Migration Towards Next Generation Optical

Access and Transport Networks

KUN WANG

Doctoral Thesis in Information and Communication Technology

School of Information and Communication Technology
KTH Royal Institute of Technology
Stockholm, Sweden, 2017
 KTH Royal Institute of Technology
School of Information and Communication Technology
SE-164 40 Kista, SWEDEN
TRITA-ICT 2017:13
ISBN: 978-91-7729-404-7

Akademisk avhandling som med tillstånd av Kungl Tekniska Högskolan framlägges till offentlig
granskning för avläggande av doktorsexamen i Informations- och kommunikationsteknik,
måndagen den 12 juni 2017 klockan 14.00 i Ka-Sal C (Sal Sven-Olof Öhrvik) Electrum, Kungl
Tekniska Högskolan Kistagången 16, Kista.

© Kun Wang, June 2017

Tryck: Universitetsservice US-AB

ii
Abstract

By 2020 there will be 50 billion connected devices over the Internet. With the fast-
increasing data traffic demand in both fixed and mobile networks, network operators
need to migrate networks towards next generation solutions. The network migration
requires the enormous investment in equipment and infrastructure, while the revenues
are not expected to grow significantly. Therefore, one of the main challenges for network
operators is to find out a proper cost-effective optical network solution that can match
future high capacity demand and flexibly support multiple network services on a
common network infrastructure.

The first part of the thesis addresses the Active Optical Network (AON) and its
migration strategies towards Next Generation Optical Access (NGOA) solutions. Several
migration strategies are proposed from the perspective of network topology, data plane
and control plane. A general methodology for Techno-Economic analysis has been
developed and applied to the Total Cost of Ownership (TCO) calculation of different
NGOA solutions. The thesis provides a complete cost evaluation of AON migration
paths, which can be used by network operators to assess the economic feasibility of
network migration.

A converged Optical Transport Network (OTN) that can serve both fixed and mobile
network services is beneficial from the cost-saving perspective. However, the different
types of services, require different network performance. The second part of the thesis
focuses on the investigation of the converged OTN that can be flexibly and timely
adjusted to satisfy varying service conditions. A programmable OTN featured with
Wavelength Division Multiplexing (WDM) in the data plane and Software Defined
Networking (SDN) in control plane has been proposed. To demonstrate the benefits of
the converged OTN, the thesis also provides a multi-domain orchestration architecture
for the multiple network services. The resource orchestration, across three network
domains: OTN, mobile network and cloud, enables agile service creation and optimized
resource allocation among the multiple domains.

Keywords: Fiber To The Home (FTTH), Mobile backhaul/fronthaul, Next Generation

Optical Access network, Optical Transport Network (OTN), Network migration,
Techno-Economic, Total Cost of Ownership (TCO), Capital expenditures (CAPEX),
Operational expenditures (OPEX), Softer Defined Networking (SDN)

iii
Sammanfattning

År 2020 kommer det att finnas 50 miljarder anslutna enheter via Internet. Med den
snabbt ökande datatrafiken i både fasta och mobila nät måste nätoperatörer migrera
nätverk till nästa generations lösningar. Nätverksmigrationen kräver den enorma
investeringen i utrustning och infrastruktur, medan intäkterna inte förväntas öka
betydligt. En av de största utmaningarna för nätoperatörer är därför att hitta en riktig
kostnadseffektiv optisk nätverkslösning som kan matcha framtida efterfrågan på hög
kapacitet och flexibelt stödja flera nätverkstjänster på en gemensam
nätverksinfrastruktur.

Den första delen av avhandlingen behandlar Active Optical Network (AON) och dess
migrationsstrategier för lösningar för nästa generationens optiska åtkomst (NGOA).
Flera migrationsstrategier föreslås ur perspektivet av nätverkstopologi, dataplan och
kontrollplan. En generell metod för Techno-Economic Analysis har utvecklats och
tillämpats på Total Cost of Ownership (TCO) beräkning av olika NGOA-lösningar.
Avhandlingen ger en komplett kostnadsbedömning av AON-migrationsvägar, som kan
användas av nätoperatörer för att bedöma den ekonomiska genomförbarheten för
nätverksmigration.

Ett konvergerat optiskt transportnätverk (OTN) som kan tjäna både fasta och mobila
nätverkstjänster är fördelaktigt från det kostnadsbesparande perspektivet. De olika
typerna av tjänster kräver dock olika nätverksprestanda. Den andra delen av
avhandlingen fokuserar på undersökningen av den konvergerade OTN som kan
anpassas flexibelt och i rätt tid för att uppfylla olika servicevillkor. En programmerbar
OTN som är utrustad med Wavelength Division Multiplexing (WDM) i dataplanet och
Software Defined Networking (SDN) i kontrollplanet har föreslagits. För att visa
fördelarna med den konvergerade OTN, ger uppsatsen också en multi-domän-
orkestreringarkitektur för de flera nätverkstjänsterna. Resursföreställningen, över tre
nätverksdomäner: OTN, mobilnät och moln, möjliggör smidig serviceuppbyggnad och
optimerad resursallokering bland de flera domänerna.

iv
Acknowledgements

My journey to the PhD degree would never have been possible without the help of many
people. It is my great pleasure to take this opportunity to thank them for the support
and advice that I have received over the past years.

Firstly, I would like to express my deepest gratitude to my main supervisor Prof. Lena
Wosinska and co-supervisor Assoc. Prof. Jiajia Chen at KTH for accepting me as PhD
student and providing me tremendous support and guidance during these years. I am
thankful to my co-supervisor Prof. Kjell Brunnström at RISE Acreo for providing me the
opportunity to pursue the industrial doctoral studies. His wide knowledge and
experience have been always of great value for me.

Thanks also to my colleagues at KTH Optical Networks Laboratory (ONLab) and all my
colleagues at RISE Acreo for their inspiring, encouragement, and creation of a pleasant
working environment.

I would like to thank Dr. Carmen Mas Machuca at Technical University of Munich, Dr.
Peter Öhlén, Dr. Ahmad Rostami, Dr. Zere Ghebretensae, Dr. Patryk Urban, Dr. Björn
Skubik at Ericsson research for their support and precious advice. Special thanks to Prof.
Mårten Sjöström and Prof. Mikael Gidlund at Mid Sweden University, for their support
and inspirations during my PhD journey.

My great appreciation to Dr. Andrey Kobyakov, Dr. Aleksandra Boskovic, Dr. Anthony
NgʹOma, and David Peters at Corning Inc., for their encouragement and support on my
PhD study.

Furthermore, I would like to thank my parents, families and friends who gave me help
and encouragement during my studies.

v
List of abbreviations
3D Tree-dimensional
5G-PPP 5G Infrastructure Public Private Partnership
ADSL Asymmetric Data Subscription Line
AON Active Optical Network
API Application-Programming Interface
ARPU Average Revenue Per User
AS Active Star
AWG Arrayed Waveguide Grating
BBU BaseBand Unit
BPMN Business Process Modelling and Notation tool
C&M Control and Management
CAPEX Capital Expenditures
CDN Content Delivery Network
CO Central Office
CoMP Coordinated multipoint transmission
CPRI Common Public Radio Interface
CU Cost Unit
CWDM Coarse Wavelength Division Multiplexing
DSL Data Subscription Line
DU Dense Urban
DWDM Dense Wavelength Division Multiplexing
EPON Ethernet Passive Optical Network
FE Fast Ethernet
FM Fault Management
FSAN Full Service Access Network
FTTH Fiber To The Home
FTTN/C/B Fiber To The Node/Curb/Cabinet/Building
GE Gigabit Ethernet
GPON Gigabit-compatible Passive Optical Network
HetNet Heterogeneous network
HFC Hybrid Fiber-Coaxial
IoT Internet of Things
LAN Local Area Network
MAN Metro Access Node

vi
MIMO Multiple Inputs Multiple Outputs
NC Node Consolidation
NE Network Equipment
NFV Network Function Virtualization
NGFI Next Generation Fronthaul Interface
NGMN Next Generation Mobile Networks
NGOA Next Generation Optical Access
NG-PON2 Next Generation-Passive Optical Network 2
NMS Network Management System
NP Network Provider
ODF Optical Distribution Frame
OLT Optical Line Terminal
ONT Optical Network Terminal
OPEX Operational Expenditures
OTN Optical Transport Network
PIP Physical Infrastructure Provider
PON Passive Optical Network
PoP Point of Presence
PtMP Point to MultiPoint
PtP Point to Point
QoE Quality of Experience
QoS Quality of Service
R Rural
RAN Radio Access Network
RF Radio Frequency
RG Residential Gateway
RN Remote Node
ROADM Reconfigurable Optical Add/Drop Multiplexer
RoE Radio over Ethernet
RRU Remote Radio Unit
SDN Software Defined Networking
SP Service Provisioning
TCO Total Cost of Ownership
TDM Time Division Multiplexing
TP Transponder
TWDM Time and Wavelength Division Multiplexing
vii
U Urban
UDWDM Ultra-Dense Wavelength Division Multiplexing
UHD Ultra High Definition
UPS Uninterrupted Power Supply
VDSL Very high speed Data Subscription Line
VR Virtual Reality
WDM Wavelength Division Multiplexing
XG-PON 10 Gigabit-compatible PON
XGS-PON 10 Gigabit compatible symmetric PON

viii
List of publications
Publications included in this thesis:
Paper I: K. Wang, A. Gavler, C. M. Machuca, L. Wosinska, K. Brunnström and J. Chen,
"Migration strategies for FTTx solutions based on active optical networks," in
IEEE Communications Magazine, vol. 54, no. 2, pp. 78-85, 2016.
Paper II: K. Wang, C. Mas Machuca, L. Wosinska, P. J. Urban, A. Gavler, K. Brunnström,
and J. Chen, "Techno-Economic Analysis of Active Optical Network Migration
Toward Next-Generation Optical Access," IEEE/OSA Journal of Optical
Communications and Networking (JOCN), vol.9, pp. 327-341, 2017.
Paper III: C. Mas Machuca, M. Kind, K. Wang, K. Casier, M. Mahloo, and J. Chen,
"Methodology for a Cost Evaluation of Migration towards NGOA Networks",
IEEE/OSA Journal of Optical Communications and Networking (JOCN), vol. 5,
no. 12, pp.1456 – 1466, 2013
Paper IV: C. Mas Machuca, K. Wang, M. Kind and K. Casier, "Total cost comparison of
Next Generation Optical Access Networks with node consolidation," IEEE 17th
European Conference on Networks and Optical Communications (NOC), 2012.
Paper V: C. Mas Machuca, K. Wang, S. Verbrugge, K. Casier, M. Kind, R. Hülsermann
and Sandro Krauss, "Cost-based assessment of NGOA architectures and its
impact in the business model," 11th Conference of Telecommunication, Media
and Internet Techno-Economics (CTTE), Ghent University, Department of
Information technology, 2012.
Paper VI: K. Wang, C. P. Larsen, A. Gavler, B. Lannoo, D. Chiaroni and M. Popov, "A
comparative model and techno-economic analysis of next generation AON
Ethernet and TDM PON," IEEE/OSA/SPIE Asia Communications and Photonics
Conference and Exhibition (ACP), 2010.
Paper VII: A. Rostami, K. Wang, Z. Ghenretensae, P. Öhlen, and B. Skubic, "First
Experimental Demonstration of Orchestration of Optical Transport, RAN and
Cloud based on SDN," Optical Fiber Communication Conference (OFC) Post
Deadline Papers, OSA Technical Digest, paper Th5A.7, 2015.

ix
Publications not included in the thesis:

1. A. Rostami, P. Öhlén, K. Wang, Z. Ghebretensae, Bjorn Skubic, M. Santos, and A. Vidal,

"Orchestration of RAN and Transport Networks for 5G: An SDN Approach," in IEEE
Communications Magazine, vol. 55, no. 4, pp. 64-70, 2017.
2. P. Öhlén, B. Skubic, A. Rostami, M. Fiorani, P. Monti, Z. Ghebretensaé, J. Mårtensson,
K. Wang, and L. Wosinska, "Data Plane and Control Architectures for 5G Transport
Networks," Journal of Lightwave Technology, vol. 34, pp.1501-1508, 2016.
3. K. Wang, M. Kihl, A. Gavler, M. Du, and C. Lagerstedt, "Power Consumption Analysis
of FTTH Networks", Proceedings of the 10th International Conference on Digital
Telecommunications (ICDT), 2015.
4. K. Wang, B. Andrén, M. Hussain, K. Brunnström, and J. Osterman, "Perception and
annoyance of crosstalk in stereoscopic 3D projector systems", Proceedings of SPIE 9011,
Stereoscopic Displays and Applications XXV, vol. 901125, 2014.
5. B. Andrén, K. Brunnström, and K. Wang, Readability of Displays in Bright Outdoor
Surroundings. SID Symposium Digest of Technical Papers, vol. 45, 2014.
6. H. Rizek, K. Brunnström, K. Wang, B. Andrén, and M. Johanson, "Subjective evaluation
of a 3D videoconferencing system," Proc. SPIE 9011, Stereoscopic Displays and
Applications XXV, 90110A, 2014.
7. S. Tavakoli, K. Brunnström, K. Wang, B. Andrén ; M. Shahid, and N. Garcia, "Subjective
quality assessment of an adaptive video streaming model," Proc. SPIE 9016, Image
Quality and System Performance XI, 90160K, 2014.
8. K. Wang, "Stereoscopic 3D Video Quality of Experience: impact of coding, transmission
and display technologies", Licentiate Thesis, Mid Sweden University, 2013.
9. K. Wang, K. Brunnström, M. Barkowsky, M. Urvoy, M. Sjöström, P. Le Callet, S.
Tourancheau and B. Andrén, "Stereoscopic 3D video coding quality evaluation with 2D
objective metrics," Proc. SPIE 8648, Stereoscopic Displays and Applications XXIV,
86481L, 2013.
10. K. Brunnström, K. Wang and B. Andrén, "Simulator sickness analysis of 3D video
viewing on passive 3D TV, "Proc. SPIE 8648, Stereoscopic Displays and Applications
XXIV, 864802, 2013.
11. M. Van der Wee, K. Casier, K. Wang, S. Verbrugge, and M. Pickavet, "Techno-Economic
Evaluation of FTTH Migration for a Network Provider: Comparison of NG-AON and
TWDM-PON," in Asia Communications and Photonics Conference, 2013.
12. V. Kulyk, S. Tavakoli, M. Folkesson, K. Brunnström, K. Wang and N. Garcia, "3D video
quality assessment with multi-scale subjective method," 2013 Fifth International
Workshop on Quality of Multimedia Experience (QoMEX), pp. 106-111, 2013.
13. K. Brunnström, I.V. Ananth, C. Hedberg, K. Wang, B. Andrén, and M. Barkowsky,
"Comparison between Different Rating Scales for 3D TV," SID Symposium Digest of
Technical Papers, vol. 44, 2013.

