-Yuk, and H. Sang-Kook, "A self-seeded reflective SOA-based optical
network unit for optical beat interference robust WDM/SCM-PON link," IEEE Photonics
Technology Letters, vol. 18, pp. 1852-1854, 2006.
[120] J. M. Kang, T. Y. Kim, I. H. Choi, S. H. Lee, and S. K. Han, "Self-seeded reflective
semiconductor optical amplifier based optical transmitter for up-stream WDM-PON link,"
IET Optoelectronics, vol. 1, pp. 77-81, 2007.
[121] F. Y. Shih, C. H. Yeh, C. W. Chow, C. H. Wang, and S. Chi, "Utilization of self-injection Fabry–
Perot laser diode for long-reach WDM-PON," Optical Fiber Technology, vol. 16, pp. 46-49,
1// 2010.
[122] H. Swook, K. Tae-Young, and P. Chang-Soo, "Direct-modulated upstream signal
transmission using a self-injection locked F-P LD for WDM-PON," in European Conference
on Optical Communication (ECOC), Glasgow, Scotland, 2005, pp. 451-452 vol.3.
[123] E. Wong, K. L. Lee, and T. Anderson, "Low-cost WDM passive optical network with directly-
modulated self-seeding reflective SOA," Electronics Letters, vol. 42, pp. 299-301, 2006.
[124] L. Marazzi, P. Parolari, G. de Valicourt, and M. Martinelli, "Network-embedded self-tuning
cavity for WDM-PON transmitter," in European Conference on Optical Communication
(ECOC), 2011, pp. 1-3.

149
References

[125] M. Martinelli, L. Marazzi, P. Parolari, M. Brunero, and G. Gavioli, "Polarization in Retracing

Circuits for WDM-PON," IEEE Photonics Technology Letters, vol. 24, pp. 1191-1193, 2012.
[126] M. J. Connelly, Semiconductor optical amplifiers: Springer Science & Business Media, 2007.
[127] G. P. Agrawal and N. K. Dutta, Semiconductor Lasers: Springer US, 1993.
[128] K. Petermann, "Noise Characterstics of Solitary Laser Diodes," in Laser Diode Modulation
and Noise, T. Okoshi, Ed., ed Dordrecht, The Netherlands: Kluwer Academic Publishers,
1998, pp. 152-208.
[129] J. Leuthold, Advanced Indium Phosphide Waveguide Mach-Zehnder Interferometer All-
optical Switches and Wavelength Converters: Theory, Design Guidelines & Experimental
Results: Hartung-Gorre, 1999.
[130] G. De Valicourt, "Conception, fabrication, et évaluation de modulateurs déportés pour les
réseaux d'accès et radio sur fibre," Télécom ParisTech, 2011.
[131] R. Bonk, Linear and Nonlinear Semiconductor Optical Amplifiers for Next-Generation
Optical Networks: KIT Scientific Publishing, 2013.
[132] G. Eisenstein, R. M. Jopson, R. A. Linke, C. A. Burrus, U. Koren, M. S. Whalen, et al., "Gain
measurements of InGaAsP 1.5 µm optical amplifiers," Electronics Letters, vol. 21, pp. 1076-
1077, 1985.
[133] T. Mukai, T. Saitoh, O. Mikami, and T. Kimura, "Fabry-Perot cavity type 1500 nm InGaAsP
BH-laser amplifier with small optical-mode confinement," Electronics Letters, vol. 19, pp.
582-583, 1983.
[134] M. J. O'Mahony, I. W. Marshall, W. J. Devlin, and J. C. Regnault, "Low-reflectivity
semiconductor laser amplifier with 20 dB fibre-to-fibre gain at 1500 nm," Electronics
Letters, vol. 21, pp. 501-502, 1985.
[135] N. A. Olsson, "Lightwave systems with optical amplifiers," Journal of Lightwave
Technology, vol. 7, pp. 1071-1082, 1989.
[136] T. Mukai, Y. Yamamoto, and T. Kimura, "S/N and Error Rate Performance in AlGaAs
Semiconductor Laser Preamplifier and Linear Repeater Systems," IEEE Transactions on
Microwave Theory and Techniques, vol. 30, pp. 1548-1556, 1982.
[137] E. J. Murphy, J. Ocenasek, C. R. Sandahl, R. J. Lisco, and Y. C. Chen, "Simultaneous single-
fibre transmission of video and bidirectional voice/data using LiNbO₃ guided-
wave devices," Journal of Lightwave Technology, vol. 6, pp. 937-945, 1988.
[138] J. K. Wheeler, J. Ocenasek, and P. P. Bohn, "Two-way transmission using electro-optical
modulator," Electronics Letters, vol. 22, pp. 479-481, 1986.
[139] C. Gibassier, J. Abiven, F. Huet, and J. Saulnier, "Simultaneous two-way light intensity
modulation optical transmission at the same bit rate without a light source in the
subscriber terminal," Electronics Letters, vol. 30, pp. 1162-1163, 1994.
[140] M. D. Feuer, J. M. Wiesenfeld, J. S. Perino, C. A. Burrus, G. Raybon, S. C. Shunk, et al.,
"Single-port laser-amplifier modulators for local access," IEEE Photonics Technology
Letters, vol. 8, pp. 1175-1177, 1996.
[141] N. Buldawoo, S. Mottet, F. Le Gall, D. Sigogne, D. Meichenin, and S. Chelles, "A
semiconductor laser amplifier-reflector for the future FTTH applications," in European
Conference on Optical Communication (ECOC), Edinburgh, UK, 1997, pp. 196-199.
[142] L. Y. Chan, C. K. Chan, T. K. Tong, S. Y. Cheung, F. Tong, and L. K. Chen, "Demonstration of
data remodulation for upstream traffic in WDM access networks using injection-locked FP
laser as modulator," in Optical Fiber Communication Conference (OFC), Anaheim, CA, USA,
2001, pp. WU5-WU5.
[143] A. Mecozzi and J. M. Wiesenfeld, "The Roles of Semiconductor Optical Amplifiers in Optical
Networks," Optics and Photonics News, vol. 12, pp. 36-42, 2001/03/01 2001.