x
14. K. Wang, M. Barkowsky, K. Brunnstrom, M. Sjostrom, R. Cousseau, P. Le Callet,
"Perceived 3D TV Transmission Quality Assessment: Multi-Laboratory Results Using
Absolute Category Rating on Quality of Experience Scale," IEEE Transactions
on Broadcasting, no.99, 2012.
15. B. Andrén, K. Wang, K. Brunnström, "Characterizations of 3D TV: Active vs Passive",
SID Symposium Digest of Technical Papers, vol. 43, 2012
16. S. Tourancheau, K. Wang, J. Bulat, R. Cousseau, L. Janowski, K. Brunnstrom, and M.
Barkowsky, "Reproducibility of crosstalk measurements on active glasses 3D LCD
displays based on temporal characterization," Proceeding of SPIE 8288, p. 82880Y,
2012.
17. M. Barkowsky, K. Brunnström, T. Ebrahimi, L. Karam, P. Lebreton, P. Le Callet, A.
Perkis, A. Raake, M. Subedar, K. Wang, L. Xing, J. You. "3D-TV System with Depth-
Image-Based Rendering: Architecture, Techniques and Challenges", Book chapter:
Subjective and objective visual quality assessment in the context of stereoscopic 3DTV.
Springer, 2012.
18. K. Wang, M. Barkowsky, R. Cousseau, K. Brunnström, R. Olsson, P. Le Callet and M.
Sjöström, "Subjective evaluation of HDTV stereoscopic videos in IPTV scenarios using
absolute category rating", Proc. SPIE 7863, 78631T, 2011.
19. S. Tombaz, P. Monti, K. Wang, A. Vastberg, M. Forzati and J. Zander, "Impact of
Backhauling Power Consumption on the Deployment of Heterogeneous Mobile
Networks," 2011 IEEE Global Telecommunications Conference - GLOBECOM, 2011.
20. M. Forzati, C. Mattsson, K. Wang and C. P. Larsen, "The uncaptured value of FTTH
networks," the 13th International Conference on Transparent Optical Networks, 2011.
21. M. Barkowsky, S.Tourancheau, K. Brunnström, K. Wang, and B. Andrén. "Crosstalk
Measurements of Shutter Glasses 3D Displays," Proceedings of SID Symposium Digest
of Technical Papers, Vol. 42, No. 1, 2011.
22. M. Barkowsky, K. Wang, R. Cousseau, K. Brunnström, R. Olsson, and P. Le Callet,
"Subjective Quality Assessment of Error Concealment Strategies for 3DTV in the
presence of asymmetric Transmission Errors", IEEE conference Packet Video
Workshop, 2010.
23. S.Tourancheau, K. Wang, J. Bulat, R. Cousseau, L. Janowski, K. Brunnström,and M.
Barkowsky. Reproducibility of crosstalkmeasurements on active glasses 3D LCD
displays based on temporal characterization. In IS& T/SPIE Electronic Imaging, 82880Y,
2010.
24. L. G. Kazovsky, C.P. Larsen, D. Breuer, A. Gavler, M. Popov, K Wang, G. Jacobsen, E.
Weis, C. Lange, S. Wong, S. Yen, V. Gudla, P. Afshar, "European and American research
toward next-generation optical access networks," 2010 12th International Conference on
Transparent Optical Networks, 2010.
25. C. P. Larsen, A. Gavler and K. Wang, "Comparison of active and passive optical access
networks," the 9th Conference of Telecommunication, Media and Internet (CTTE),
Ghent, 2010

xi
Table of Contents
0 Thesis Organization.............................................................................. 1

1 Introduction ............................................................................................ 3
1.1 Overall aim ................................................................................... 3
1.2 Background and motivation ...................................................... 4
1.2.1 Fixed Access Network ..................................................... 4
1.2.2 Mobile / Radio Access Network (RAN) ........................ 5
1.2.3 Optical Transport Network (OTN) ................................ 6
1.3 Problem statements and contributions .................................... 7
1.4 Summary of original work ......................................................... 9

2 Optical Access Network ..................................................................... 14

2.1 AON ............................................................................................ 14
2.2 TDM-PON .................................................................................. 15
2.3 Next Generation Optical Access (NGOA) ............................. 16
2.3.1 10 Gigabit TDM-PON .................................................... 16
2.3.2 WDM-PON ...................................................................... 17
2.3.3 TWDM-PON ................................................................... 18

3 Network migration strategies for AON .......................................... 20

3.1 Topology Migration .................................................................. 20
3.1.1 Star / Tree migration ...................................................... 21
3.1.2 Mesh/Ring migration ..................................................... 22
3.2 Data plane migration ................................................................ 23
3.2.1 AON PtP to WDM-PON ................................................ 24
3.2.2 AON AS to WDM-backhaul ......................................... 25
3.2.3 AON PtP to TWDM-PON ............................................. 25
3.3 Control Plane migration ........................................................... 26

4 Techno-Economic analysis of AON migration .............................. 29

4.1 Methodology .............................................................................. 29
4.2 Preconditions and use cases .................................................... 31
4.2.1 Time frame....................................................................... 32
4.2.2 Business Models ............................................................. 32
4.2.3 Service Requirements..................................................... 33
4.2.4 Service penetration ......................................................... 34
4.2.5 Geographical model ....................................................... 35
4.2.6 Network scope ................................................................ 36
4.3 Technological scenarios ............................................................ 37
4.4 Cost modeling ............................................................................ 38
4.4.1 CAPEX ............................................................................. 38
4.4.2 OPEX ................................................................................ 39
4.4.3 CAPEX and OPEX division by business role ............. 42
xii
4.5 Techno-Economic Analysis Results ........................................ 43
4.5.1 Yearly TCO over 20 years.............................................. 44
4.5.2 TCO breakdown in the migration year ....................... 45
4.5.3 Node consolidation analysis ......................................... 46

5 Converged Optical Transport Network .......................................... 48

5.1 Converged network services ................................................... 49
5.1.1 Mobile backhaul/fronthaul services ............................ 49
5.2 OTN architecture ....................................................................... 51
5.2.1 Data Plane ........................................................................ 51
5.2.2 Control plane................................................................... 52
5.3 Network Orchestration ............................................................. 54
5.3.1 Use case 1: Elastic Mobile Broadband Services .......... 54
5.3.2 Use case 2: Distributed synchronization services ...... 55
5.3.3 Multi-domain orchestration architecture .................... 56
5.3.4 Orchestration of converged services............................ 58

6 Conclusions and future works.......................................................... 60

6.1 Conclusions ................................................................................ 60
6.2 Future work ................................................................................ 62

7 References ............................................................................................. 63

xiii
 Thesis Organization 1

0 Thesis Organization
The thesis addresses a broad set of aspects related to the optical access
and next generation optical transport network. It is organized as
follows.

- Chapter 1 introduces the background knowledge about optical

access and transport network, the motivation and contributions
of the thesis. Brief summaries of each paper included in the thesis
are also presented along with the detailed information about the
author’s contribution in each article.

- Chapter 2 presents a background for different optical access

network solutions, including both existing fiber access solutions,
and Next Generation Optical Access (NGOA) networks for future
deployments.

- Chapter 3 discusses fiber access network migration strategies

from data plane, control plane and network topology
perspectives. The chapter specifically focuses on the Active
Optical Network (AON) migration strategies, which is based on
the author’s contribution in Paper I.

- Chapter 4 concentrates on the Techno-Economic (TE) analysis of

AON migration. It includes a TE methodology for assessing
optical access and aggregation network, Total Cost of Ownership
(TCO) modeling and calculation, TE migration case analysis from
AON to different NGOA networks. This chapter is based on
author’s contribution in Paper II, Paper III, Paper IV, Paper V,
and Paper VI.

- Chapter 5 presents a next generation Optical Transport Network

(OTN) architecture considering both data plane and control plane
aspects. It can accommodate the converged services from both
fixed and mobile network. A multi-service orchestration across
radio access network, cloud network and OTN is also given to
demonstrate the benefits of the proposed OTN solution. This
chapter is based on author’s contribution in Paper VII.
Thesis Organization 2

- Chapter 6 concludes the thesis and proposes some possible topics

for a future work.
 Introduction 3

1 Introduction
Optical fiber communication, as a future proof technology, has its
unique advantages of delivering ultra-high capacity over long distance.
It is the underlying infrastructure of today’s Internet. Most of Internet’s
data at some point must flow through an optical fiber network. It has
been an ideal technology to build nationwide and global backbone
telecommunication networks.

By 2020 there will be 50 billion connected devices over the Internet [1].
Emerging services such as Internet of Things (IoT) [2], Ultra High
Definition (UHD), Three-Dimensional (3D) videos, cloud services, 5G
mobile services are dramatically increasing network capacity demand in
all segments of telecommunication networks. We are at a junction point
where the larger fiber optic community is seeking new communication
technologies to extend the usage of fiber communications from only the
long-haul transmission and core network to other network domains (e.g.
mobile backhaul/fronthaul, metro/access network and data centers) in a
cost-effective way.

1.1 Overall aim

In order to address challenges of huge data traffic increase, capability of
providing emerging services, customers’ satisfaction and more revenue
growth, it is essential for network operators/providers to migrate their
networks towards next generation optical network solutions.

The aim of this thesis is to investigate optimum network migration

strategies and programmable optical network solutions that can meet
future high capacity demand and flexibly support multiple network
services on a common network infrastructure in a cost-effective way.

The thesis addresses both technology and economic aspects of the

network to identify the efficient migration strategies. An innovative
Optical Transport Network (OTN) solution featured Software Defined
Networking (SDN) is also investigated for accommodating multi-
domain network services. The objective of this thesis is to provide tools
and comprehensive guidance for the network operators/providers to
design and evaluate innovative optical network architectures towards
future fixed and mobile convergence.
 Introduction 4

1.2 Background and motivation

There are different existing broadband access technologies that can be
divided into fixed access and mobile/radio access solutions. Both fixed
and mobile networks can offer Internet access, voice and video services.
Generally speaking, fixed access network provides higher capacity than
mobile network, however, the mobile network has superior
performance for mobility services. Both fixed and radio access networks
are supported by OTN in the metro/ aggregation and core network
segment.

1.2.1 Fixed Access Network

The fixed access solutions are mainly based on copper lines and optical
fibers. The major wired access technologies include: Digital Subscriber
Line (DSL), Hybrid Fiber-Coaxial (HFC, normally referred to as the
broadband Internet service offered via cable TV network), and Fiber To
The Home (FTTH).

The traditional DSL network (e.g. Asymmetric DSL, ADSL [3]) can only
offer low speed access up to 8 Mbit/s. Although newer generation DSL
technologies (commercial available), e.g. Very high speed DSL (VDSL2)
[4] can provide higher bitrate than traditional DSL, e.g. 100Mbit/s, the
transmission distance over copper lines can be very short, e.g. up to 500
meters [5]. Recently, with new system technologies [6], even 1 Gbit/s
and 10 Gbit/s are achievable, but again, the transmission distance has to
be shorter and shorter, e.g. 70 meters for 1 Gbit/s and 30 meters for 10
Gbit/s [7]. A similar trend is also observed in the coaxial cable based first
mile access solutions, where the higher bitrate transmitted over coaxial
cables, the shorter transmission distance is allowed.

Because of the electrical properties of copper cable, copper-based access

solutions have greater losses at higher bitrate. Fiber doesn’t have this
problem. It is good at delivering ultra-high capacity over long distance.
Therefore, with the future increasing capacity demand, there is a
foreseen evolution of the access network, where the reach of the fiber
will be extended closer to the end-users, while the copper network will
be reduced accordingly. In an intermediate stage, this transition is
happening as a Fiber To The Node/Curb/Cabinet/Building (FTTN/C/B)
solution. The final stage of this evolution will be the FTTH, where fiber
will reach all the way to the end-users’ premises.
 Introduction 5

The most common FTTH solutions deployed today are Active Optical
Network (AON) [8] Gigabit-capable Passive Optical Network (GPON)
[9] and Ethernet PON (EPON) [8]. AON has been standardized since
2005. It has been widely deployed in Europe, while Sweden in
particular is completely dominated by AON [10]. Several network
operators/providers in Europe also deploy AON, e.g. in Denmark,
Norway, the Netherlands, and Austria. However, during the past 5-10
years, PON architectures became the dominating technology for the new
FTTH deployments. [11][12].

Today, most of fixed broadband access service providers can offer an

average bitrate up to 100Mbit/s [13][14][15]. Although a higher bitrate
service can be offered for a certain number of users by using statistical
multiplexing [16]. However, in a longer term, when a majority of the
users in the network start demanding higher capacity, the current
system will not be able to handle it.

By the end of 2016, annual global IP traffic will pass the zettabyte (ZB)
threshold (to 1 billion terabytes), and will reach 2.3 ZB per year by 2020
[17]. Emerging services, such as UHD TV, Virtual Reality (VR), are
gradually eating up capacity of existing networks and driving bitrate
demands beyond 100 Mbit/s in a short future. According to the
European Commission Digital Agenda [18] by 2020, all European
households should have Internet access at bitrate at least 30 Mbit/s, and
50% or more at bitrate above 100 Mbit/s. In order to achieve this goal,
current access networks have to be migrated towards a future proof
solution that is capable of delivering the new demanding services.

1.2.2 Mobile / Radio Access Network (RAN)

Besides fixed access network, RAN can also offer Internet access and
voice services to end-users. Mobile communication network has evolved
many generations since it was born in 1970s. From the plain voice and
texts based services in the 1st generation (1G)/2G network, today’s
3G/4G network has successfully shifted to data centric services. As a
part of a mobile telecommunication network, RAN can offer similar
services as the fixed access network. It provides connectivity between
the core network and end-users. However, in contrast to the fixed
networks, the transmission media of RAN is air, and it can provide
uninterrupted services during high-speed movement).
 Introduction 6

Mobile data traffic has increased 1000 times during the past few years
[19] and this trend is not going to change. By 2020, the mobile traffic
volumes are expected to grow rapidly, and the number of connected
devices will be 10-100 times higher than today [20]. The next generation
mobile network, 5G [20][21] aims to address these issues and provide
unlimited access to information for people and a large variety of
connected devices. Many organizations like Next Generation Mobile
Networks (NGMN) [22] and 5G Infrastructure Public Private
Partnership (5G-PPP) [23] have defined aggressive performance targets
for the 5G systems to fulfil future requirements, including access bitrates
up to 10 Gbit/s [21]. Those targets not only bring challenges on the
wireless technology side, but also raise many challenges on the wired
backhaul/fronthaul networks. A good transport network solution is a
key aspect of 5G mobile network.

1.2.3 Optical Transport Network (OTN)

Whereas mobile network plans to upgrade the backhaul/fronthaul
network to accommodate future 5G mobile services, fixed access
networks are also facing the need of the network upgrade. So far, the
fixed network and mobile network have been built separately. The
network infrastructure, equipment, topology, functions and network
requirements are largely different. It leads to two independent
investments and network operation. Separate evolution paths on fixed
access and mobile backhaul/fronthaul networks would lead to
tremendous investments. Therefore, one very important question to
network operator/provider comes up: Is there a unified transport
network solution that can serve both fixed network and mobile
backhaul/fronthaul?

OTN has been already used in metro and core network for many years.
It is widely considered as a future-proof technology that can provide
reliable, scalable, and ultra-high capacity transport services. With the
needs of evolution in both fixed and mobile network, now it is an
opportunity for operators to consider a converged OTN solution.

Many researchers have investigated converged network solutions [24].

10G-EPON was proposed in [25] for the convergence between fixed
access and mobile backhaul network in 2012. In [26], the Wavelength
Division Multiplexing (WDM) technology is proposed to converge not
only fixed and mobile network, but also converge OTN in core and
metro segment with access network segment. Next Generation-PON2
 Introduction 7

(NG-PON2) [27] is mentioned in [28] as a converged solution, where

mobile digital fronthaul data can be transmitted by point-to-point Dense
WDM channels, and the fixed access network users can be transported
in the multiple Time Division Multiplexing (TDM) channels in NG-
PON2.