150
 References

[144] J. Leuthold, "Linear and Nonlinear Semiconductor Optical Amplifiers," in Optical Fiber
Communication Conference (OFC), San Diego, CA, USA, 2010, p. OThI3.
[145] J. Leuthold, J. Jaques, and S. Cabot, "All-optical wavelength conversion and regeneration,"
in Optical Fiber Communication Conference (OFC), Los Angeles, CA, USA, 2004, p. WN1.
[146] A. Yariv, Optical Electronics, Fourth ed.: Saunders College Publishing, 1991.
[147] U. Mankong, "Amplified subcarrier multiplexed Passive Optical Network using a
Semiconductor Optical Amplifier," in Electrical Engineering/Electronics, Computer,
Telecommunications and Information Technology, 2009. ECTI-CON 2009. 6th International
Conference on, 2009, pp. 983-986.
[148] K. Carney, R. Lennox, R. Maldonado-Basilio, S. Philippe, F. Surre, L. Bradley, et al., "Method
to improve the noise figure and saturation power in multi-contact semiconductor optical
amplifiers: simulation and experiment," Optics Express, vol. 21, pp. 7180-7195,
2013/03/25 2013.
[149] A. Borghesani, N. Fensom, A. Scott, G. Crow, L. M. Johnston, J. A. King, et al., "High
Saturation Power (>16.5dBm) and Low Noise Figure (<6dB) Semiconductor Optical
Amplifier for C-band Operation," in Optical Fiber Communication Conference (OFC),
Atlanta, GA, USA, 2003, p. ThO1.
[150] G. de Valicourt, D. Make, J. Landreau, M. Lamponi, G. H. Duan, P. Chanclou, et al., "High
Gain (30 dB) and High Saturation Power (11 dBm) RSOA Devices as Colorless ONU Sources
in Long-Reach Hybrid WDM/TDM-PON Architecture," IEEE Photonics Technology Letters,
vol. 22, pp. 191-193, 2010.
[151] S. Mazzucato, H. Carr, re, X. Marie, T. Amand, M. Achouche, et al., "Gain, amplified
spontaneous emission and noise figure of bulk InGaAs/InGaAsP/InP semiconductor optical
amplifiers," IET Optoelectronics, vol. 9, pp. 52-60, 2015.
[152] D. C. Kim, H.-S. Kim, K. S. Kim, B.-s. Choi, J.-S. Jeong, and O. K. Kwon, "10 Gbps SOA-REAM
using Monolithic Integration of Planar Buried-Heterostructure SOA with Deep-Ridge
Waveguide EA Modulator for Colourless Optical Source in WDM-PON," in European
Conference on Optical Communication (ECOC), Geneva, Switzerland, 2011, p.
Tu.5.LeSaleve.5.
[153] P. Ossieur, C. Antony, A. Naughton, S. Porto, N. A. Quadir, A. Clarke, et al., "Hybrid DWDM-
TDMA PONs for Next Generation Access," in Optical Fiber Communication Conference
(OFC), Los Angeles, CA, USA, 2012, p. OW1B.7.
[154] X. Fengnian, W. Jian, V. Menon, and S. R. Forrest, "Monolithic integration of a
semiconductor optical amplifier and a high bandwidth p-i-n photodiode using asymmetric
twin-waveguide technology," IEEE Photonics Technology Letters, vol. 15, pp. 452-454,
2003.
[155] C. Caillaud, P. Chanclou, F. Blache, P. Angelini, B. Duval, P. Charbonnier, et al., "Integrated
SOA-PIN Detector for High-Speed Short Reach Applications," Journal of Lightwave
Technology, vol. 33, pp. 1596-1601, 2015/04/15 2015.
[156] L. H. Spiekman, "Semiconductor optical amplifiers in access networks," in Opto-Electronics
and Communications Conference (OECC), Hong Kong, China, 2009, pp. 1-2.
[157] G. de Valicourt, A. Nkansah, D. Wake, F. Poingt, R. Brenot, and N. J. Gomes, "Performance
Comparison of Single- and Two-Section RSOAs in Radio-Over-Fiber Links," IEEE Photonics
Technology Letters, vol. 25, pp. 190-193, 2013.
[158] S. Ó. Dúill, L. Marazzi, P. Parolari, R. Brenot, C. Koos, W. Freude, et al., "Efficient modulation
cancellation using reflective SOAs," Optics Express, vol. 20, pp. B587-B594, 10 Dec. 2012.
[159] X.-H. Jia, Z.-M. Wu, and G.-Q. Xia, "Detailed theoretical investigation on relative intensity
noise reduction enhancement based on reflective SOAs," Optics & Laser Technology, vol.
44, pp. 1240-1246, 2012.

151
References

[160] L. Marazzi, P. Parolari, A. Boletti, A. Gatto, M. Martinelli, and R. Brenot, "Highly-nonlinear

RSOA RIN compression," in European Conference on Networks and Optical
Communications (NOC), Milan, Italy, 2014, pp. 115-119.
[161] M. Fujiwara, H. Suzuki, and K. Iwatsuki, "Reducing the backreflection impact by using gain-
saturated SOA in WDM single-fiber loopback access networks," in Optical Fiber
Communication Conference (OFC), Anaheim, CA, USA, 2006, p. 3 pp.
[162] J. Sang-Min, H. Moon-Ki, J. Sun-Young, Y. Seung-Min, and H. Sang-Kook, "15-Gbit/s slow
adaptive uplink OFDMA-PON employing channel stabilization technique using low
frequency seed carrier modulation and gain saturated SOA at OLT," in European
Conference on Optical Communication (ECOC), 2013, pp. 1-3.
[163] S. L. Chuang, Physics of photonic devices vol. 80: John Wiley & Sons, 2012.
[164] M. B. Panish, "Heterostructure Injection Lasers," IEEE Transactions on Microwave Theory
and Techniques, vol. 23, pp. 20-30, 1975.
[165] Z. I. Alferov and R. Kazarinov, "Inventor's Certificate No. 181737," ed: Application, 1963.
[166] H. Kroemer, "A proposed class of hetero-junction injection lasers," Proceedings of the IEEE,
vol. 51, pp. 1782-1783, 1963.
[167] B. R. Bennett, R. A. Soref, and J. A. D. Alamo, "Carrier-induced change in refractive index
of InP, GaAs and InGaAsP," IEEE Journal of Quantum Electronics, vol. 26, pp. 113-122, 1990.
[168] N. Holonyak, R. M. Kolbas, R. D. Dupuis, and P. D. Dapkus, "Quantum-well heterostructure
lasers," IEEE Journal of Quantum Electronics, vol. 16, pp. 170-186, 1980.
[169] S. D. Hersee, B. de Cremoux, and J. P. Duchemin, "Some characteristics of the GaAs/GaAlAs
graded‐index separate‐confinement heterostructure quantum well laser structure,"
Applied Physics Letters, vol. 44, pp. 476-478, 1984.
[170] Y. Arakawa and A. Yariv, "Quantum well lasers--Gain, spectra, dynamics," IEEE Journal of
Quantum Electronics, vol. 22, pp. 1887-1899, 1986.
[171] J. Ohtsubo, Semiconductor lasers: stability, instability and chaos vol. 111: Springer, 2012.
[172] G. de Valicourt, M. A. Violas, D. Wake, F. Van-Dijk, C. Ware, A. Enard, et al., "Radio-Over-
Fiber Access Network Architecture Based on New Optimized RSOA Devices With Large
Modulation Bandwidth and High Linearity," IEEE Transactions on Microwave Theory and
Techniques vol. 58, pp. 3248-3258, 2010.
[173] B. Mersali, G. Gelly, A. Accard, J. L. Lafragette, P. Doussiere, M. Lambert, et al., "1.55 mu
m high-gain polarisation-insensitive semiconductor travelling wave amplifier with low
driving current," Electronics Letters, vol. 26, pp. 124-125, 1990.
[174] P. Doussiere, P. Garabedian, C. Graver, D. Bonnevie, T. Fillion, E. Derouin, et al., "1.55 μm
polarisation independent semiconductor optical amplifier with 25 dB fiber to fiber gain,"
IEEE Photonics Technology Letters, vol. 6, pp. 170-172, 1994.
[175] P. Koonath, K. Sangin, C. Woon-Jo, and A. Gopinath, "Polarization-insensitive quantum-
well semiconductor optical amplifiers," IEEE Journal of Quantum Electronics, vol. 38, pp.
1282-1290, 2002.
[176] J. R. Kim, J. S. Lee, S. Park, M. W. Park, J. S. Yu, S. D. Lee, et al., "Spot-size converter
integrated polarization insensitive semiconductor optical amplifiers," IEEE Photonics
Technology Letters, vol. 11, pp. 967-969, 1999.
[177] H. Lianping, R. Zhong, S. Yi, and Y. Siyuan, "Angled-Facet Spot-Size-Converter Integrated
Semiconductor Optical Amplifiers Using Asymmetric Twin Waveguide Technology," IEEE
Photonics Technology Letters, vol. 20, pp. 563-565, 2008.
[178] T. Saitoh and T. Mukai, "1.5µm GaInAsP traveling-wave semiconductor laser amplifier,"
IEEE Journal of Quantum Electronics, vol. 23, pp. 1010-1020, 1987.
[179] M. Nakamura and S. Tsuji, "Single-mode semiconductor injection lasers for optical fiber
communications," IEEE Journal of Quantum Electronics, vol. 17, pp. 994-1005, 1981.