1.3 Problem statements and contributions

Problem 1
Active Optical Network (AON) is a widely deployed fiber access
network solution in Europe. Most of existing AON deployments are
based on Fast Ethernet (FE), which is able to offer a sustainable bitrate
up to 100 Mbit/s per customer. AON operators have strong demands to
investigate possible AON migration strategies that on one hand can
satisfy the ever-growing capacity demand driven by new services, and
on the other hand best fit their current network characteristics and
future network planning.

Contribution 1
The first contribution of the thesis is the investigation of the various
AON migration strategies, i.e., data plane, topology and control plane
aspects. Paper I of the thesis proposes WDM-PON and Time and
Wavelength Division Multiplexed PON (TWDM-PON) as a data plane
migration path for AON Point to Point (PtP) and AON Active Star (AS),
respectively. Mesh/ring topology and node consolidation are proposed
as topology migration options. For the control plane migration, SDN
based solutions are proposed. A guidance for network
operators/providers is provided. The selected migration strategy can be
either based on only one aspect or a combination of different aspects.

Problem 2
In order to offer higher bandwidth to customers, network
operators/providers are looking at the possibilities and advantages of
Next Generation Optical Access (NGOA) solutions. An important
question for network operators/providers is: How to compare and
evaluate those candidates from both technical and economic point of
view?

Contribution 2
The second contribution of the thesis is the general methodology for
performing Techno-Economic analysis of the fiber access network
Introduction 8

solutions. Paper II and Paper III of the thesis propose a funnel process
based methodology. The method includes three major steps. The first
step defines important preconditions (e.g. business models, service
demands). The second step defines proper technological solutions that
can satisfy the preconditions defined in the first step. The third step
merges previous two steps together, and then, translates the
technological and business parameters into costs. This contribution
provides a tool to assess the Total Cost of Ownership (TCO) of different
solutions and find out a technology/architecture that is the most suitable
for the certain needs of network operators and business cases.

Problem 3
Network operators/providers are facing the challenge of huge data
traffic increase, which requires continuous investment in their network
equipment and infrastructure. AON has been massively deployed in
the past. There is a need of migration. One of the key issues for AON
operators is: which migration strategy and NGOA technologies requires
lowest Capital Expenditures (CAPEX) for the migration, and which one
will have lowest Operational Expenditures (OPEX) in the future?

Contribution 3
The third contribution of the thesis is the quantitative investigation of
the Total Cost of Ownership (TCO) of different AON migration
technological solutions, and the comprehensive techno-economic
evaluation. Paper II and studies in Paper IV, Paper V, Paper VI of the
thesis conducted techno-economic analyses of several AON migration
paths. The TCO results show that the CAPEX of migration to TWDM-
PON and WDM-PON is higher than the other considered migration
paths. However, when migration is finished, the OPEX per year in
TWDM-PON and WDM-PON are lower than others. Moreover, the
longer operation period after migration is considered, the higher
economic benefits of the AON migration to TWDM-PON and WDM-
PON can be achieved. The AON migration costs were also analyzed
with respect to different business roles, such as PIP, NP and users. The
TCO results of node consolidation show that the cost saving in the
aggregation network is large enough to cover the increased migration
costs in the access network. In particular, the benefit of node
consolidation during AON migration becomes significant in the rural
areas.
 Introduction 9

Problem 4
A converged OTN which can serve both fixed and mobile network
services is preferred by network operators/providers from the cost-
saving perspective. However, the service requirements may differ
significantly between different types of services, e.g., residential user,
enterprise access, mobile backhaul/fronthaul. In order to satisfy all kinds
of services, the converged OTN needs to be flexibly and timely adjusted
according to the varying service conditions. Therefore, it is a great
challenge to design an OTN supporting a wide range of services on a
common network infrastructure.

Contribution 4
The fourth contribution of the thesis is a programmable OTN solution
that can flexibly and automatically adjust to the converged network
services. An innovative OTN solution featured with WDM in the data
plane and SDN in control plane is proposed in Paper VII of this thesis.
The OTN supports the converged services from both mobile and fixed
networks. Paper VII also introduces a multi-domain orchestration
architecture for the converged network services. Different service
requests are sent to a network orchestrator which coordinates OTN,
mobile and cloud network domains. Then the orchestrator controls the
data plane through the corresponding controller in every domain. This
global orchestration process enables agile and efficient resource
allocation across multiple domains.

1.4 Summary of original work

In this section the publications included in the thesis and the author’s
contributions are briefly summarized.

Paper I. K. Wang, A. Gavler, C. M. Machuca, L. Wosinska, K.

Brunnström and J. Chen, "Migration strategies for FTTx
solutions based on active optical networks," in IEEE
Communications Magazine, vol. 54, no. 2, pp. 78-85, 2016.

Summary:
Active Optical Network, one of the most deployed fiber
access solutions in Europe, is facing the need to be upgraded
to satisfy the ever-growing bandwidth demand driven by
new applications and services. The paper proposes several
migration strategies for AON from data plane, topology,
Introduction 10

and control plane perspectives, and investigates their impact

on the TCO.

Contribution of the author:

Original research ideas, network modeling, design and
proposal of all migration strategies, performance analysis,
TCO analysis, writing of the first draft, rebuttal letter, and
updated versions of the manuscript.

Paper II. K. Wang, C. Mas Machuca, L. Wosinska, P. J. Urban, A.

Gavler, K. Brunnström, and J. Chen, "Techno-Economic
Analysis of Active Optical Network Migration Toward
Next-Generation Optical Access," IEEE/OSA Journal of
Optical Communications and Networking (JOCN), vol.9, pp.
327-341, 2017.

Summary:
Active Optical Network has been widely deployed in
Europe. However, with the increasing traffic demand, the
capacity of the existing AONs is becoming insufficient. This
paper proposes and analyzes several migration paths
towards NGOA networks offering a minimum 300Mbit/s
sustainable bitrate and 1Gbit/s peak bitrate to every end-
customer. Furthermore, this paper provides detailed
descriptions of the network cost modeling and the processes
for AON migration. The TCO for the proposed migration
paths are evaluated taking into account different migration
starting times, customer penetration, node consolidation
and business roles in the fiber access networks.

Contribution of the author:

Original research ideas, development of TCO evaluation
methodology, cost modeling, network modeling, Techno-
Economic analysis, node consolidation analysis, results
interpretation, writing of the first draft, rebuttal letter, and
updated versions of the manuscript.

Paper III. C. Mas Machuca, M. Kind, K. Wang, K. Casier, M. Mahloo,

and J. Chen, "Methodology for a Cost Evaluation of
Migration towards NGOA Networks", IEEE/OSA Journal of
Introduction 11

Optical Communications and Networking (JOCN), vol. 5,

no. 12, pp.1456 – 1466, 2013.

Summary:
In this paper, a complete methodology is presented for
evaluating the TCO of the migration towards a Next
Generation Optical Access network. It contains a detailed
description of which key aspects have to be considered,
which process they affect and how they are translated into
costs in a logical manner. Finally, it also shows how this
methodology has been applied to particular selected cases
and how it gives a detailed view of all costs involved in
migration. This approach opens opportunities to cooperate
in techno-economic research using it as a base.

Contribution of the author:

Contributions for idea development, TCO evaluation
methodology development, cost modeling, network
topology modeling, and part of manuscript writing.

Paper IV. C. Mas Machuca, K. Wang, M. Kind and K. Casier, "Total

cost comparison of Next Generation Optical Access
Networks with node consolidation," IEEE 17th European
Conference on Networks and Optical Communications
(NOC), 2012.

Summary:
This work conducted cost assessment (both CAPEX and
OPEX) of three different NGOA architectures: 80 and 40
channel TWDM-PON, 2-stage WDM-PON and WDM
backhaul AON. The study focused on a green field
deployment scenario. Two types of node consolidation were
studied: hard consolidation and soft consolidation.

Contribution of the author:

Cost modeling and analysis of two NGOA architectures:
two-stage WDM-PON and active optical network
backhauled by WDM. Contributions for idea development
and manuscript writing.
Introduction 12

Paper V. C. Mas Machuca, K. Wang, S. Verbrugge, K. Casier, M.

Kind, R. Hülsermann and Sandro Krauss, "Cost-based
assessment of NGOA architectures and its impact in the
business model," 11th Conference of Telecommunication,
Media and Internet Techno-Economics (CTTE), Ghent
University, Department of Information technology, 2012.

Summary:
This extended the work from Paper II, investigated both
TCO of different NGOA solutions and the impact of
business model on the TCO. Five NGOA were studied:
WDM-PON, ultra-dense WDM-PON, TWDM-PON, two-
stage WDM-PON and WDM backhauled AON. The results
show that there is a large geographical impact on PIP than
NP. NPs have more cost in dense urban area than PIPs. The
OPEX for NP is considerable higher than PIP.

Contribution of the author:

Cost modeling and cost analysis of three NGOA
architectures: wavelength-routed WDM-PON, two-stage
WDM-PON and active optical network backhauled by
WDM. Contributions for idea development and manuscript
writing.

Paper VI. K. Wang, C. P. Larsen, A. Gavler, B. Lannoo, D. Chiaroni

and M. Popov, "A comparative model and techno-economic
analysis of next generation AON Ethernet and TDM-PON,"
IEEE/OSA/SPIE Asia Communications and Photonics
Conference and Exhibition (ACP), 2010.

Summary:
A reference model covering active and passive networks has
been developed for techno-economic evaluation. The paper
compared CAPEX of 10G Passive Optical Network (PON)
and 1G AON – both capable of delivering average 1Gbit/s to
end-users.

Contribution of the author:

Original research idea, development of reference model for
comparing PON and AON, techno-economic modeling and
Introduction 13

analysis, writing manuscript and presentation in the

conference.

Paper VII. A. Rostami, K. Wang, Z. Ghenretensae, P. Öhlen, and B.

Skubic, "First Experimental Demonstration of Orchestration
of Optical Transport, RAN and Cloud based on SDN,"
Optical Fiber Communication Conference (OFC) Post
Deadline Papers, OSA Technical Digest, paper Th5A.7, 2015.

Summary:
The paper presents the first resource orchestration across
DWDM optical transport, RANs and cloud domains based
on SDN. The orchestration enables agile service creation
and optimized resource allocation.

Contribution of the author:

Contributions for idea development, architecture design,
software implementation of service use cases, multi-domain
controllers and orchestrators, experiment setup, results
preparation and analysis.
 Optical Access Network 14

2 Optical Access Network

The most common optical access networks deployed today are Active
Optical Network (AON) and Time Division Multiplexing Passive
Optical Network (TDM-PON). This chapter starts with an introduction
to the exiting AON and TDM-PON in Section 2.1 and 2.2, followed by
Next Generation Optical Access (NGOA) networks in Section 2.3.

2.1 AON
There are two variants of AONs: Point-to-Point (PtP) Ethernet and
Active Star (AS). The PtP architecture is also referred to as homerun
(shown in Figure 2-1 (a)). In this architecture, each subscriber has a
dedicated fiber connection between the home Residential Gateway (RG)
(which for instance can be Optical Network Terminal (ONT) in case of
the Fiber To The Home (FTTH)) and the Ethernet switch at Optical Line
Terminal (OLT) located in the traditional access node, also referred to as
Central Office (CO).

Unlike the PtP architecture, the AON AS has a point-to-multipoint

(PtMP) fiber topology, employing active Remote Node (RN) connected
to the CO and multiple households as illustrated in Figure 2-1 (b). The
RN can be located either in a cabinet or inside the building, e.g. a
basement of a multi-dwelling unit. The Ethernet switch at the RN
aggregates the traffic from a group of subscribers, and it is connected by
a feeder fiber to another Ethernet switch at the CO. Two or more feeder
fibers can be deployed to provide resiliency, but the amount of fibers
used in the AON AS architecture is significantly reduced compared to
the PtP case. Figure 2-1 (c) shows a FTTB/C/N (Fiber To The Build-
ing/Curb/Node) architecture based on AON AS. The optical signals are
terminated at the RN, which is connected to the households via legacy
copper cables.

The earlier deployments of AONs offer a bitrate up to 100 Mbps per

ONT. New deployments of AON have already begun to offer 1 Gbit/s
per subscriber [29][30]. The available optical components and Ethernet
equipment are able to support up to 120 km reach. However, due to the
topological characteristics of AON, especially PtP, the viability of the
solution is limited by the economic factors when considering the
deployment of fiber infrastructures. It is not cost efficient to have an
AON PtP solution over a long-reach fiber infrastructure.
 Optical Access Network 15

Figure 2-1. Current AON based FTTx solutions

2.2 TDM-PON
From fiber topology perspective, PON is similar to the AON AS, where
both of them are point-to-multipoint (PtMP). In contrast to AON AS,
PON only uses passive splitter in the RN, as it is shown in Figure 2-2.
The most common PON deployments today are using Time Division
Multiplexing (TDM). It has two flavors, namely Gigabit-compatible
PON (GPON) [9] and Ethernet PON (EPON) [8]. GPON is capable to
provide 2.5Gbit/s downstream and 1.25 Gbit/s upstream, while EPON
can offer 1.25 Gbit/s in both downstream and upstream direction. The
bandwidth is shared among all users that are connected to the same
PON OLT port. With TDM technology, every ONT is able to
receive/transmit its data in an assigned time slot. In the downstream, the
OLT broadcasts data throughout the fiber links to all users. In the
upstream ONTs can send their data only at the timeslots that OLT
assignes to them.

The physical reach (distance) between OLT and ONT is typically within
20 km for current generation TDM-PON. The reach is related to the
splitting ratio of the power splitter and the transmitter power at OLT.
The higher splitting ratio gives more subscribers that can be supported
by a single PON / feeder fiber. However, it also reduces the received
signal power at every ONT, and therefore the reach has to be shortened.
 Optical Access Network 16

In order to tackle this issue an amplifier can be used either in the CO or

RN. With power amplifications, the reach can be extended up to 60 km
[31].

Figure 2-2. Current FTTH solution: TDM-PON

The number of users that can be connected to a single PON or feeder

fiber depends on many parameters in the PON network planning, e.g.
power budget, reach (distance) and average bitrate. Normally it is below
64 for GPON and EPON. The more users are connected to the same
PON, the shorter distance between the OLT and user can be supported,
and the lower average bitrate can be obtained by every user.

2.3 Next Generation Optical Access (NGOA)

With the expected increase of traffic demand, the current generation
optical access network will no longer be able to satisfy end-users'
demand in a near future. Different NGOA technologies and
architectures have been extensively investigated by many researchers
and organizations. For example, the Full Service Access Network
(FSAN) group [32] has defined the Next Generation of GPON in 2010,
referred to as NG-PON1 [33] as a near/medium term evolution option,
and NG-PON2 [27] in 2016 as a long term evolution option.

2.3.1 10 Gigabit TDM-PON

The 10G TDM-PON is able to provide higher data throughput and
larger split ratio (i.e. more users can share the same feeder fiber) while
reusing the fiber network infrastructure from the existing PONs. ITU-T
10 Gigabit-compatible PON (also known as XG-PON or NG-PON1) and
10GE-PON were standardized in 2010 [33] and 2009 [34], respectively.