152
 References

[180] M. P. C. M. Krijn, G. W. t Hooft, M. J. B. Boermans, P. J. A. Thijs, T. van Dongen, J. J. M.

Binsma, et al., "Improved performance of compressively as well as tensile strained
quantum-well lasers," in Conference on Lasers and Electro-Optics (CLEO), Anaheim, CA,
USA, 1992, p. CPD23.
[181] J. O’Gorman, A. F. J. Levi, T. Tanbun‐Ek, D. L. Coblentz, and R. A. Logan, "Temperature
dependence of long wavelength semiconductor lasers," Applied Physics Letters, vol. 60,
pp. 1058-1060, 1992.
[182] E. P. O’Reilly and M. Silver, "Temperature sensitivity and high temperature operation of
long wavelength semiconductor lasers," Applied Physics Letters, vol. 63, pp. 3318-3320,
1993.
[183] J. C. L. Yong, J. M. Rorison, and I. H. White, "1.3 µm quantum-well InGaAsP, AlGaInAs, and
InGaAsN laser material gain: a theoretical study," IEEE Journal of Quantum Electronics, vol.
38, pp. 1553-1564, 2002.
[184] T. Fuyuki, R. Yoshioka, K. Yoshida, and M. Yoshimoto, "Long-wavelength emission in photo-
pumped GaAs1−xBix laser with low temperature dependence of lasing wavelength,"
Applied Physics Letters, vol. 103, p. 202105, 2013.
[185] K. Alberi, O. D. Dubon, W. Walukiewicz, K. M. Yu, K. Bertulis, and A. Krotkus, "Valence band
anticrossing in GaBixAs1−x," Applied Physics Letters, vol. 91, p. 051909, 2007.
[186] C. A. Broderick, P. E. Harnedy, and E. P. O'Reilly, "Theory of the Electronic and Optical
Properties of Dilute Bismide Quantum Well Lasers," IEEE Journal of Selected Topics in
Quantum Electronics, vol. 21, pp. 287-299, 2015.
[187] M. Yoshimoto, W. Huang, G. Feng, and K. Oe, "New semiconductor alloy GaNAsBi with
temperature-insensitive bandgap," Physica Status Solidi (b), vol. 243, pp. 1421-1425, 2006.
[188] C. Bilel, M. M. Habchi, A. Ben Nasr, I. Guizani, A. Rebey, and B. El Jani, "Investigation of the
doping and Stark effects on the band structure and optical absorption of 1.55 μm
GaNAsBi/GaAs MQWs," Current Applied Physics, vol. 16, pp. 340-347, 3// 2016.
[189] G. P. Agrawal and N. A. Olsson, "Self-phase modulation and spectral broadening of optical
pulses in semiconductor laser amplifiers," IEEE Journal of Quantum Electronics, vol. 25, pp.
2297-2306, 1989.
[190] J. Wang, Pattern Effect Mitigation Techniques for All-optical Wavelength Converters Based
on Semiconductor Optical Amplifiers: Univ.-Verlag Karlsruhe, 2008.
[191] H. A. Kramers, "Diffusion of Light by Atom," Transaction of Volta Centenary Congress, vol.
2, pp. 548-557, 1927.
[192] R. d. L. Kronig, "On the Theory of Dispersion of X-rays," Journal of Optical Society of
America, vol. 12, pp. 547-557, 1926.
[193] S. A. Gebrewold, L. Marazzi, P. Parolari, R. Brenot, S. P. Duill, R. Bonjour, et al., "Reflective-
SOA Fiber Cavity Laser as Directly Modulated WDM-PON Colorless Transmitter," IEEE
Journal of Selected Topics in Quantum Electronics, vol. 20, pp. 1-9, 2014.
[194] E. Wong, L. Ka Lun, and T. B. Anderson, "Directly Modulated Self-Seeding Reflective
Semiconductor Optical Amplifiers as Colorless Transmitters in Wavelength Division
Multiplexed Passive Optical Networks," Journal of Lightwave Technology, vol. 25, pp. 67-
74, 2007.
[195] L. Marazzi, P. Parolari, R. Brenot, G. de Valicourt, and M. Martinelli, "Network-embedded
self-tuning cavity for WDM-PON transmitter," Optics Express, vol. 20, pp. 3781-3786,
2012/02/13 2012.
[196] M. Sang-Rok, K. Myeong Gyun, S. Byung-Il, and L. Chang-Hee, "2.5 Gb/s injection seeded
DWDM - PON using feed forward noise suppression," in Opto-Electronics and
Communications Conference (OECC), Kyoto, Japan, 2013, pp. 1-2.

153
References

[197] N. Nadarajah, L. Ka Lun, and A. Nirmalathas, "Upstream Access and Local Area Networking
in Passive Optical Networks Using Self-Seeded Reflective Semiconductor Optical
Amplifier," Photonics Technology Letters, IEEE, vol. 19, pp. 1559-1561, 2007.
[198] L. Marazzi, P. Parolari, M. Brunero, A. Gatto, M. Martinelli, R. Brenot, et al., "Up to 10.7-
Gb/s High-PDG RSOA-Based Colorless Transmitter for WDM Networks," IEEE Photonics
Technology Letters, vol. 25, pp. 637-640, 2013.
[199] Q. Deniel, F. Saliou, L. Anet Neto, N. Genay, B. Charbonnier, D. Erasme, et al., "Up to 10
Gbit/s transmission in WDM-PON architecture using External Cavity Laser based on Self-
Tuning ONU," in Optical Fiber Communication Conference (OFC), Los Angeles, CA, USA,
2012, p. JTh2A.55.
[200] R. Brenot, S. Barbet, D. Make, C. Fortin, F. Lelarge, F. Pommereau, et al., "Evidence of gain
modulation in km-long lasers," in IEEE International Semiconductor Laser Conference
(ISLC), 2012, pp. 108-109.
[201] M. Presi, A. Chiuchiarelli, R. Corsini, and E. Ciaramella, "Self-seeding of semiconductor
lasers for next-generation WDM Passive Optical Networks," in International Conference on
Transparent Optical Networks (ICTON), Cartagena, Spain, 2013, pp. 1-4.
[202] M. J. Connelly, "Wideband semiconductor optical amplifier steady-state numerical
model," IEEE Journal of Quantum Electronics, vol. 37, pp. 439-447, 2001.
[203] S. O'Duill, L. Marazzi, P. Parolari, W. Freude, C. Koos, and J. Leuthold, "Modulation
Cancellation Properties of Reflective SOAs," in European Conference on Optical
Communication (ECOC), Amsterdam, The Netherlands, 2012, p. We.2.E.1.
[204] A. M. de Melo and K. Petermann, "On the amplified spontaneous emission noise modeling
of semiconductor optical amplifiers," Optics Communications, vol. 281, pp. 4598-4605,
9/15/ 2008.
[205] W. Jin, A. Maitra, C. G. Poulton, W. Freude, and J. Leuthold, "Temporal Dynamics of the
Alpha Factor in Semiconductor Optical Amplifiers," Journal of Lightwave Technology, vol.
25, pp. 891-900, 2007.
[206] A. P. Bogatov, P. G. Eliseev, and B. N. Sverdlov, "Anomalous interaction of spectral modes
in a semiconductor laser," IEEE Journal of Quantum Electronics, vol. 11, pp. 510-515, 1975.
[207] S. S. Girard, M. M. Piche, C. Hongxin, G. G. Schinn, O. Wang-Yuhl, and B. B. Bouma, "SOA
Fiber Ring Lasers: Single- Versus Multiple-Mode Oscillation," IEEE Journal of Selected
Topics in Quantum Electronics, vol. 17, pp. 1513-1520, 2011.
[208] S. A. Gebrewold, L. Marazzi, P. Parolari, M. Brunero, R. Brenot, D. Hillerkuss, et al.,
"Colorless Self-Seeded Fiber Cavity Laser Transmitter for WDM-PON," in Conference on
Lasers and Electro-Optics (CLEO), San Jose, CA, USA, 2014, p. STu1J.4.
[209] S. L. Woodward, P. P. Iannone, K. C. Reichmann, and N. J. Frigo, "A spectrally sliced PON
employing Fabry-Perot lasers," Photonics Technology Letters, IEEE, vol. 10, pp. 1337-1339,
1998.
[210] P. Healey, P. Townsend, C. Ford, L. Johnston, P. Townley, I. Lealman, et al., "Spectral slicing
WDM-PON using wavelength-seeded reflective SOAs," Electronics Letters, vol. 37, pp.
1181-1182, 2001.
[211] Q. Deniel, F. Saliou, P. Chanclou, D. Erasme, and R. Brenot, "Up to 45km-long amplified
self-seeded RSOA based external cavity for 2.5Gb/s WDM PON transmission," in Optical
Fiber Communication Conference (OFC), Anaheim, CA, USA, 2013, pp. 1-3.
[212] S. Gebrewold, R. Bonjour, S. Barbet, A. Maho, R. Brenot, P. Chanclou, et al., "Self-Seeded
RSOA-Fiber Cavity Lasers vs. ASE Spectrum-Sliced or Externally Seeded Transmitters—A
Comparative Study," Applied Sciences, vol. 5, pp. 1922-1941, 2015.
[213] F. Saliou, G. Simon, P. Chanclou, A. Pizzinat, H. Lin, E. Zhou, et al., "WDM PONs based on
colorless technology," Optical Fiber Technology, vol. 26, pp. 126-134, 12// 2015.

154
 References

[214] A. Banerjee, Y. Park, F. Clarke, H. Song, S. Yang, G. Kramer, et al., "Wavelength-division-

multiplexed passive optical network (WDM-PON) technologies for broadband access: a
review [Invited]," Journal of Optical Networking, vol. 4, pp. 737-758, 2005/11/01 2005.
[215] Y. Horiuchi, "Economical Solutions of the WDM-PON system," in Optical Fiber
Communication Conference (OFC), Los Angeles, CA, USA, 2012, pp. 1-3.
[216] L. F. Henning, P. Monteiro, and A. de Almeida Prado Pohl, "Comparison of LED and RSOA
performance in WDM-PONs," in International Conference on Telecommunications (ICT),
Lisbon, Portugal, 2014, pp. 124-128.
[217] N. J. Frigo, P. P. Iannone, P. D. Magill, T. E. Darcie, M. M. Downs, B. N. Desai, et al., "A
wavelength-division multiplexed passive optical network with cost-shared components,"
IEEE Photonics Technology Letters, vol. 6, pp. 1365-1367, 1994.
[218] J. M. Buset, Z. A. El-Sahn, and D. V. Plant, "Experimental demonstration of a 10 Gb/s RSOA-
based 16-QAM subcarrier multiplexed WDMPON," Optics Express, vol. 22, pp. 1-8,
2014/01/13 2014.
[219] Q. W. Zhang, E. Hugues-Salas, Y. Ling, H. B. Zhang, R. P. Giddings, J. J. Zhang, et al., "Record-
high and robust 17.125 Gb/s gross-rateover 25 km SSMF transmissions of real-time dual-
band optical OFDM signals directly modulated by 1 GHz RSOAs," Optics Express, vol. 22,
pp. 6339-6348, 2014/03/24 2014.
[220] E. Wong, K. L. Lee, and T. B. Anderson, "Directly Modulated Self-Seeding Reflective
Semiconductor Optical Amplifiers as Colorless Transmitters in Wavelength Division
Multiplexed Passive Optical Networks," Journal of Lightwave Technology, vol. 25, pp. 67-
74, 2007/01/01 2007.
[221] U. R. Duarte, J. B. Rosolem, R. S. Penze, and M. A. Romero, "Analysis of ASE-Related
Impairments on Wavelength-Reuse WDM-PONs Based on Self-Seeded Reflective SOAs,"
IEEE/OSA Journal of Optical Communications and Networking, vol. 6, pp. 773-781,
2014/09/01 2014.
[222] T. Pfeiffer, "Next Generation Mobile Fronthaul Architectures," in Optical Fiber
Communication Conference (OFC), Los Angeles, CA, USA, 2015, p. M2J.7.
[223] L. Daewon, S. Hanbyul, B. Clerckx, E. Hardouin, D. Mazzarese, S. Nagata, et al.,
"Coordinated multipoint transmission and reception in LTE-advanced: deployment
scenarios and operational challenges," IEEE Communications Magazine, vol. 50, pp. 148-
155, 2012.
[224] (2014). Common Public Radio Interface (CPRI); Interface Specification V6.1. Available:
www.cpri.info
[225] M. Zhu, X. Liu, N. Chand, F. Effenberger, and G.-K. Chang, "High-Capacity Mobile Fronthaul
Supporting LTE-Advanced Carrier Aggregation and 8×8 MIMO," in Optical Fiber
Communication Conference (OFC), Los Angeles, CA, USA, 2015, p. M2J.3.
[226] P. Parolari, L. Marazzi, M. Brunero, M. Martinelli, R. Brenot, A. Maho, et al., "C- and O-
Band Operation of RSOA WDM PON Self-Seeded Transmitters up to 10Gb/s [Invited],"
IEEE/OSA Journal of Optical Communications and Networking, vol. 7, pp. A249-A255,
2015/02/01 2015.
[227] S. D. Le, Q. Deniel, F. Saliou, A. Lebreton, and P. Chanclou, "16x2.5 Gbit/s and 5 Gbit/s
WDM PON based on self-seeded RSOA," in International Conference on Transparent
Optical Networks (ICTON), Cartagena, Spain, 2013, pp. 1-4.
[228] L. Marazzi, P. Parolari, M. Brunero, A. Mellerio, R. Brenot, and M. Martinelli, "Build-up
analysis of an RSOA-based self-seeded transmitter," in European Conference on Optical
Communication (ECOC), Cannes, France, 2014, pp. 1-3.