XG-PON can support asymmetric traffic at 10 Gbit/s for downstream

and 2.5 Gbit/s for upstream. The symmetric version of XG-PON, called
10 Gigabit compatible symmetric PON (XGS-PON), is also consented
 Optical Access Network 17

recently in 2016 [35]. 10GE-PON also has two versions, namely, the
asymmetric version supporting 10 Gbit/s in downstream and 1 Gbit/s in
upstream, and the symmetric version supporting 10 Gbit/s in both
directions.

2.3.2 WDM-PON
WDM-PON is one of the selected NGOA technology for NG-PON2 [27].
Figure 2-3 shows a typical WDM-PON architecture. In the OLT side,
several colored optical transceivers modulate their data streams at
different wavelengths, and then these different wavelengths are
multiplexed into a single feeder fiber. In the RN of WDM-PON, unlike
the TDM-PON, the power splitter is not required anymore. It is
replaced by a wavelength splitter, e.g. Arrayed Waveguide Grating
(AWG) which routes a dedicated wavelength to every end-user. In
every ONT there is an optical transceiver which can receive/transmit the
dedicated wavelength. Although the topology of the physical fiber
infrastructure of WDM-PON is PtMP, at the wavelength level, the
virtual topology is based on PtP .

Figure 2-3. NGOA solution: WDM-PON

Currently there are a few WDM-PON deployments [36][37] that provide

bitrate up to 100 Mbit/s for every subscriber. The bitrate depends on the
capability of the network equipment (e.g. transceivers in the OLTs and
ONTs). The new generation WDM-PON is able to provide 1 Gbit/s per
ONT. In the NG-PON2 standard, the WDM-PON link can even provide
10 Gbit/s per ONT.

In TDM-PON (as shown in Figure 2-2) one transceiver in the OLT is

serving multiple ONTs, the optical power of the OLT transceiver is
divided multiple times, each ONT only receives a fraction of the optical
 Optical Access Network 18

power from the OLT transceiver. On the other hand, in the WDM-PON
case (as indicated in Figure 2-3), one transceiver in the OLT only serves
one ONT, each ONT gets full power from an OLT transceiver. Therefore
the physical reach of WDM-PON is longer than of TDM-PON. The
number of users that can be multiplexed in a same feeder fiber in WDM-
PON is very much dependent on the channel spacing of the WDM
system, which can be further divided into Coarse WDM (CWDM),
Dense WDM (DWDM) and Ultra-Dense WDM (UDWDM). More
information about different WDM-PON systems can be found in
[38][39].

2.3.3 TWDM-PON
Time and Wavelength Division Multiplexed PON (TWDM-PON)
combines TDM and WDM technology. It is defined by ITU-T [38] as a
primary technology for NG-PON2. The NG-PON2 defined TWDM-PON
system has 40 Gbit/s capacity carried by 4 wavelengths, each of which
can deliver up to 10 Gbit/s peak rate for any user.

Figure 2-4. NGOA solution: TWDM-PON

Figure 2-4 shows an example of TWDM-PON implementation. It can be

simply considered as an integration of four 10G TDM PONs. By using
WDM technology, several 10G TDM-PONs operating at different
wavelength are multiplexed into one feeder fiber. Then four different
10G TDM-PON get separated from each other at RN2 where the
wavelength splitter is located. At RN1 there are power splitters that are
the same as used in 10G TDM-PON systems. Every ONT is equipped
with a transceiver which is tunable. It enables the TWDM-PON system
to assign any of the four different 10G TDM-PON resources to the users.
Optical Access Network 19

One of the key advantages of TWDM-PON is the compatibility with the

legacy PON systems, e.g. GPON and XGPON. Therefore, it allows
operators to reuse their existing equipment and fiber infrastructure. NG-
PON2 supports 40 km reach from OLT to ONT. The longer reach can be
supported by implementing amplifiers either at OLT or RN locations.
 Network migration strategies for AON 20

3 Network migration strategies for

AON
In Chapter 2 different optical access network technologies were briefly
reviewed. This chapter will concentrate on AON and systematically
investigate possible migration strategies from current AON
deployments to NGOA networks. This Chapter is addressing the first
contribution of the thesis described in Section 1.3, which has been
published in Paper I.

AON is a widely deployed FTTx solution in Europe. In 2012 it

represented 78% of the market share [40], which corresponds to a high
technology penetration in Europe. Therefore, AON deployments are
facing the need to be upgraded in order to satisfy the ever-growing
bandwidth demand driven by new applications and services.
Meanwhile, network providers want to save both Capital Expenditures
(CAPEX) and Operational Expenditures (OPEX) from the NGOA
networks to ensure that there is profit coming from the investments.

A comprehensive view of network architecture should include various

aspects, such as data plane, control plane and network topology. This
chapter will consider all these three aspects, concentrate on AON and
systematically investigate possible migration strategies from current
AON deployments to NGOA networks.

3.1 Topology Migration

There are several physical network topologies, e.g. star, tree, bus, ring.
Current fixed access networks were built with a star/tree topology that
is inherited from the legacy telephone networks. Paper I of the thesis
proposed two options for the topology migration: 1) keeping star/tree
topology but carrying out Node Consolidation (NC) and 2) upgrading
the topology to mesh/ring. These two trends have both pros and cons,
and provide different features. We will discuss these two topology
migration cases in Section 3.1.1 and 3.1.2.
 Network migration strategies for AON 21

3.1.1 Star / Tree migration

Figure 3-1 shows an example of the star/tree topology. The COs (also
called local exchange offices) host telephone exchange equipment. At
the time when the first generation FTTx networks were built, the COs
were also used for hosting OLTs and Ethernet switches. For AON PtP,
end-users are directly connected to the CO via star topology. For AON
active star and TDM-PON deployments, end-users are first connected to
the remote nodes (RN) via star topology, and then further connected to
the CO. Metro Access Node (MAN) itself is also a type of CO, but with a
larger size, more functions and a longer distance to end-users than the
normal COs. It aggregates several COs and connects to core edge node
via star/tree topology. In some cases MANs are interconnected [41], as
shown in Figure 3-1 by the dashed lines, forming a logical ring
topology.

Figure 3-1: Star/Tree topology

The star/tree topology used in the access networks was introduced tens
of years ago, key design parameters, such as the distance between the
nodes (including CO, RN, MAN), the number of nodes, and the service
area of each node were limited by the physical reach of copper cable and
switching capabilities of the legacy network infrastructure and
equipment. When optical access network based on such topology is
deployed, network operators may want to take advantage of long reach
and low loss properties of fiber to redesign the network topology, so
that the distance between nodes can be longer, the service area (the
number of customers) of every node can become larger, and therefore
less network nodes are required [41]. These changes can lead to cost
savings for housing and maintenance of access node. This concept is
referred to as Node Consolidation (NC).
 Network migration strategies for AON 22

Figure 3-2: Star/Tree topology with Node consolidation [Paper I]

Figure 3-2 illustrates the concept of NC. In contrast to Figure 3-1, NC

reduces the number of COs, so that the costs associated to those nodes,
e.g. housing, energy and maintenance, can be saved. All network
equipment will therefore be moved to the MANs which serve larger
areas than the traditional COs. The fiber links in the traditional COs
have to be reconnected to MAN. There are many ways to do it. For
example, power or wavelength splitters can be used to aggregate
multiple links into one fiber that continues to the MAN. In this case the
CO is transformed into a RN. The RN can be either an outdoor cabinet
or underground enclosure. The underground deployment of RN can
maximize the benefit of reducing number of nodes because it minimizes
the cost for the RN footprint.

The network migration from current AON to the NC scenario should

take into account the technology evolution of data plane, which will be
further introduced in Section 3.2. For example, transforming an old CO
to a passive RN requires replacement of active Ethernet switches by
passive splitter/filters, and therefore new network components and
multiplexing technologies must be introduced.

3.1.2 Mesh/Ring migration

The mesh/ring topology has more links between nodes than the
star/tree. Nodes are interconnected either in mesh or in a closed loop so
that there are 2 or more paths between a pair of nodes. Figure 3-3 shows
an example of mesh/ring topology.

Unlike the NC approach, the mesh/ring topology migration keeps all

existing network nodes and fiber connections, and adds more
links/fibers between the nodes, so that the new network topology
becomes mesh/ring. The solid lines in Figure 3-3 represent fiber links
 Network migration strategies for AON 23

that already existed in the legacy star/tree topology, while the dashed
lines show the potential links that can be added to form the mesh/ring
structure. The more links added, the higher node degree of mesh/ring
topology.

Figure 3-3. Mesh/Ring topology [Paper I]

Compared to star/tree, mesh/ring topology has better connectivity and

offers better resiliency. When failures occur, alternative paths between
nodes can potentially be found, and hence the impact of the failures can
be reduced. The network resiliency is extremely important for business
customers, public services, healthcare, mobile backhaul/fronthaul, etc.

Mesh/ring topology better suits services like Content Delivery Networks

(CDN) [42] than the NC approach in the star/tree topology. The
mesh/ring can take advantage of the distributed nodes that are close to
the end-users and make use of meshed connectivity to share caches
among neighbouring nodes. Thus, users can get fast video streaming
and enhanced Quality of Experience (QoE). Meanwhile, mesh/ring
topology can keep traffic locally [43], which offloads the data traffic in
metro/core network, and hence improves the network efficiency.

3.2 Data plane migration

Data plane, also known as forwarding/transport plane, is responsible for
forwarding the network traffic between the nodes. The technologies
used in the data plane have a large impact on the cost and performance
of the network (throughput, latency, speed). AON, GPON, EPON, and
DSL are all using different data plane technologies. They are
implemented in the network equipment, but have to fit the properties of
physical infrastructures. In this section, we investigate migration
 Network migration strategies for AON 24

strategies from the data plane perspective, taking into account the
characteristics of the existing AON deployments.

3.2.1 AON PtP to WDM-PON

Figure 3-4 shows a proposed migration path from current AON PtP to
WDM-PON. In the case of node consolidation, a number of COs are
supposed to be closed down to save the costs, therefore all active
network equipment has to be moved into a consolidated MAN. Arrayed
Waveguide Gratings (AWGs) are placed in the location of traditional
COs to replace the AON PtP switches. The AWGs are passive
components and they can be installed either underground or in a
cabinet, where electrical power is not needed. The AWG aggregates a
number of customers into one feeder fiber, which connects to the WDM-
PON OLT (Ethernet switch with another AWG and colored optical
interfaces). Every user connected to the AWG is assigned a dedicated
wavelength from the OLT. All residential gateways at households and
Ethernet switches at MAN must be replaced with the WDM optical
interfaces.

Figure 3-4: Migration path from AON PtP to WDM-PON [Paper II]

AON PtP migration to WDM-PON has the advantage of reusing

existing fiber infrastructure from the legacy AON. The fiber connection
between CO and MAN may also require additional investment if it is
not available from the legacy network. However, thanks to the WDM
technology, the additional amount of fibers required for network
upgrade can be reduced and can be installed in the existing ducts.
 Network migration strategies for AON 25

Consequently, the huge trenching costs can be avoided. There are some
fiber splicing and reconnection work involved at COs.

3.2.2 AON AS to WDM-backhaul

The WDM backhaul solution can be an option for the AON AS
migration as shown in Figure 3-5. The Ethernet switches at RN are
backhauled by WDM technology. New WDM OLTs should be installed
at MAN. AWGs replace the Ethernet switches at the old COs, which
become completely passive. Every switch at RN gets an individual
wavelength from the OLT, and then gray optical transceivers (that don’t
need to involve WDM technology) are used to connect end-users in the
legacy AON AS. Therefore, no changes are needed in the residential
gateway. However, if the residential gateway and switches at RN in
legacy AON AS only has low data-rate (e.g.100Mbps) capability, the
upgrades to Gigabit residential gateway and Gigabit Ethernet (GE)
switches are still required to deal with higher bandwidth.

Figure 3-5: Migration path from AON active star to WDM-backhaul [Paper II]

3.2.3 AON PtP to TWDM-PON

A fully passive solution for the AON AS data plane migration can be
also achieved by using TWDM-PON as shown in Figure 3-6. The active
RNs are replaced by passive power splitters, while the Ethernet
switches at the old COs are replaced by AWGs. Therefore, both RNs and
COs can be potentially eliminated to support node consolidation. This
migration path also requires fiber reconnection at both RNs and COs, as
well as new TWDM OLTs at MAN. At customers’ side, replacement of
RGs/ONTs is required in order to adapt to TWDM technology.
 Network migration strategies for AON 26

Figure 3-6. Migration path from AON active star to TWDM-PON [Paper II]

3.3 Control Plane migration

The control plane is the brain of network equipment. It configures and
manages the equipment and tells the data plane how to deliver data
traffic between network nodes. In this section, we investigate the
migration strategy in the control plane, which can be carried out
together with the network evolutions in the data plane and topology
introduced in the previous sections.

In the traditional network equipment both data plane and control plane
are embedded in one physical box. The devices are geographically
distributed in different locations in the network. With such a distributed
control plane, it is difficult to get a complete view of entire network
topology, equipment and links status, etc. It has limited memory and
computation power, and it does not have a complete view of the entire
network topology. Therefore, the control plane functions can be very
limited and inefficient.

Software Defined Networking (SDN) is a new concept for the control

plane aiming at simplifying network operations, and managing the
network in a cost-effective and flexible way [44]. It separates the control
plane from the traditional network equipment, and uses a powerful,
programmable, logically centralized SDN controller to control the
network.
Network migration strategies for AON 27

Figure 3-7. Control plane migration towards SDN based network [Paper I]

Paper I proposes two options for the AON control plane migration, as
shown in Figure 3-7. The first option, Figure 3-7(a), uses separate SDN
controllers for each network domain and network service. A network
orchestrator on top of all controllers is responsible for a unified resource
coordination across different domains. The second option, Figure 3-7(b),
only uses a single centralized SDN controller to coordinate network
devices across different network domains. To increase the resiliency of
the centralized SDN controller, multiple redundant controllers can be
setup in different locations.

The first option may fit better than the second one for a large network
operator involving many network nodes, data plane technologies (e.g.,
PON in access, IP/Ethernet in metro/core network) and service
functionalities (e.g., broadband access for residential/business users,
mobile backhaul/fronthaul). The use of separated controllers has
advantages in terms of security, scalability, etc. on expense of efficient
network resource utilization and flexibility. For example, in the case of
cross domain network service, the domain controllers need to wait for
Network migration strategies for AON 28

the decision from the orchestrator coordinating the service provisioning,

because each domain controller is only aware of the details regarding its
own domain. Consequently, it may add delay and complexity of the
service provisioning.

The second control plane migration option may be appropriate for a

small network operator who has less diversity in respect to data plane
technology applied in the network. For example, AON is using Ethernet
as data plane technology, meanwhile Ethernet is also widely used in the
home, enterprise, aggregation and core networks. Therefore, a network
operator can use a logically centralized SDN controller to manage all
network devices in all segments. The use of single controller enables a
global view of all network devices and the entire network topology,
consequently, can provide quick and efficient resources allocation and
optimal routing path across the whole network.
 Techno-Economic analysis of AON migration 29

4 Techno-Economic analysis of AON

migration
Evolution of optical access networks promises to bring higher
bandwidth to more customers. However, it also introduces challenges
related to the selection of the right option and taking into account both
technical and economic aspects of the different solutions. Thus, different
next generation optical access (NGOA) solutions and corresponding
migration paths need to be properly compared in order to select the
most beneficial option for a certain scenario.