155
References

[229] M. Brunero, L. Marazzi, P. Parolari, A. Maho, R. Brenot, and M. Martinelli, "Experimental

evaluation of burst-mode operation of a RSOA-based self-seeded transmitter," in
European Conference on Optical Communication (ECOC), Cannes, France, 2014, pp. 1-3.
[230] U. R. Duarte, M. A. Romero, J. B. Rosolem, R. S. Penze, and A. A. Leonardi, "Enhancement
of chromatic dispersion and post-filtering effects tolerances on spectrum-sliced WDM-
PONs using self-seeded reflective-SOAs," in International Telecommunications Symposium
(ITS), Sao Paulo, Brazil, 2014, pp. 1-5.
[231] S.-H. Yoo, S.-G. Mun, J.-Y. Kim, and C.-H. Lee, "1.25 Gb/s Broadcast Signal Transmission in
WDM-PON Based on Mutually Injected Fabry-Perot Laser Diodes," Journal of the Optical
Society of Korea, vol. 16, pp. 101-106, 2012/06/01 2012.
[232] T. Pfeiffer, "Next Generation Mobile Fronthaul and Midhaul Architectures [Invited],"
IEEE/OSA Journal of Optical Communications and Networking, vol. 7, pp. B38-B45,
2015/11/01 2015.
[233] F. Saliou, P. Chanclou, B. Charbonnier, B. Le Guyader, Q. Deniel, A. Pizzinat, et al., "Up to
15km Cavity Self Seeded WDM-PON System with 90km Maximum Reach and up to
4.9Gbit/s CPRI Links," in European Conference on Optical Communication (ECOC),
Amsterdam, The Netherlands, 2012, p. We.1.B.6.
[234] S. Koenig, R. Bonk, H. Schmuck, W. Poehlmann, T. Pfeiffer, C. Koos, et al., "Amplification
of advanced modulation formats with a semiconductor optical amplifier cascade," Optics
Express, vol. 22, pp. 17854-17871, 2014/07/28 2014.
[235] G. Simon, F. Saliou, P. Chanclou, Q. Deniel, D. Erasme, and R. Brenot, "70km external cavity
DWDM sources based on O-band Self Seeded RSOAs for transmissions at 2.5Gbit/s," in
Optical Fiber Communication Conference (OFC), San Francisco, CA, USA, 2014, p. W3G.5.
[236] P. Parolari, L. Marazzi, M. Brunero, M. Martinelli, R. Brenot, A. Maho, et al., "10-Gb/s
Operation of a Colorless Self-Seeded Transmitter Over More Than 70 km of SSMF," IEEE
Photonics Technology Letters, vol. 26, pp. 599-602, 2014.
[237] N. J. Frigo, "Recent progress in optical access networks," in Optical Fiber Communication
Conference (OFC), San Jose, CA, USA, 1996, pp. 142-143.
[238] K. H. Han, E. S. Son, H. Y. Choi, K. W. Lim, and Y. C. Chung, "Bidirectional WDM PON using
light-emitting diodes spectrum-sliced with cyclic arrayed-waveguide grating," IEEE
Photonics Technology Letters, vol. 16, pp. 2380-2382, 2004.
[239] A. Maho, G. Simon, S. Barbet, F. Lelarge, F. Saliou, P. Chanclou, et al., "Demystification of
the Self-Seeded WDM Access," Journal of Lightwave Technology, vol. 34, pp. 776 - 782
2015.
[240] S. D. Le, A. Lebreton, F. Saliou, Q. Deniel, B. Charbonnier, and P. Chanclou, "Up to 60 km
Bidirectional Transmission of a 16 Channels × 10 Gb/s FDM-WDM PON Based on Self-
Seeded Reflective Semiconductor Optical Amplifiers," in Optical Fiber Communication
Conference (OFC), San Francisco, CA, USA, 2014, p. Th3G.8.
[241] K. Y. Cho, Y. Takushima, and Y. C. Chung, "10-Gb/s Operation of RSOA for WDM PON," IEEE
Photonics Technology Letters, vol. 20, pp. 1533-1535, 2008.
[242] K. Hoon, "10-Gb/s Operation of RSOA Using a Delay Interferometer," IEEE Photonics
Technology Letters, vol. 22, pp. 1379-1381, 2010.
[243] D. Torrientes, P. Chanclou, F. Laurent, S. Tsyier, Y. Chang, B. Charbonnier, et al., "RSOA-
Based 10.3 Gbit/s WDM-PON with Pre-Amplification and Electronic Equalization," in
Optical Fiber Communication Conference (OFC), San Diego, California, 2010, p. JThA28.
[244] X. Chongjin, D. Po, S. Randel, D. Pilori, P. Winzer, S. Spiga, et al., "Single-VCSEL 100-Gb/s
short-reach system using discrete multi-tone modulation and direct detection," in Optical
Fiber Communications Conference (OFC), Anaheim, CA, USA, 2015, pp. 1-3.