Techno-Economic (TE) analysis for NGOA networks is a complex task,

as it needs to involve a good knowledge from both technology and
economics. This chapter starts with introducing a general methodology
for performing TE study and TCO calculation in 4.1. Then the
methodology is elaborated with a focus on AON migration. Sections 4.2,
4.3 and 4.4 discuss predictions and use cases, technological scenarios
and cost modeling, respectively. The results are presented in 4.5.

This chapter is addressing the second and third contributions of the

thesis described in Section 1.3, which are based on a series of work
published in Paper II, Paper III, Paper IV, Paper V, Paper VI.

4.1 Methodology
TE evaluation is a combination of technological solutions and economic
analysis. It associates technological parameters with business and
economic parameters, and then translates them into costs and revenues.
To get a complete and reliable outcome, a good knowledge of both
technologies and economics is needed. The TE analysis can be used to
assess the economic feasibility of a technology, or to compare different
technological solutions in terms of the economic performance.

TCO is commonly used to present TE results. Paper II and Paper III of

the thesis has developed a general methodology for performing a
complete TCO estimation, as it shown in Figure 4-1. It is a funnel
process that comprises three major steps. Every step consists of several
building blocks that can be modified according to the focus of a specific
TE analysis.
Techno-Economic analysis of AON migration 30

Figure 4-1. A funnel process based techno-economic framework

In the first step, important preconditions and use cases should be

defined as a common ground for the TE study. They are mostly based
on the business and finance interests of the study, e.g. market shares,
business models, regulations, service demands and predictions. The top
layer of the Figure 4-1 gives an example of preconditions and cases that
are used to assess the TCO of communication networks. It contains 6
blocks. The time frame defines a time window of all scenarios and
models in the TE analysis. The impact of time frame on the TCO
depends on the customer penetration, which gives the market share of
customers every year. A telecommunication network normally consists
of several segments, e.g. home/in-building network, access network,
aggregation / metro network, and core network. Network scope defines
which segments are considered for a specific TE analysis. For example,
in TE study conducted in Paper VI, only access segment is included,
while in Paper II both access and aggregation/metro segments are
considered. The geographic model divides customers into several
groups based on their geographical locations. By doing that, the TE
study can estimate the geographical impacts on the economic
performance. The services block defines both the service types and the
quality of the service (e.g. bitrate, latency) that all different technological
 Techno-Economic analysis of AON migration 31

solutions have to provide. In the communication network value chain,

according to different business models, the costs and revenues can be
either owned by a single actor (e.g. traditional telecom incumbent) or
shared among several actors (e.g. service provide, network provider,
and infrastructure provider). The TCO results of the TE study are
calculated based on the input of the business model. Besides the
aforementioned 6 areas, other preconditions, e.g. regulatory limits and
return of investment, can be also added according to the TE assessment
objectives.

In the second step, technological solutions come into play. Proper

solutions should be designed and developed to match the preconditions
and use cases defined in the first step. A technological solution is broken
down to 3 categories: topology, data plane, and control plane. Every
category can have its own technological varieties. Chapter 3 has detailed
descriptions about these 3 categories. Step 2 can be used in a TE study
for assessing technological solutions that are completely different in all 3
categories. It can also compare options that only have differences within
a single category, in order to find out the impact of a specific parameter
on the TCO.

In the third step, preconditions and technological solutions are merged

together. TCO tools that consist of CAPEX models and OPEX models
are used to translate technological solutions developed in the second
step into costs and revenues. Details about CAPEX and OPEX models
are described in section 4.4.

Every block in Figure 4-1 can be the focus of one specific TE study. For
example, when the focus of the TE study is to compare the TCO
performance of different data plane technologies, the blocks in the
preconditions and use cases are kept unchanged as a common set of
inputs, and then only the “data plane model” block in the technological
scenarios vary from one case to the other. Similarly, when the focus of
the TE study is to investigate the impact of various business cases on a
certain technological solution, the blocks in the technological scenarios
are kept constant, while the business model block in Step 1 varies for
detailed investigation.

4.2 Preconditions and use cases

Starting from this section, we will go through the abovementioned
methodology with a detailed TE study case that focuses on the TCO
comparison of AON migration towards different NGOA solutions.
 Techno-Economic analysis of AON migration 32

4.2.1 Time frame

The time frame plays an important role in a network TE study. It
defines, for instance, when and how long a network is deployed, when
the deployment of the next generation network will start, and how long
it will take for migrating users from legacy network to the new one, etc.

For example, in Paper II, a time frame of 20 years is considered. In order

to find out the impact of migration starting year on the TCO, Paper II
also includes two scenarios that have different migration starting time.
In the first scenario, the legacy AON deployment is running from 1st
year to 9th year, the network migration process towards NGOA takes
place at the 10th year (referred to as the migration year), and the
migration process is assumed to take one year to complete, which is
realistic in certain areas [45]. In the second scenario, the migration
process starts at the 15th year. During the migration, the legacy network
and partly migrated NGOA are running simultaneously. After all the
migration processes finish, the legacy network is fully closed.

4.2.2 Business Models

Several business models for the optical access network are described in
[46][47]. Figure 4-2 shows an example [46]. Traditional
telecommunication services are based on a “vertical integration” model
(Figure 4-2 (a)), which means only one entity (actor) owns network
equipment and infrastructure and is responsible for all other things in
the network value chain, e.g. providing services, delivering, operating
the network. However, today’s broadband markets are transforming.
More actors come into play. New services and their providers, such as
Google [48], are expanding their business from service level to network
and physical infrastructure level. Other actors such as municipalities
have also started investing in the physical infrastructure of new telecom
networks, such as FTTH [46].

Actors can be categorized in 3 types according to their business roles in

different network layers. The Physical Infrastructure Provider (PIP)
owns and maintains the passive infrastructures such as ducts, fiber
cables, passive filters and optical distribution frames, etc. The Network
Provider (NP) is responsible for the active network equipment, such as
OLTs, RGs, amplifiers and cooling equipment. The service provider
delivers the digital services (e.g., Internet, video streaming, e-health,
cloud services, etc.).
 Techno-Economic analysis of AON migration 33

(a) (b) (c)

Content
S S S S S S
P P P P P P Application
Vertically
Integrated
network layers

Operator IP

N N N
NP OLT, switch
P P P
wavelength
SP
& Cabinet, PoP, ODF
NP
& Splitter, filter, closure
PIP PIP
PIP Fiber

Trech, duct

Figure 4-2. Examples of business model for the optical access network [46]

An open access network model is shown in Figure 4-2 (b) and (c). It
conceptually separates these 3 roles. In every business role category,
there can be one or multiple actors competing. In the Figure 4-2 (b), for
example, one PIP is running the fiber infrastructures. Several NPs can
rent a part of the fiber network to provide data delivery services to the
services layer. Service providers who are running the same or different
services are competing for end customers. However, those SPs can run
their business either on different NPs (Figure 4-2 (b)) or single NP
(Figure 4-2 (c)).

Service providers get revenues from end-users for their services, e.g.
Internet access, TV, cloud. Then SPs pay connection fee to the NPs for
the data delivery and network connectivity services, and NPs pay to the
PIP for the fiber access. The PIP can use the payments to build new fiber
infrastructure so that more end-users can be connected. Thus, an
ecosystem is formed.

4.2.3 Service Requirements

An optical access network can accommodate multiple services, such as
telephony, TV, video streaming, Internet access, IoT and mobile
backhaul/fronthaul. Different services have different requirements on
the network quality (also called Quality of Services, QoS), e.g. bitrate
and latency. In this section bitrate is modeled as an example of service
requirement.
 Techno-Economic analysis of AON migration 34

There are two measures for capacity, namely peak bitrate and
sustainable bitrate. The peak rate means the maximum bitrate a
customer may get from the network (e.g., during off-peak time when
other customers rarely use the network). The sustainable bitrate is the
guaranteed bitrate that is always reserved for every subscriber
whenever a customer connects to the network. The occurrence of peak
rate is uncertain. Many services like IP TV, Voice of IP, have to rely on
stable bitrate, therefore peak rate cannot be used for network planning
and design. In this thesis, sustainable bitrate is used as a common
baseline for assessing all the NGOA network planning and
dimensioning.

Figure 4-3 [49] shows three capacity evolution scenarios over 20 years.
In the final year of the evolution, the network should be able to offer
every customer a sustainable bitrate of 150 Mbit/s for minimum
scenario, 300 Mbit/s for medium scenario and 500 Mbit/s for maximum
scenario. The medium traffic evolution curve is used by Paper II, Paper
III, Paper IV, Paper V of the thesis for the TCO modeling.

Capacity Demand Evolution

Sustainable bitrate per customer (Mbit/s)

500
450 Min. scenario

400 Med. scenario

350 Max. scenario
300
250
200
150
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Year
Figure 4-3. Evolution of sustainable capacity over 20 years

4.2.4 Service penetration

In the cost assessment, it is important to define service penetration rate.
The penetration rate indicates the percentage of users who subscribe for
the network access in the area. Figure 4-4 shows three examples of the
penetration curves [50][51] over 20 years. The increase of number of
customers in scenario 1 is slower than in the other 2 scenarios, it reaches
63% in the 20th year. Scenario 3 presents a fast-growing broadband
 Techno-Economic analysis of AON migration 35

market, where the customer penetration aggressively hit 81%. Scenario 2

is used by Paper II, Paper III, Paper IV, and Paper V of the thesis,
where 74% of the users are connected to the network in the 20th year. In
the network migration study published in Paper II, two migration
starting years are investigated. One is the 10th year when customer
penetration is 10%, and the other is 15th year when the penetration is
about 40%.

Penetration Evolution
90%

80% Scenario 1
Scenario 2
Penetration percentage

70%
Scenario 3
60%

50%

40%

30%

20%

10%

0%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Year
Figure 4-4. Evolution of penetration curves over 20 years [51]

The penetration rate not only indicates the number of customers

expected to use the network, but also reflects the percentage of the
network/infrastructure that needs to be installed for accommodating
expected customers. Therefore, the penetration curve can be also used
for designing deployment scenarios and dimensioning the network
equipment etc.

4.2.5 Geographical model

The geographical model applied in Paper II, Paper III, Paper IV and
Paper V is based on the telecommunication network structure in
Germany [52]. In order to investigate the impact of Node Consolidation
(NC) on the TCO, the same geographical area is modeled in two
different scenarios: with and without NC (Non-NC). The Non-NC
scenario reflects the current situation of the legacy telecommunication
network. It consists of approximately 7500 service areas. One service
area corresponds to a traditional access node that can be either a Central
Office (CO) or Metro Access Node (MAN) as shown in Figure 3-1. In
the NC scenario, all COs are removed and only MANs remain as shown
in Figure 3-2. Therefore, the number of nodes in the NC case is reduced
from 7500 (in Non-NC case) to 1000.
 Techno-Economic analysis of AON migration 36

Furthermore, according to the population density, both Non-NC and

NC scenarios are divided in to 3 types of areas: Dense Urban (DU),
Urban (U) and Rural (R). The reference areas are characterized by a
number of households and area size, as shown in TABLE I.

TABLE I: PARAMETERS FOR THE AREA TYPES [Paper II]

Number of users per Area size (km2) per Density
Node Consolidation Type Area Type
node node (users per km2)
Non-NC DU 15600 5.00 3120
Non-NC U 8640 24.00 360
Non-NC R 3060 56.67 54
NC DU 44500 14.26 3120
NC U 51000 141.67 360
NC R 33000 611.11 54

4.2.6 Network scope

The network scope clearly defines which parts of the network are
included in the TE analysis. In Paper II, Paper III, Paper IV and Paper
V of the thesis two network segments: access and aggregation are
considered. The rest of networks (e.g. home network, core network) are
excluded. In Paper VI, only access segment is included for the TE
analysis.

Figure 4-5. Network model: access and aggregation [Paper II]

The definition of access and aggregation network in NC scenario is

different from the one in Non-NC scenario as shown in Figure 4-5.
Depending on the geographical location of the OLT placement, the
access network in a Non-NC scenario is defined from the end-point to
the CO, and the aggregation network is from the CO to the core Point of
Presence (PoP). However, in a NC scenario, the CO is assumed to be
closed, OLT is placed in the MAN, therefore, the access segment is
defined from end-point to MAN, and the aggregation is from the MAN
to the core PoP. In order to have a fair comparison between NC and
 Techno-Economic analysis of AON migration 37

Non-NC scenarios, a TE study should consider both access and

aggregation segments, i.e., from the end-point and the core PoP.

4.3 Technological scenarios

In the previous section, the preconditions and study cases of TE analysis
have been defined. This section illustrates how technological solutions
can be developed to fit those preconditions considering examples of
AON migration towards different NGOA solutions.

The technological solution can be divided into 3 parts: topology, control

plane and data plane. There can be multiple variants of each category.
Therefore, the different combinations from the 3 categories can lead to a
large number of technological solutions.

Chapter 3 has depicted the migration strategies of AON in detail with a

few variants of every category. However, in this TE analysis example,
the main focus is set on comparing the TE performance of NGOA data
plane solutions. Hereby, a number of data plane variants are modeled in
this section, while topology and control plane models are simply
modeled with one scenario respectively.

The topology model applied in Paper II, Paper IV and Paper Vis based
on star / tree topology with node consolidation evolution. The model for
control plane adopts the traditional option where the control plane
functions are placed locally at different nodes. However, those
distributed nodes can be remotely configured by centralized C&M
system. According to the network scope defined in 4.2.6, the thesis also
assumes that each access network segment and metro / aggregation
segment have their own C&M system to control the network equipment
in the own domain.

For the data plane model, this thesis investigates five NGOA scenarios
referred to as Point-to-Point WDM-PON, TWDM-PON, WDM-backhaul,
and two Gigabit Ethernet (GE) network upgrade scenarios. Every
NGOA scenario can offer 300Mbit/s sustainable capacity per user, and at
least 1Gbit/s peak rate.

The two considered starting architectures are AON PtP and AS, as
illustrated in Figure 2-1(a) and (b). Both are in Non-NC scenario and
equipped with Fast Ethernet (FE) in the first mile, giving a maximal
bitrate of 100Mbit/s.
 Techno-Economic analysis of AON migration 38

TABLE II summarizes the studied migration paths in Paper II. In order

to investigate the NC impact on AON migration, the five targeted
NGOA scenarios are divided into two groups. The NC migration group
contains three migration paths (MG#1,MG#2 and MG#3), while the Non-
NC migration group includes two paths (MG#4 and MG#5). The three
scenarios have been introduced in Chapter 3. However, this section
provides detailed technical information that is directly related to the
network planning and dimensioning.

TABLE II STUDIED MIGRATION PATHS (MG)

Starting Architecture Target Architecture Topology
MG#1 AON PtP (FE) WDM-PON (80ch) Non-NC NC
MG#2 AON AS (FE) TWDM-PON (1:32, 40ch) Non-NC NC
MG#3 AON AS (FE) WDM-backhaul (40ch.) Non-NC NC
MG#4 AON PtP (FE) AON PtP (GE) Non-NC Non-NC
MG#5 AON AS (FE) AON AS (GE) Non-NC Non-NC
*FE (Fast Ethernet), GE (Gigabit Ethernet)

4.4 Cost modeling

This section shows an example of cost modeling for the optical access
network. It translates technological scenarios to costs. The cost
assessment of the network is based on the TCO, which consists of both
CAPEX and OPEX.