156
 References

[245] T. Takahara, T. Tanaka, M. Nishihara, Y. Kai, L. Lei, T. Zhenning, et al., "Discrete Multi-Tone
for 100 Gb/s optical access networks," in Optical Fiber Communication Conference (OFC),
San Francisco, CA, USA, 2014, pp. 1-3.
[246] L. Nadal, M. Svaluto Moreolo, J. M. Fabrega, A. Dochhan, H. Griesser, M. Eiselt, et al., "DMT
Modulation With Adaptive Loading for High Bit Rate Transmission Over Directly Detected
Optical Channels," Journal of Lightwave Technology, vol. 32, pp. 4143-4153, 2014.
[247] P. S. Chow, J. M. Cioffi, and J. Bingham, "A practical discrete multitone transceiver loading
algorithm for data transmission over spectrally shaped channels," IEEE Transactions on
communications, vol. 43, pp. 773-775, 1995.
[248] T. Pollet and M. Peeters, "Synchronization with DMT modulation," IEEE Communications
Magazine, vol. 37, pp. 80-86, 1999.
[249] T. Nakagawa, M. Matsui, T. Kobayashi, K. Ishihara, R. Kudo, M. Mizoguchi, et al., "Non-
Data-Aided Wide-Range Frequency Offset Estimator for QAM Optical Coherent Receivers,"
in Optical Fiber Communication Conference (OFC), Los Angeles, CA, USA, 2011, p. OMJ1.
[250] R. Schmogrow, B. Nebendahl, M. Winter, A. Josten, D. Hillerkuss, S. Koenig, et al., "EVM as
a Performance Measure for Advanced Modulation Formats," IEEE Photonics Technology
Letters, vol. 24, pp. 61-63, 2012.
[251] J. Kani, H. Kawata, K. Iwatsuki, A. Ohki, and M. Sugo, "Design and demonstration of Gigabit
spectrum-sliced WDM systems employing directly modulated super luminescent diodes,"
in Optical Fiber Communication Conference (OFC), Anaheim, CA, USA, 2005, p. 3 pp. Vol.
3.
[252] G. J. Pendock and D. D. Sampson, "Transmission performance of high bit rate spectrum-
sliced WDM systems," Journal of Lightwave Technology, vol. 14, pp. 2141-2148, 1996.
[253] J. L. Wei, A. Hamié, R. P. Gidding, E. Hugues-Salas, X. Zheng, S. Mansoor, et al., "Adaptively
modulated optical OFDM modems utilizing RSOAs as intensity modulators in IMDD SMF
transmission systems," Optics Express, vol. 18, pp. 8556-8573, 2010/04/12 2010.
[254] N. S. André, K. Habel, H. Louchet, and A. Richter, "Adaptive nonlinear Volterra equalizer
for mitigation of chirp-induced distortions in cost effective IMDD OFDM systems," Optics
Express, vol. 21, pp. 26527-26532, 2013/11/04 2013.
[255] L. M. Zhang and F. R. Kschischang, "Staircase Codes With 6% to 33% Overhead," Journal of
Lightwave Technology, vol. 32, pp. 1999-2002, 2014.
[256] R. Bonk, R. Brenot, C. Meuer, T. Vallaitis, A. Tussupov, J. C. Rode, et al., "1.3/1.5 µm QD-
SOAs for WDM/TDM GPON with Extended Reach and Large Upstream/Downstream
Dynamic Range," in Optical Fiber Communication Conference (OFC), San Diego, CA, USA,
2009, p. OWQ1.
[257] P. Parolari, L. Marazzi, M. Brunero, A. Gatto, M. Martinelli, R. Brenot, et al., "10-Gb/s
amplified self-seeding WDM PON transmission exploiting RSOAs," in European Conference
on Networks and Optical Communications (NOC), 2014, pp. 47-50.
[258] T. Komljenovic, D. Babić, and Z. Sipus, "47-km 1.25-Gbps transmission using a self-seeded
transmitter with a modulation averaging reflector," Optics Express, vol. 20, pp. 17386-
17392, 2012/07/30 2012.
[259] M. Sprem, T. Komljenovic, D. Babic, and Z. Sipus, "Colorless optical sources for fiber-optic
access networks of new generation," in International Symposium ELMAR (ELMAR), 2014,
pp. 1-4.
[260] C. Ning and Z. Lei, "Efficient mobile fronthaul for 24x20MHz LTE signals using self-seeded
RSOA with adaptive crosstalk cancellation," in European Conference on Optical
Communication (ECOC), Valencia, Spain, 2015, pp. 1-3.

157
References

[261] l. zhou and N. Cheng, "Cost Effective Self-Seeded WDM Mobile Fronthaul with Adaptive
Crosstalk Cancellation and Bit Loading OFDM," in Asia Communications and Photonics
Conference (ACP), Hong Kong, 2015, p. ASu3E.1.
[262] J. L. Pleumeekers, M. A. Dupertuis, T. Hessler, P. E. Selbmann, S. Haacke, and B. Deveaud,
"Longitudinal spatial hole burning and associated nonlinear gain in gain-clamped
semiconductor optical amplifiers," IEEE Journal of Quantum Electronics, vol. 34, pp. 879-
886, 1998.
[263] J. N. Fehr, M. A. Dupertuis, T. P. Hessler, L. Kappei, D. Marti, F. Salleras, et al., "Hot phonons
and Auger related carrier heating in semiconductor optical amplifiers," IEEE Journal of
Quantum Electronics, vol. 38, pp. 674-681, 2002.
[264] J.-N. Fehr, M.-A. Dupertuis, T. P. Hessler, L. Kappei, D. Marti, P. E. Selbmann, et al., "Direct
observation of longitudinal spatial hole burning in semiconductor optical amplifiers with
injection," Applied Physics Letters, vol. 78, pp. 4079-4081, 2001.
[265] G. de Valicourt, Next Generation of Optical Access Network Based on Reflective-SOA:
INTECH Open Access Publisher, 2012.
[266] L. W. Casperson and J. M. Casperson, "Power self-regulation in double-pass high-gain laser
amplifiers," Journal of Applied Physics, vol. 87, pp. 2079-2083, 2000.
[267] R. Brenot, F. Pommereau, O. Le Gouezigou, J. Landreau, F. Poingt, L. Le Gouezigou, et al.,
"Experimental Study of the Impact of Optical Confinement on Saturation Effects in SOA,"
in Optical Fiber Communication Conference (OFC), Anaheim, CA, USA, 2005, p. OME50.
[268] J. Mork, A. Mecozzi, and G. Eisenstein, "The modulation response of a semiconductor laser
amplifier," IEEE Journal of Selected Topics in Quantum Electronics, vol. 5, pp. 851-860,
1999.
[269] Z. Vujicic, R. P. Dionisio, A. Shahpari, N. B. Pavlovic, and A. Teixeira, "Efficient Dynamic
Modeling of the Reflective Semiconductor Optical Amplifier," IEEE Journal of Selected
Topics in Quantum Electronics, vol. 19, pp. 1-10, 2013.
[270] A. R. Totovic, J. V. Crnjanski, M. M. Krstic, and D. M. Gvozdic, "Numerical Study of the
Small-Signal Modulation Bandwidth of Reflective and Traveling-Wave SOAs," Journal of
Lightwave Technology, vol. 33, pp. 2758-2764, 2015.
[271] C. Antonelli, A. Mecozzi, H. Zhefeng, and M. Santagiustina, "Analytic Study of the
Modulation Response of Reflective Semiconductor Optical Amplifiers," Journal of
Lightwave Technology, vol. 33, pp. 4367-4376, 2015.
[272] R. Brenot, J. G. Provost, O. Legouezigou, J. Landreau, F. Pommereau, F. Poingt, et al., "High
modulation bandwidth reflective SOA for optical access networks," in European
Conference on Optical Communication (ECOC), 2007, pp. 1-2.
[273] H.-S. Kim, B.-S. Choi, K.-S. Kim, D. C. Kim, O. K. Kwon, and D.-K. Oh, "Improvement of
modulation bandwidth in multisection RSOA for colorless WDM-PON," Optics Express, vol.
17, pp. 16372-16378, 2009/09/14 2009.
[274] E. Udvary, "Investigation of semiconductor optical amplifier direct modulation speed," in
International Conference on Transparent Optical Networks (ICTON), Graz, Austria, 2014,
pp. 1-4.
[275] P. Linning, M. Helard, and S. Haese, "On Bit-Loading for Discrete Multi-Tone Transmission
Over Short Range POF Systems," Journal of Lightwave Technology, vol. 31, pp. 4155-4165,
2013.
[276] G. D. Forney and M. V. Eyuboglu, "Combined equalization and coding using precoding,"
IEEE Communications Magazine, vol. 29, pp. 25-34, 1991.
[277] S. C. J. Lee, "Discrete Multimode Modulation for Short-Range Optical Communication,"
PhD, EOC, DEE, TUE, the Netherlands, 2009.