4.4.1 CAPEX
CAPEX can be categorized into three major parts: infrastructure,
network equipment and home equipment as described below:

1. Infrastructure contains access infrastructure, and in-house

infrastructure. The access infrastructure cost includes fiber cables, ducts,
trenching, fiber splicing and fusion, optical distribution frames (ODFs),
and passive components in the access network such as optical branching
boxes, cabinets, power splitters, wavelength filters and related
installations for all those components. The in-house infrastructure cost
consists of in-house cabling, optical sockets and installation required at
the customer premises.

2. Network Equipment refers to the active equipment located in the

access network segment. It may include the Ethernet switches, OLTs,
backplane switch fabric, cooling equipment, amplifiers, Uninterrupted
Power Supply (UPS), etc.
 Techno-Economic analysis of AON migration 39

3. Home Equipment mainly refers to the cost of RG. A typical RG

includes optical transceiver, Local Area Network (LAN) function, four
Gigabit Ethernet interfaces, one plain old telephone service interface
and one radio frequency interface.

4.4.2 OPEX
OPEX assessment considers several cost driving processes such as
Service Provisioning (SP), Fault Management (FM), energy
consumption, maintenance, floor space, etc.

1. Energy includes energy consumed by any network equipment, the

cooling devices, UPS and RG. Some items that are belonging to the
infrastructure can also consume energy, e.g. active outdoor cabinet.
TABLE III shows an example of electricity prices used in Paper II,
where the price differs depending on the year and location (indoor
versus outdoor, or urban versus rural). Business roles may also have
impact on the electricity price, for example, the price can differ between
residential users and business users.

TABLE III PRICE OF ENERGY OVER 10 YEARS FOR DIFFERENT LOCATIONS [53]
Location type Y10 Y11 Y12 Y13 Y14 Y15 Y16 Y17 Y18 Y19 Y20
Indoor 73 75 77 79 82 84 87 89 92 95 98
Outdoor 86 89 91 94 97 100 103 106 109 112 116
Residential 52 53 55 57 58 60 62 64 66 68 70
*unit [CU(cost unit) per kWyear], one CU is corresponding to the cost of a standard GPON ONT

2. Service provisioning (SP) is the cost associated with any activities

related to adding, changing and cancelling the customer services. It is
related to many factors such as fiber management (e.g. patching,
splicing), remote configuration, human resources, travelling, etc. The SP
model is developed with the Business Process Modelling and Notation
tool (BPMN)[54]. An example of SP process is shown Figure 4-6 [55].

The BPMN tool uses flowchart to present the business workflow and
processes. There are three types of block in Figure 4-6. The circular
blocks denote the starting or ending points of the process. The
rectangular blocks indicate process steps or activities. The diamond
blocks represent “Yes/No” questions, where a decision is required.
There are 3 major SP processes in Figure 4-6: add, change, and cancel
services. For adding new service process, two activities are triggered:
the first one is related to the physical connectivity (data plane), and the
second one is related to the configuration of the customer service profile
and service path through the network (C&M plane). Technicians need
Techno-Economic analysis of AON migration 40

first check whether there is a physical connection or not. If the answer is

no, technicians must travel to different locations and manually set up
the physical connections. Therefore, the network topology (number of
nodes, location of nodes, etc.) also has impact on the SP costs. Once the
connection is ready the network equipment has to be configured, and
new services should be added.

Figure 4-6. An example of service provisioning process in flowchart diagram [55]

As for service cancellation, the processes are modelled according to the

reasons of requests, e.g., network operator/provider change or other
reasons. In case of operator/provider change request (a customer
switching from operator A to operator B), the current operator/provider
has to check whether there is a bit-stream access for the new network
operator/provider. If so, only Control and Management (C&M) plane
activity is involved, the service can be re-configured through the
Network Management System (NMS). Otherwise, both data plane and
C&M plane activities have to be included. E.g. manually disconnect
physical connections, reconfigure network equipment,
decommissioning services, etc.

The service change processes can be divided into two types: location
change or service change (e.g. capacity upgrade). In case of location
changing, the process is similar to adding new services. In case of
service change, only C&M activity is involved, the requests can be done
remotely though NMS.
 Techno-Economic analysis of AON migration 41

3. Fault Management (FM) is the cost associated with the failure

monitoring, detection, troubleshooting, component replacement and
reparation.

The work presented in [55] also introduces an overall failure

management process as show in Figure 4-7. It contains 3 major steps.
When the FM process is triggered by either customer calls or
system/network alarms, firstly the faults are diagnosed (or try to be
resolved) in C&M plane, e.g. NMS. The second step is the physical
reparation (data plane). Typically technicians are going to the location of
the failure and repair it. Finally, the process is closed with some
additional C&M plane administrative steps.
PTI/Field (NP)
technicians

Fault Local
Diagnosis reparation line test
report
+ +
technicians

Config.
NMS/IT

Diagnosis Config.
report failure successful
solving [1] test ? NO
YES
physical YES
technical service (TS)

YES NO Technical Service

Trouble failure
Helpline (CS) and

solved ?
NO
Helpline
? Create 2nd level
Customer call TT YES Technical Service
(diagnosis) Close
Helpline Failure
Extended TT
1st level identified failure
? NO diagnosis [3]
Customer Service FMDB [2]
NO
Management

Fault alarm Problem

Network

Fault Fault solved

Create ? YES
detection diagnosis
TT
by NMS and isolation
NMS/IT technicians
TT: Trouble Ticket ; CS: Customer Service ; TS: Technical Service ; NP: Network Production [1] Mostly from a central NMS/IT
Fault detection is the process of receiving alarms and knowing that there is a network problem that should be identified. location without traveling
Fault diagnosis is the process of identifying a set of candidates to be faulty and have caused the received alarms. [2] FMDB: Failure Message Data Base
Fault isolation is the identification from the set of candidates of the faulty element after some remote tests. [3] e.g. outside failure localisation by
technical service

Figure 4-7. An example of fault management model [55]

In order to calculate the cost of SP and FM processes, every gateway in

the BPMN model is associated with a statistical occurrence, and every
activity is associated with an execution cost. The statistical occurrence
indicates the probability of a decision that could happen. The execution
cost is calculated from the amount of resources on average consumed in
the activity multiplied by the unit cost of the resources. Different
network topologies, data plane, and C&M plane models have a large
impact on the costs. For example, advanced C&M solutions can simplify
and automate the SP and FM processes; node consolidation topology
has fewer nodes in the field and therefore leads to lower travel and
human resource costs; data plane components may have different
statistical occurrence of the failure.
 Techno-Economic analysis of AON migration 42

4. Maintenance includes the maintenance of both network equipment

and physical infrastructure. Paper IV describes the cost model for
maintenance, where 3 cost factors are considered: personnel cost,
hardware maintenance cost, and system software maintenance.

5. Floor space refers to the cost that is required for the placement of
OLTs, racks, ODFs, cooling equipment, UPS, etc. One example of the
floor space cost can be the real estate rent in CO and MAN. The floor
space is related to the footprint and quantity of the network equipment.
For example the PON takes less floor space than AON PtP in CO
because the number of required OLTs in PON is less than AON PtP. The
number of fibers connected in the CO or MAN also has impact on the
floor space. The more connected fibers, the more required space for fiber
management tools, e.g. ODFs. The cost of floor space can be related to
the geographical locations and time. For example, the rental price in a
city area may be more expensive than in a rural area, and it may
increase gradually over the years.

Figure 4-8. TCO breakdown, cost category view [Paper II]

In Paper II of the thesis, six major parts of CAPEX and OPEX are
considered in the TE analysis as shown in Figure 4-8. Maintenance and
floor space aspects are covered in the Paper IV. Paper VI only focuses
on the CAPEX part of the network cost.

4.4.3 CAPEX and OPEX division by business role

In Paper II of the thesis, all six categories of CAPEX and OPEX are
defined in relation to the different business roles: PIP, NP and User, as
shown in Figure 4-9.
 Techno-Economic analysis of AON migration 43

Figure 4-9. TCO breakdown by business role [Paper II]

PIP CAPEX refers to the material and deployment cost associated to

access infrastructure that belongs to PIP. PIP OPEX includes FM and
energy consumption of PIP owned infrastructure. PIP FM involves
events such as reparation of fiber cut, AWG or power splitter failures.
Energy cost of PIP is associated with the energy bills related to the
infrastructure, such as lighting, cooling and maintenance of the cabinet.

NP CAPEX includes the expenses of both network equipment in CO,

MAN, RN and RGs at customer premises owned by NP. NP OPEX
consists of the costs for FM of network equipment and RG, energy
consumption of network equipment, and SP.

User CAPEX considered in this thesis only includes to the in-house

infrastructure, because we assume that the NP owns the RG even if it is
located in the user’s premises. However, there can be alternative
business model, where customer owns and pays for the RG. Therefore,
in that case, RG can be considered as user CAPEX. The in-house
Infrastructure is usually paid by users, construction companies, real
estate companies or house/building management companies. User
OPEX refers to the energy bill related to RG, which is normally paid by
the customer no matter who owns RG.

4.5 Techno-Economic Analysis Results

This section summarizes the results from TE studies conducted in Paper
II, Paper IV, Paper V. The cost values presented in this thesis are
normalized to the cost of a GPON ONT, referred to as one Cost Unit
(CU). The TCO results of Non-NC scenarios have been normalized to
the same service area as NC cases.
 Techno-Economic analysis of AON migration 44

4.5.1 Yearly TCO over 20 years

In Paper II, TCO for five different AON migration paths towards
NGOA are compared and analyzed, as shown in Figure 4-10. The TCO
per year shows the investment evolution over the entire operation time
of the network, which is considered to be 20 years. The network
migration is assumed to start in the 10th year. The number of connected
users increases yearly and follows the second scenario of penetration
curve in Figure 4-4. The TCO calculation is based on one MAN node
service area.

Figure 4-10. Yearly TCO for a dense urban area in access network, when the
migration starts in the 10th year. [Paper II]

The MG#1 and MG#2 paths have significantly higher costs than the
other options. They are both in NC scenario and involve introducing the
new WDM/TWDM technology during the migration. The remaining
migration paths (MG#3, MG#4, MG#5) are at the similar cost level. The
migration paths from FE to GE in Non-NC scenario (MG#4, MG#5) have
lower cost because there is no new technology introduced during the
migration, and the network topology remains the same as the legacy
AON. However, both GE PtP and GE AS have obviously higher costs
after the migration year, i.e., between the 11th and the 20th year. This
will gradually reduce their investment savings gained in the beginning
of the migration.

In Non-NC scenario (MG#4 and MG#5) the cost of feeder fiber (from CO
to MAN) is excluded from Figure 4-10, because it is modeled as a part of
the aggregation network, which is addressed in Section 4.5.3.
 Techno-Economic analysis of AON migration 45

4.5.2 TCO breakdown in the migration year

TCO break down (migration in 10th year)

120
SP

100 Energy

FM
80 Network Equip.
Thousands CU

RG
60
Infrastructure

40

20

0
AON PtP-> AON AS -> AON AS -> AON PtP AON AS
WDM-PON TWDM-PON WDM-backhaul FE->GE FE -> GE
(NC) MG#1 (NC) MG#2 (NC) MG#3 (Non-NC) MG#4 (Non-NC) MG#5

Figure 4-11. TCO breakdown for a dense urban area in the 10th year [Paper II].

Figure 4-11 zooms in for a close-up of the TCO in the 10th year. The TCO
includes six categories as described in Figure 4-8. The MG#1 has the
highest migration costs mainly due to the investment on new network
equipment (55% of TCO). The MG#2 also introduces modern TWDM
technology, however the costs of the network equipment only accounts
for 19% of TCO. The WDM-PON with 80 wavelength channels can only
support up to 80 users while the TWDM-PON can increase this number
to 1280 (32 TDM slots * 40 wavelength channels). Although the WDM-
PON OLT is less complex and costly than the TWDM-PON OLT, the
amount of required WDM-PON OLTs is much larger than the TWDM-
PON OLTs, which leads to the highest network equipment cost and
energy consumption among all migration paths. The dominant cost for
MG#2 migration path is the infrastructure (38% of TCO). New
infrastructure such as power splitters, AWGs have to be deployed in
both RNs and COs, which involves massive installations. In the OPEX
part, SP is a major factor for all migration paths. The SP involves
technician travel, manual cable patching, reconfiguration, etc. In a
migration scenario, customers need to be switched from the legacy
network to the new network. The number of SP events is proportional to
the number of users, therefore, it leads to a high SP cost.
 Techno-Economic analysis of AON migration 46

TCO break down (migration in 10th year)

120
User OPEX User CAPEX
100
PIP OPEX PIP CAPEX

80
Thousands CU

NP OPEX NP CAPEX

60

40

20

0
AON PtP-> AON AS -> AON AS -> AON PtP AON AS
WDM-PON TWDM-PON WDM-backhaul FE->GE FE -> GE
(NC) MG#1 (NC) MG#2 (NC) MG#3 (Non-NC) MG#4 (Non-NC) MG#5
Figure 4-12. TCO breakdown by business roles for a dense urban area in the 10th
year [Paper II]

Paper II of the thesis also analyzes the TCO in the 10th year according to
the different business roles, as shown in Figure 4-12. The NP is a
dominant player in the network migration. The NP investment
attributes to more than 80% of overall TCO for most of the migration
paths, except for the MG#2 where the NP accounts for 60% and PIP
accounts for 35%. Although PIP investment is the largest part of TCO in
the initial greenfield deployment [56][57], the migration paths proposed
in Paper II can maximize the reuse of legacy infrastructure and avoid
the unnecessary reinvestment on the PIP infrastructure.

4.5.3 Node consolidation analysis

In this section, we focus on the impact of NC on the TCO results. Paper
II compares NC scenarios to Non-NC scenarios with the consideration
of the full network scope described in Figure 4-5. The aggregation
network cost was provided by Deutsche Telekom [53][58]. The
aggregation network cost of the NC and Non-NC scenarios are
different. For Non-NC scenarios, the aggregation network cost consists
of two parts: aggregation I and aggregation II as shown in Figure 4-5.
For NC scenarios, only aggregation I is presented because the
aggregation II segment is merged into NC access network.
Techno-Economic analysis of AON migration 47

2.5
Delta Aggregation
2.0 Delta Access

cost losses/ savings [CU/ user] 1.5

1.0

0.5

0.0

-0.5
AON PtP --> WDM-PON

AON AS --> WDM-backhual

AON AS --> TWDM-PON

AON PtP --> WDM-PON

AON AS --> WDM-backhual

AON AS --> TWDM-PON

AON PtP --> WDM-PON

AON AS --> WDM-backhual

AON AS --> TWDM-PON

-1.0

DU U R
Figure 4-13. Cost difference NC versus Non-NC in access and aggregation network
[Paper II]

Figure 4-13 depicts the cost difference between the NC and Non-NC
scenarios. Three geographical area types, Dense Urban (DU), Urban (U)
and Rural (R) are investigated. The cost difference in the access network
segment is presented the as blue bar in the figure, and the cost
difference in aggregation network segment is shown as red bar. For the
access segment (blue bars), the values are negative. It means that the NC
approach requires more investment in the access segment than the Non-
NC approach, and hence it does not bring cost benefits. For the
aggregation network segment (red bars), the values are positive. It
indicates that the NC scenarios can bring the obvious cost savings
compared to Non-NC scenarios in all types of areas. The cost saving
become significantly higher in rural area due to the low sharing factor.
By summing up the cost losses from the access segment and gains from
the aggregation segment, it can be concluded that NC is beneficial for
the AON network migration, especially in the rural areas.
 Converged Optical Transport Network 48

5 Converged Optical Transport

Network
With the fast-increasing data traffic demand in both fixed and mobile
networks, both networks are facing the need of the network upgrade.
The network migration requires enormous investment, while the
Average Revenue Per User (ARPU) is almost flat or even declining [24].
Therefore, upgrading cost is one of the most important aspects for
network operators/providers when they are considering network
evolution.