158
 References

[278] A. Garcia-Armada, "SNR gap approximation for M-PSK-Based bit loading," IEEE
Transactions on Wireless Communications, vol. 5, pp. 57-60, 2006.
[279] J. M. Ciofli, "A multicarrier primer," ANSI T1E], vol. 4, pp. 91-157, 1991.
[280] P. Ciblat and M. Ghogho, "Blind NLLS carrier frequency-offset estimation for QAM, PSK,
and PAM modulations: performance at low SNR," IEEE Transactions on Communications,
vol. 54, pp. 1725-1730, 2006.
[281] R. Bhandari, "A useful generalization of the Martinelli effect," Optics Communications, vol.
88, pp. 1-5, 1992/03/01 1992.

159
 Acknowledgement
This dissertation presents research carried out at the Institute of Electromagnetic Fields (IEF),
ETH Zurich, Switzerland. It was partially funded by the European project ERMES (GA 288542-
2012). I would like to acknowledge all the people that helped me to achieve the results presented
in this work.
Foremost, I would like to express my deep gratitude to my supervisor Prof. Juerg Leuthold. I
would like to thank him for his precious advice and guidance during the course of this research.
He is a visionary and he knew how to garner the positive out of all situations. He was always open
minded and supportive and gave me the freedom, which helped me to be self-supportive. I would
also like to thank Prof. Mario Martinelli for agreeing to be my co-examiner. I am grateful for the
interesting discussions we had and his significant feedbacks on the dissertations.
I also would like to mention my appreciation to Dr. Sean O’Duill for taking the time to advise me
at the start of this work. I also thank Dr. Romain Brenot from III-VI Labs for the device samples
used in this dissertation. I acknowledge Dr. Paola Parolari, Dr. Lucia Marazzi, Dr. Marco Brunero
and Dr. Alberto Gatto for the experimental results, which were important for simulation
optimization. I appreciate Dr. Philippe Chanclou and Simon Gael for the fruitful discussions.
I would also like to acknowledge all the help I received from my colleagues here at IEF. I thank
Dr. David Hillerkuss for his feedbacks to improve my publications and support in the lab. I am
indebted to Roman Bonjour, Claudia Hoessbacher, Arne Josten, Benedikt Baeuerle, Wolfgang
Heni, Yannick Salamin and Felix Abrecht for both friendship and scientific help.
Finally yet importantly, I would like to thank God for giving me the wisdom and power. Next, I
express my profound gratitude to my parents who sacrificed a lot form me. They always kept me
in their prayers. I also would like to thank all my brothers, sisters and families both here in
Switzerland and back home in Ethiopia for their love and support.

161
 List of Publications

Journal Papers

[J1] S. A. Gebrewold, R. Bonjour, R. Brenot, D. Hillerkuss, J. Leuthold, "Bit- and Power-

Loading – A Comparative Study on Maximizing the Capacity of Colorless DMT
Transmitters," submitted to Optics Express.

[J2] R. Bonjour, M. Singleton, S. A. Gebrewold, Y. Salamin, F. C. Abrecht, B. Baeuerle, A.

Josten, P. Leuchtmann, C. Hafner, J. Leuthold, "Ultra-Fast Millimeter Wave Beam
Steering," IEEE Journal of Quantum Electronics, vol. 52, pp. 1-8, 2016.

[J3] S. A. Gebrewold, R. Bonjour, S. Barbet, A. Maho, R. Brenot, P. Chanclou,M. Brunero,

L. Marazzi, P. Parolari, A. Totovic, D. Gvozdic, D. Hillerkuss, C. Hafner, J. Leuthold,
"Self-Seeded RSOA-Fiber Cavity Lasers vs. ASE Spectrum-Sliced or Externally Seeded
Transmitters—A Comparative Study," Applied Sciences, vol. 5, pp. 1922-1941, 2015.

[J4] A. Maho, G. Simon, S. Barbet, F. Lelarge, F. Saliou, P. Chanclou, P. Parolari, L. Marazzi,

M. Brunero, M. Martinelli, S. A. Gebrewold, J. Leuthold, R. Brenot, "Demystification
of the Self-Seeded WDM Access," Journal of Lightwave Technology, vol. 34, pp. 776 -
782, 2015.

[J5] R. Bonjour, S. A. Gebrewold, D. Hillerkuss, C. Hafner, and J. Leuthold, "Continuously

tunable true-time delays with ultra-low settling time," Optics Express, vol. 23, pp.
6952-6964, 2015.

[J6] L. Marazzi, P. Parolari, M. Brunero, M. Martinelli, P. Chanclou, Q. Deniel, F. Saliou, D.

S. Le, G. Gavioli, R. Brenot, S. Barbet, A. Maho, S. A. Gebrewold, D. Hillerkuss, J.
Leuthold, "O-Band 10-Gb/s Operation of a Reflective Semiconductor Optical Amplifier
Based Self-Seeded Transmitter for Optical Access Applications," Fiber and Integrated
Optics, vol. 33, pp. 173-183, 2014.

[J7] S. A. Gebrewold, L. Marazzi, P. Parolari, R. Brenot, S. P. Duill, R. Bonjour, D. Hillerkuss,

C. Hafner, J. Leuthold, "Reflective-SOA Fiber Cavity Laser as Directly Modulated WDM-
PON Colorless Transmitter," IEEE Journal of Selected Topics in Quantum Electronics,
vol. 20, pp. 1-9, 2014.

[J8] F. Saliou, G. Simon, P. Chanclou, M. Brunero, L. Marazzi, P. Parolari, M. Martinelli, R.

Brenot, A. Maho, S. Barbet, G. Gavioli, G. Parladori, S. A. Gebrewold, J. Leuthold,
"125km Long Cavity based on Self-Seeded RSOAs Colorless Sources for 2.5Gbit/s
DWDM Networks," Journal of Lightwave Technology, vol. 33, pp. 1602 - 1607 , 2014.