OTN has been already used in metro and core network for many years.
It is widely considered as a future proof technology that can provide
reliable, scalable, and ultra-high capacity transport services. The need
for evolution in both fixed and mobile networks brings an opportunity
for network operators to consider a converged OTN solution that covers
core, metro and access segments. The unified OTN can support a large
variety of services coming from both fixed and mobile customers. Thus,
it not only saves the CAPEX but also reduces the OPEX compared to
running fixed and mobile network in parallel.

Additionally, the innovative technologies, such as SDN, enable another

way to obtain cost-saving. Traditional network designs lead to complex
hardware stacks, over-provisioning of capacity, difficulty to re-
configure and high cost. The concept of SDN allows service provider to
reduce OPEX by simplifying network operations and by automated
network provisioning.

This chapter introduces a programmable OTN architecture with SDN

features. The OTN can accommodate converged services from both
fixed and mobile networks as well as rapid re-provisioning for
dynamically changed service requirements.

This chapter describes the fourth contribution of the thesis, which has
been published in Paper VII.
 Converged Optical Transport Network 49

5.1 Converged network services

A converged OTN should be able to serve both fixed and mobile
networks. Several NGOA architectures have been discussed in Chapter
3 with a focus on fixed network services, but the unified OTN
architecture should also include backhaul/fronthaul services for the
mobile networks.

5.1.1 Mobile backhaul/fronthaul services

RAN provides connectivity between the core network and end-users, it
consists of multiple radio access nodes/cells. The connection between
cells and end-users is wireless (through air), while the connection
between cells and mobile core is provided by backhaul/fronthaul
network, which can be either wired or wireless.

There are many different types of cells, in general they can be divided
into two major categories: macro cells and small cells. A macro cell has
higher radio power and covers a larger service area than a small cell. A
typical macro cell radius is a few kilometers (maximal coverage up to 35
km) [59]. The small cells can be divided in three categories, i.e.,
microcells, picocells and femtocells [60].The small cells are characterized
by lower power and smaller coverage than the macro cells and typically
have a range from 10 meters to several hundred meters. A RAN
consisting of different types of access nodes is called as heterogeneous
network (HetNet). In a HetNet RAN, the macro cells mainly focus on
providing coverage, i.e., connectivity to cover mobile users in a large
service area, whereas the small cells aim for improving cellular signal
quality and capacity where needed. The macro cells usually are
deployed outdoor, either on a telecommunication tower or on the top of
the buildings. The Small cells are normally used for in-building
environment, enterprises, campus, venues, stadiums as well as public
spaces.

Figure 5-1 (a) shows an example of a traditional RAN architecture. A

macro cell and its Base Band Unit (BBU) are co-located in the field
nearby the Central Office (CO), and provide the wireless broadband and
voice services to all subscribers within its coverage. A small cell and its
BBU is deployed closer to the end-user, e.g. in-building, in order to
provide enhanced wireless capacity and Quality of Services (QoS). The
links between the BBU and the mobile core network are called mobile
backhaul.
Converged Optical Transport Network 50

Figure 5-1: RAN architectures in a HetNet deployment

A new RAN architecture, called Centralized-RAN or Cloud-RAN (C-

RAN), has been raised up by China Mobile in 2011 [61]. It has attracted
tremendous interest from industry due to the advantages of, e.g. energy
and operational efficiency, cost-saving, high spectrum efficiency, and
users’ capacity improvement. An example of C-RAN is shown in Figure
5-1(b). Unlike the traditional RAN, in C-RAN BBUs are not co-located
with the cell sites anymore. The BBUs are moved to a centralized
location, e.g. CO or Metro Access Node (MAN). But the antennas and
Radio Frequency (RF) components (also referred to as Remote Radio
Head, RRH) are still distributed in the field. This centralized BBU
location is called BBU hotel or pool, where many BBUs are co-located
together. With Network Function Virtualization (NFV) technology, BBU
can be even virtualized in a cloud.

The connections between BBU and mobile core network are the same as
the traditional RAN, which can be an IP based backhaul link. On the
other hand, the connections between BBU and RRH are called fronthaul,
which need a special transport protocol. The current C-RAN uses
Common Public Radio Interface (CPRI) [62] as a fronthaul interface,
which has stringent requirements on the link capacity and latency. Next
Generation Fronthaul Interface (NGFI) [63], e.g. Radio over Ethernet
(RoE) [64], evolved CPRI [65], is under investigation by different
standardization groups. The NGFI solutions are based on different BBU
functional split options and can become more flexible and cost-effective
than the CPRI.
 Converged Optical Transport Network 51

Whereas mobile network is working on 5G technologies and

architectures, the backhaul/fronthaul network for the RAN requires also
evolution. New technologies such as massive Multiple Inputs Multiple
Outputs (MIMO) [66], Coordinated multipoint transmission (CoMP)
[67], not only significantly improve the network performance and QoE,
but also makes the 5G RAN superior to the traditional RAN in terms of
cost-saving, energy consumption, resource utilization, interference
control, etc. In order to implement those 5G technologies and achieve
their benefits, a high quality (e.g. high capacity, low latency)
backhaul/fronthaul network is essential.

5.2 OTN architecture

5.2.1 Data Plane
OTN data plane must support for higher capacity, increased number of
transport clients, support a wider range of performance requirements.
Several converged OTN scenarios have been proposed, e.g., in [28][68]
[69]. NG-PON2 is standardized with the consideration of matching both
needs from fixed access and mobile backhaul/fronthaul, making it
possible to have one network deployment for both fixed and mobile
services.

Figure 5-2: Data plane structure of the converged OTN

Figure 5-2 shows an example of converged OTN solution from [69]. The
convergence is not only related to different services from RAN, fixed
residential and business access network, but also across different
network segments from the access to the core network. On the right side
of the Figure 5-2, the OTN is connected to multiple services e.g. BBU
pool for the mobile fronthaul, Edge router for fixed access, etc. DWDM
 Converged Optical Transport Network 52

is the main technology used in the converged OTN solution. It consists

of DWDM switches, reconfigurable optical add/drop multiplexers
(ROADM) and tunable transponders (TP). DWDM switch is used to
aggregate wavelengths from both fixed and mobile access network so
that wavelength resources can be flexibly allocated. ROADMs are used
to add/drop wavelengths to different access points, and then the TP
converts the wavelength used in the DWDM system to the wavelength
used in the local services.

5.2.2 Control plane

The main advantage of the converged OTN is its ability to accommodate
multiple services. However, for different types of services, e.g.,
residential access, enterprise access, mobile backhaul/fronthaul, the
service demand may differ significantly. Even for the users within the
same type of service (e.g. residential access) the demand can vary
largely between the users and may also depend on the time period.
Therefore, it is important that the converged OTN can be flexibly and
timely adjusted to adapt to varying service conditions.

Different services have their own requirements related to network QoS

parameters, e.g. capacity, latency. The changes of network parameters
usually need the reconfiguration of network resources. In a traditional
network environment where the control plane is distributed, it is a
difficult task to reconfigure the network and provision services. It
involves many operational efforts, and usually takes long time.

The concept of SDN has been introduced in Chapter 3. In an SDN-

controlled network, a centralized controller controls and manages
network equipment and infrastructure via an application-programming
interface (API) at the southbound of the controller, as shown in Figure 5-
3. Then, the controller can expose the network resource information to
higher layer applications through the northbound API. In contrast to the
traditional control plane, SDN approach allows network applications to
automatically optimize network resources across heterogeneous
domains and quickly instantiate/change end-to-end services.

Figure 5-3 shows an example of overall OTN architecture proposed in

Paper VII. The control plane of the OTN is implemented with a
centralized SDN controller, which controls and manages the operation
of OTN data plane. In Paper VII, the controller is implemented with
open-source software, Open-Day-Light (ODL) [71] as the basis, and
extends it with several functions to optimize it for control of large-scale
Converged Optical Transport Network 53

DWDM networks. The controller has the detailed information about all
OTN network resources (e.g. equipment details, network topology,
available wavelength resources, link capacities). According to the
requirement of the application, the optical path computation element
(PCE) inside the controller can find an optimum path that can provide
the best performance for the application. Then, the corresponding nodes
along the path will be configured, network resources will be allocated,
and finally an optical transport link will be established from one OTN
access point to another (e.g. connecting BBU hotel/pool with mobile
access network in Figure 5-3).

Figure 5-3: OTN architecture: control plane and data plane

The SDN also allows for creating one or more layers of

abstraction/virtualization on top of the controller. The process of
exposing resources to the orchestrator/or network application is referred
to as network abstraction/virtualization. The purpose of network
abstraction is to hide unnecessary complex details from upper
applications so that the applications can use OTN more efficiently.
There are different ways to abstract network resources. In [70] three
abstraction models are presented and the tradeoffs between complexity
and performance are compared. In this thesis, a “big switch” abstraction
model is presented as shown in Figure 5-3. The OTN consists of many
network nodes and different types of equipment. A layer of abstraction
on top of the SDN controller can hide all the details of that network and
present the whole network as one single node (switch). In the abstracted
“big switch” only the access points and service edges are exposed to the
 Converged Optical Transport Network 54

upper application/orchestrator. This improves the modularity and

scalability of the control mechanism.

5.3 Network Orchestration

To demonstrate and evaluate benefits of the converged multi-purpose
OTN, Paper VII of the thesis implemented two use cases that make
smart use of the multi-service testbed..

5.3.1 Use case 1: Elastic Mobile Broadband Services

The first case is an Elastic Mobile Broadband Service (EMBS). OTN
provides mobile fronthaul services to the EMBS. The service capacity
can be dynamically and automatically scaled up and down when and
where needed.

Figure 5-4 illustrates the EMBS in a C-RAN HetNet deployment case.

There are two disjoint RAN service areas: Area 1 and Area 2. Each area
is equipped with two Remote Radio Units (RRU): a macro cell and a
small cell. While the macro cell provides the coverage across the area,
the small cell is used for providing additional capacity in the area if
needed. The total of 4 RRUs (across two areas) are served by a
centralized BBU hotel. In a traditional deployment, every RRU requires
a BBU resource. In the EMBS case, only 3 BBUs are used due to the
resource sharing between two small cells. When extra capacity demand
is identified in an area (e.g. more users in one area than the other), the
corresponding small cell is activated and connected to the BBU.

A practical use case of the EMBS can be: Area 1 is a residential area and
Area 2 is an office area. Two macro cells are always on and provide
continuous coverage in each area. During the working hours, e.g. 8:00 –
17:00, there are more users in the office area than in the residential area,
the small cells in the residential area can be reduced, and the
corresponding BBU resources and transport link resources can be
assigned to the small cells in the office area. On the other hand, during
the evening or weekend, more users are supposed to be in the
residential area than the in the office area. Thus, the BBU capacity and
transport resources can be shifted to the residential area accordingly.
 Converged Optical Transport Network 55

Figure 5-4: Elastic Mobile Broadband Service (EMBS) over OTN

5.3.2 Use case 2: Distributed synchronization services

With the fast development of hardware virtualization, cloud computing
and storage is becoming more and more popular [71][72]. Nowadays,
cloud services such as Dropbox [73], Google Drive [74], Microsoft
SharePoint [75] are available for both private users and enterprises, and
are already commonly used by people every day. Those cloud services
synchronize the local user data to/from remote servers in a data center,
and keep the same files up-to-date on all distributed devices.

Figure 5-5 shows a distributed synchronization service (DSS) in an OTN

use case. Individual user’s data is first synchronized to / from a local
cloud storage server which provides fast cloud sharing services inside
the enterprise network. Then, the local server is synchronized to / from
remote servers through the OTN. The remote sever can be located either
in a data center or another location of an enterprise campus.
 Converged Optical Transport Network 56

Figure 5-5: Distributed synchronization service over OTN

5.3.3 Multi-domain orchestration architecture

To accommodate EMBS and DSS on the same OTN, these two services
need to be orchestrated and the network infrastructure need to be
coordinated. Paper VII introduced a multi-domain orchestration
architecture based on SDN approach, which supports dynamic service
creation across a heterogeneous set of resources in a resource-optimized
manner.

Figure 5-6 shows a hierarchical view of converged multi-domain

orchestration architecture. It is composed of three domains with
different types of resources: an OTN based on DWDM-centric transport
network, a mobile broadband network based on LTE, and a cloud
domain. The DWDM based OTN has a central role in the architecture, it
provides flexible connectivity services for the other two domains.

There are three separate domain controllers at the bottommost layer.

Each controller is responsible for the control and management of the
network resources in its own domain. The orchestrator at the middle
layer of the hierarchy is responsible for a unified resource coordination
across the three domains. The orchestrator creates a global view of
network resources/capabilities from all three domains and exposes it to
higher layer service (e.g. EMBS and DSS). Different services have their
own requirements related to the network infrastructure. The
orchestrator translates these requests to the available resources and
configures the data plane accordingly through the corresponding
controllers. This global orchestration enables agile and efficient resource
allocation across the three domains.
Converged Optical Transport Network 57

Figure 5-6: Hierarchical architecture of the multi-domain multi-service testbed

[Paper VII]

The implementation of OTN controller has been described in Section

5.2.2. For the other two domains, domain-specific controllers for the
mobile and cloud domain have been proposed in Paper VII. The RAN
controller is responsible for mobile cell activation and configuration, the
assignment of BBU resources to active RRUs (i.e., cells) as well as
management of users’ handovers among the cells. For the cloud
domain, the local and remote servers are controlled by the cloud
controller. The controller functions include detecting connectivity
between severs, data replication, etc.

The EMBS application continuously monitors the capacity demand in

the RAN, and decides when and which cell needs the BBU resource and
the OTN connectivity resource between the cell and the BBU. EMBS
sends requests to the orchestrator. Since the orchestrator has the global
view of the available network resources across all domains, it is able to
allocate appropriate BBU resources through the RAN controller and
connectivity resources through the OTN controller. Finally, controllers
configure the data plane equipment and setup the services.

The DSS application tracks the changes in the enterprise/data center

network. Depending on a certain level of data changes, the DSS
application sends data replication requests to the orchestrator. The
 Converged Optical Transport Network 58

orchestrator then checks the availability of OTN resources, and asks the
OTN controller to establish the transport connectivity between the
enterprise network and the remote data center network. When the
connectivity is ready, the orchestrator sends commitment request to the
cloud controller to complete the data replication process.

5.3.4 Orchestration of converged services

To demonstrate the benefit of converged OTN and make an efficient
usage of OTN resources, in Paper VII of the thesis a multi-service
orchestration allowing the EMBS and DSS to operate on the shared OTN
infrastructure has been proposed.