[J9] P. Parolari, L. Marazzi, M. Brunero, A. Gatto, M. Martinelli, P. Chanclou, Q. Daniel, F.

Saliou, D. S. Le, R. Brenot, S. Barbet, F. Lelarge, S. A. Gebrewold, S. P. Duill, D.

163
List of Publications

Hillerkuss, J. Leuthold, G. Gavioli, P. Galli, "Self-tuning transmitter for fibre-to-the-

antenna PON networks," Optical Switching and Networking, vol. 14, pp. 25-31, 2014.

Conference Contributions

[C1] S. A. Gebrewold, R. Brenot, R. Bonjour, A. Josten, B. Baeuerle, D. Hillerkuss, J.

Leuthold, "Colorless Low-Cost RSOA Based Transmitters Optimized for Highest
Capacity Through Bit- and Power-Loaded DMT," in Optical Fiber Communication
Conference (OFC), Anaheim, CA, USA, 2016, p. Tu2C.4.

[C2] F. Saliou, G. Simon, P. Chanclou, M. Brunero, L. Marazzi, P. Parolari, M. Martinelli, R.

Brenot, A. Maho, S. Barbet, G. Gavioli, G. Parladori, S. A. Gebrewold, L. Leuthold,
"Self-Seeded RSOAs WDM PON Field Trial for Business and Mobile Fronthaul
Applications," in Optical Fiber Communication Conference (OFC), Los Angeles, CA,
USA, 2015, p. M2A.2.

[C3] J. Leuthold, R. Bonjour, S. A. Gebrewold, S. Dash, W. Heni, C. Hoessbacher, C. Haffner,

Y. Salamin, B. Baeuerle, A. Josten, P. Ma, P. Leuchtmann, Y. Fedoryshyn, D. Hillerkuss,
"The Path towards 100 Gbit/s Wireless Communications [Keynote] " in Progress In
Electromagnetics Research Symposium (PIERS) 2015 p. 1050

[C4] M. Brunero, P. Parolari, L. Marazzi, M. Martinelli, R. Brenot, A. Maho, S. Barbet, G.

Gavioli, G. Parladori, S. A. Gebrewold, L. Leuthold, F. Saliou, G. Simon, P. Chanclou,
"WDM PON RSOA-based self-tuning transmitters: An insight from the EU FP7 ERMES
project," in International Conference on Transparent Optical Networks (ICTON),
Budapest, Hungary, 2015, pp. 1-4.

[C5] S. A. Gebrewold, R. Bonjour, D. Hillerkuss, C. Hafner, J. Leuthold, and R. Brenot,

"Adaptive Subcarrier multiplexing maximizing the performance of a bandwidth-
limited colorless self-seeded reflective-SOA," in Photonics in Switching (PS), Florence,
Italy, 2015, pp. 163-165.

[C6] R. Bonjour, M. Singleton, S. A. Gebrewold, B. Baeuerle, A. Josten, Y. Salamin, P.

Leuchtmann, D. Hillerkuss, C. Hafner, J. Leuthold, "Time-space division multiplexing
enabled by ultra-fast beam steering," in International Topical Meeting on Microwave
Photonics (MWP), Paphos, Cyprus, 2015, pp. 1-3.

[C7] R. Bonjour, S. A. Gebrewold, D. Hillerkuss, C. Hafner, and J. Leuthold, "Ultra-fast

tunable true-time delay using complementary phase-shifted spectra (CPSS)," in
Optical Fiber Communications Conference and Exhibition (OFC), Los Angeles, CA, USA,
2015, pp. W2A.67.

[C8] S. A. Gebrewold, L. Marazzi, P. Parolari, M. Brunero, R. Brenot, D. Hillerkuss, C.

Hafner, J. Leuthold, "Colorless Self-Seeded Fiber Cavity Laser Transmitter for WDM-
PON," in Conference on Lasers and Electro-Optics (CLEO), San Jose, CA, USA, 2014, p.
STu1J.4.

164
 List of Publications

[C9] P. Chanclou, A. Pizzinat, F. Le Clech, T. L. Reedeker, Y. Lagadec, F. Saliou, B. Le

Guyader, L. Guillo, Q. Deniel, S. Gosselin, S. D. Le, T. Diallo, R. Brenot, F. Lelarge, L.
Marazzi, P. Parolari, M. Martinelli, S. P. Duill, S. A. Gebrewold, D. Hillerkuss, L.
Leuthold, G. Gavioli, P. Galli, "Optical fiber solution for mobile fronthaul to achieve
cloud radio access network," in Future Network and Mobile Summit
(FutureNetworkSummit), Lisbon, Portugal, 2013, pp. 1-11.

165
 Curriculum Vitae
Simon Arega Gebrewold, M.Sc
Born June 1st, 1987, Debrezeit, Ethiopia, nationality: Ethiopian
E-mail: [email protected], phone: +41799484314.
Work Experience
03/2013 – 06/2016 ETH Zurich, Switzerland
Institute of Electromagnetic Fields (IEF)
PhD Student (Research Associate)
 Development of a theoretical model of RSOAs and
self-seeded WDM-PON transmitters
 Experimental characterization of self-seeded RSOA
transmitter
 Implementation of adaptive DMT, advanced
modulation formats (QAM) and DSP for maximizing
transmission capacity
 Implementation of DSP for millimeter wave beam-
steering
 Work package management of the EU project ERMES
 Tutorials for lectures: optical communication
fundamentals, optics and photonics

08/2011 – 09/2011 LIMO Lissotschenko Mikrooptik GmbH, Germany

Research and Development Department
Internship
 Development of laser alignment device

09/2008 – 08/2010 Hawassa University, Ethiopia

Department of Electrical Engineering
Assistant lecturer
 Course on fundamental of electronic circuits
Education
12/2012 – 06/2016 ETH Zurich, Switzerland
Institute of Electromagnetic Fields (IEF)
PhD Study 1 0F

 Preliminary thesis title: “RSOA Based Colorless WDM-

PON Transmitters” (ERMES Project)

1
The first 3 months of the PhD were at IPQ, KIT, Germany, then moved to ETH Zurich with the
group.

167
Curriculum Vitae

10/2010 – 09/ 2012 Karlsruhe Institute of Technology (KIT), Germany

Karlsruhe School of Optics and Photonics (KSOP)
M.Sc. Degree in Optics and Photonics
 Grade: 1.1 2 1F

 Final thesis: “Investigation of 1 to 10 Gbit/s OFDM

System Performance for Wireless Backhauling”
(CONDOR Project)

12/2004 – 07/2008 Haramaya University, Ethiopia

Department of Electrical Engineering
B.Sc. Degree in Electrical Engineering
 Grade 3: 3.76
2F

2
The German grading scale ranges from 1 to 6, where 1 is the highest and 6 is the lowest.
3
The Ethiopian scale of grades ranges from 1 to 4, with 4 being the highest grade.

168