In a traditional OTN, to operate both EMBS and DSS requires 5

wavelength channels (lightpaths). With help of SDN, the converged
OTN only uses 3 lightpaths, where one lightpath is shared by 2 small
cells in EMBS and the synch process of DSS. A coordination policy has
been developed in Paper VII to resolve the possible conflicts on the
shared link between EMBS and DSS. The policy prioritizes the EMBS.
When there is a competition for the OTN resources, the EMBS always
has the priority to get connectivity. A state diagram in Figure 5-7 shows
the algorithm of policy implemented in the orchestrator. It consists of 4
states. States 1-3 is related to the EMBS and State 4 is related to the DSS.
State 1 represents the default mode of the EMBS, where only two macro
cells are provided connectivity by OTN. In this State, EMBS has low
capacity demand, no additional small cell and OTN resources are
needed. States 2 and 3 represent cases where a high capacity demand is
observed in Area 1 or 2. Besides two active macro cells, an additional
OTN link is required to activate the small cell in the corresponding area.
State 4 refers to the case where OTN resources are provided to two
macro cells and DSS. State 4 occurs only when the capacity demand in
EMBS is low, i.e., none of the small cells is using the OTN.
Converged Optical Transport Network 59

Figure 5-7: State diagram of the orchestration process: texts on arrows show
conditions for the transition to a new stage [Paper VII]
 Conclusions and future works 60

6 Conclusions and future works

This thesis investigated network migration towards next generation
optical access and transport solutions. The study provides the following
major outcomes:

• Paper I of the thesis suggested several AON migration strategies

from data plane, topology, and control plane perspective, and
provided qualitative evaluation of these strategies with respect to
both CAPEX and OPEX.

• Paper II and Paper III of the thesis developed a general

methodology for performing Techno-Economic analysis of the
fiber access network solutions.

• Paper II, Paper IV and Paper V of the thesis quantitatively

evaluated several migration paths from existing AON to NGOA.
The key elements of both CAPEX and OPEX are assessed and
compared.

• Paper VII of the thesis proposed an innovative OTN solution

featured with WDM in the data plane and SDN in the control
plane. The OTN can flexibly and automatically adjust to the
converged network services.

6.1 Conclusions
Paper I investigated AON migration strategies from various aspects, i.e.,
data plane, topology and control plane aspects. It proposed WDM-PON
and TWDM-PON as data plane migration path for AON PtP and AON
AS, respectively. Both paths have the advantage of reusing the existing
fiber infrastructure from legacy AON, and therefore a large part of
CAPEX can be saved. Mesh/ring topology and node consolidation were
proposed as topology migration options. The first one is making use of
the distributed nodes to improve network resiliency and traffic locality
feature, while the second one aims at reducing the number of access
network nodes. For the control plane migration, SDN solutions were
proposed. It simplifies network control and management processes,
reduces operational complexity, and improves the network efficiency.
For a network operator/provider, the selected migration strategy can be
either based on only one of the aspects or a combination of different
aspects.
Conclusions and future works 61

The thesis developed a general methodology for performing Techno-

Economic analysis of the fiber access network solutions based on the
work in Paper II and Paper III of the thesis. The methodology includes
three major steps. The first step defines important preconditions such as
business models, service demands, customer penetration, geographical
model. The second step defines proper technological solutions that can
satisfy the preconditions. The technological solution can be developed
from data plane, control plane and topology perspective. The third step
merges previous two steps together, and then, translates the
technological and business parameters into costs. The outcome of this
work provides a tool to assess the TCO of different solutions and find
out a technology/architecture that is the most suitable for network
operators’ own business cases.

Paper II, Paper IV, Paper V and Paper VI performed several TE studies
and provided quantitative evaluations on various optical access
network solutions. Paper II, Paper IV and Paper V focused on the
network migration scenarios. Paper VI assessed the CAPEX of Gigabit
Ethernet AON and 10G TDM-PON in a greenfield deployment scenario.
The TCO results show that the CAPEX of migration to TWDM-PON and
WDM-PON is higher than the other considered migration paths.
However, the OPEX per year in TWDM-PON and WDM-PON are lower
than others. Moreover, the longer operation period after migration is
considered, the higher economic benefits of the AON migration to
TWDM-PON and WDM-PON can be achieved. The impact of different
business roles on the TCO was investigated in Paper II and Paper V.
The results indicate that NP is responsible for more than 60% of the total
migration cost and thus, NP can be considered as the dominant player
in the AON migration. PIP has less OPEX because most of the
infrastructure is passive. The TCO results of node consolidation show
that the cost savings in the aggregation network is large enough to cover
the increased migration costs in the access network. Therefore, node
consolidation is beneficial for the AON network migration, especially in
the rural areas.

Paper VII of the thesis developed an innovative OTN solution featured

with WDM in the data plane and SDN in control plane. The OTN can
flexibly accommodate converged services from both mobile and fixed
networks, and automatically adjust itself to the different service
requirements. The thesis also introduces a multi-domain orchestration
architecture for cross-domain services. This global orchestration process
 Conclusions and future works 62

enables agile and efficient creation of diverse services across multiple

domains.

6.2 Future work

The follow-up work of this thesis could include:

• A quantitative investigation of the impact of Software Defined

Networking (SDN) on the TCO of optical metro/access networks.
The SDN is proposed in this thesis as control plane migration
options. However, the quantitative results are missing. It would
be interesting to perform a Techno-Economic (TE) study of SDN
based optical metro/access network.

• An investigation of the impact of business model, customer

demand, timing etc. on the TCO results of the network migration.
This thesis focuses on the TE comparison of different optical
metro/access technologies, while other important parameters that
are not highly technology-dependent such as business models
and strategies, traffic models, regulatory aspects, customer
penetrations, migration timing and duration of the migration are
simplified in the thesis. However, these parameters are very
important for the TE study and may affect TCO results
differently. In order to get a more reliable and realistic TCO
results, the impact and sensitivity of these parameters should be
further studied.

• Cross-domain TE study is desired for future converged network.

For example, a TE analysis of different OTN solutions should
cover fixed network, mobile backhaul/fronthaul network, and
cloud domains. Currently, most of TE analysis is only focused on
an individual network domain, where the TE result may indicate
that a certain technological solution is cost-effective for the
specific network domain. . However, the solution may lead to a
higher cost in other network domains. Today some network
domains tend to be converged, and hence it is important for TE
study to consider the converged network domains, so that the
evaluated technological solutions can have the best TCO
performance across multiple network domains
 References 63

7 References
[1] "Towards 50 billion connected devices", Ericsson Discussion
Paper, 2010, available online:
https://www.ericsson.com/au/res/region_RASO/docs/2010/ericsso
n_50_billion_paper.pdf

[2] Dave Evans, "The Internet of Things, How the Next Evolution of
the Internet is Changing Everything", Cisco White paper, 2011.

[3] ITU-T Recommendation, G Series, G.992.1 Amendment 1,

“Asymmetric digital subscriber line (ADSL) transceiver,” 2003.

[4] ITU-T Recommendation, G Series, G.993.2, “Very high speed

digital subscriber line transceivers 2,” 2015.

[5] Oksman, V., Schenk, H., Clausen, A., Cioffi, J.M., Mohseni, M.,
Ginis, G., Nuzman, C., Maes, J., Peeters, M., Fisher, K.D. and
Eriksson, P.E., "The ITU-T's new g.vector standard proliferates
100 mb/s dsl," IEEE Communications Magazine, vol. 48, no. 10,
pp. 140-148, 2010.

[6] Nokia press release, “Nokia and Deutsche Telekom show how
XG-FAST technology can extend copper network speeds and
meet future data demands,” 2016.

[7] W. Coomans, R. B. Moraes and K. Hooghe, "XG-FAST: Towards

10 Gb/s copper access," IEEE Globecom Workshop, USA, 2014.

[8] IEEE 802.3.ah Standard, “IEEE Ethernet in the First Mile”, 2005

[9] ITU-T Recommendation, G Series, G.984 series of

recommendations on Gigabit-capable Passive Optical Network,
2008.

[10] Claus Popp Larsen, “Access Network Technologies:

Deployments in Sweden” (invited), BAT Symposium at ECOC,
Germany, 2007.

[11] Paul-François Fournier (France Telecom), “From FTTH pilot to

pre-rollout in France”, 2007.
References 64

[12] J. Finn, "PON Technology in the Verizon Network," IEEE Global

Telecommunications Conference, USA, 2008.

[13] British Telecom Broadband service options 2017:

http://www.productsandservices.bt.com/products/broadband-
packages/?s_intcid=con_hpni_pns_bbpkgs .

[14] AT&T Broadband service options 2017:

https://www.att.com/internet/.

[15] Time Warner Cable broadband service options 2017:

https://www.timewarnercable.com/en/plans-
packages/internet/internet-service-plans.html.

[16] D. Wetherall, “Computer Communication and Networks, lecture

1,” University of Washington, course material, CSE561 Lecture
Notes and Handouts, 2003.
https://courses.cs.washington.edu/courses/cse561/04au/lectures/le
cture01-plan.txt

[17] Cisco white paper, “The Zettabyte Era—Trends and Analysis”,

2016.

[18] European Commission Digital Agenda scoreboard 2013,

http://europa.eu/pol/pdf/flipbook/en/digital_agenda_en.pdf

[19] AT&T 2014 annual report, 2014.

https://www.att.com/Investor/ATT_Annual/2014/downloads/att_
ar2014_annualreport.pdf

[20] J. F. Monserrat, et al., “Rethinking the mobile and wireless

network architecture: the METIS research into 5G,” Proceedings
of European Conference on Networks and Communications
(EuCNC), 2014.

[21] E. Dahlman, G. Mildh, S. Parkvall, J. Peisa, J. Sachs, and Y. Selén,

“5G radio access,” Ericsson Technology Review, 2014

[22] Next Generation Mobile Networks (NGMN), “5G white paper,”

2015.

[23] The 5G infrastructure Public Private Partnership (5G-PPP), “5G-

PPP 5G Architecture,” 2016.
References 65

[24] S. Gosselin et al., "Fixed and Mobile Convergence: Needs and

Solutions," European Wireless, 20th European Wireless
Conference, Spain, 2014.

[25] N. Madamopoulos, S. Peiris, N. Antoniades, D. Richards, B.

Pathak, G. Ellinas, R. Dorsinville, and M. A. Ali, "A Fully
Distributed 10G-EPON-based Converged Fixed–Mobile
Networking Transport Infrastructure for Next Generation
Broadband Access," Journal of Optical Communications and
Networking, vol. 4, no. 5, pp. 366-377, 2012.

[26] B. Skubic and I. Pappa, "Energy consumption analysis of

converged networks: Node consolidation vs metro
simplification," Optical Fiber Communication Conference and
Exposition and the National Fiber Optic Engineers Conference
(OFC/NFOEC), USA, 2013.

[27] ITU-T Recommendations, G.989 Series Recommendations, “40-

Gigabit-Capable Passive Optical Network (NG-PON2),” 2015.

[28] T. Pfeiffer, "Next Generation Mobile Fronthaul Architectures,"

Optical Fiber Communication Conference, USA, 2015.

[29] O. Malik, “In the Netherlands, 1 Gbps Broadband Will Soon Be

Everywhere”, 2010. https://gigaom.com/2010/02/19/in-
netherlands-1-gbps-broadband-will-soon-be-everywhere/

[30] O. Malik," So Where Else in the World Can You Get 1 Gbps to the
Home?" 2010. https://gigaom.com/2010/02/11/so-where-else-in-
the-world-can-you-get-1-gbps-to-the-home/

[31] B. Zhu and D. Nesset, "GPON reach extension to 60 km with

entirely passive fibre plant using Raman amplification," 35th
European Conference on Optical Communication, 2009.

[32] Full Service Access Network (FSAN) Group Forum. Available:

http://www.fsan.org/

[33] ITU-T Recommendation G.987, 10-Gigabit-capable passive

optical network (XG-PON) systems: Definitions, abbreviations,
and acronyms. 2010.

[34] IEEE 802.3av 10Gbit/s Ethernet Passive Optical Networks

standard, 2009.
References 66

[35] ITU-T Recommendation G.9807.1, 10-Gigabit-capable symmetric

passive optical network (XGS-PON), 2016.

[36] Ger Bakker, “UNET WDM PON strategies: Examining why,

where, and how UNET is deploying WDM PON”, IRR FTTx
Summit, Germany, 2009.

[37] Han Hyub Lee et al, “WDM PON experience and direction”,
FSAN Workshop, Laforet Biwako, Japan, 2009.

[38] D. Nesset, "NG-PON2 technology and standards." Journal of

Lightwave Technology, vol. 33, no. 5, pp. 1136-1143, 2015.

[39] K. Grobe, M. Roppelt, A. Autenrieth, J.-P. Elbers, and M. Eiselt,

“Cost and energy consumption analysis of advanced WDM-
PONs,” IEEE Communications Magazine, vol. 49, no. 2, pp. 25-32,
2011.

[40] Valerie Chaillou, “Inventory of FTTH/B in Europe”, IDATE

DigiWorld Institute, 2012. Available:
http://www.idate.org/en/News/Inventory-of-FTTH-B-in-
Europe_765.html

[41] C. Lange, R. Huelsermann, D. Kosiankowski, F. Geilhardt, A.

Gladisch. "Effects of network node consolidation in optical access
and aggregation networks on costs and power consumption", in
conference Optical Metro Networks and Short-Haul Systems II,
Proceedings of the SPIE, vol. 7621, 2010

[42] E. Nygren, R.K. Sitaraman, and J. Sun. "The Akamai Network: A

Platform for High-Performance Internet Applications, ACM
SIGOPS Operating Systems Review, vol. 44, no. 3, pp. 2-19, 2010.

[43] V. Nordell, A. Aurelius, A. Gavler, A. Arvidsson, M. Kihl,

"Concurrency and locality of content demand," International
Conference on Smart Communications in Network Technologies,
2013.

[44] Open Networking Foundation, “Software-Defined Networking:

The New Norm for Networks,” Open Networking Foundation
White Paper, 2012.

[45] EU FP7 Optical Access Seamless Evolution (OASE) project

deliverable D5.3: “Techno-economic assessment studies”, 2011.
References 67

Available:
http://cordis.europa.eu/docs/projects/cnect/5/249025/080/delivera
bles/001-OASED53WP5TUM220113v10.pdf

[46] M. Forzati, C.P. Larsen, and C. Mattsson, “Open access networks,

the Swedish experience,” 12th International conference on
transparent optical networks, 2010.

[47] European Commission, “Broadband Investment Guide”,

Publications Office of the European Union, 2014.

[48] R. Hack, "The Google fiber project," Journal of Computing

Sciences in Colleges, vol. 28, no. 5, 140-140, 2013.

[49] EU FP7 Optical Access Seamless Evolution (OASE) project

deliverable D 2.1, “Requirements for European next-generation
optical access networks,” 2011. Available: http://www.ict-
oase.eu/public/files/OASE_D2_1_WP2_DTAG_010611_V1_0.pdf

[50] M. D. Schlesinger, T. Heger, T. Monath and M. Kind, "FTTH

infrastructure roll out - Sensitivity analysis of monthly
termination end point fees," 10th Conference of
Telecommunication, Media and Internet Techno-Economics
(CTTE), Germany, 2011.

[51] M. Van der Wee, S. Verbrugge, M. Tahon, D. Colle, and M.

Pickavet, "Evaluation of the Techno-Economic Viability of Point-
to-Point Dark Fiber Access Infrastructure in Europe," Journal of
Optical Communication Networks, vol.6, no.3, pp. 238-249, 2014.

[52] R. Hülsermann and C. Lange, "Topology Analysis of the

Access/Aggregation Network Structure in Future Optical Access
Networks," 2010 ITG Symposium on Photonic Networks,
Germany, 2010

[53] EU FP7 Optical Access Seamless Evolution (OASE) project.

Available: http://cordis.europa.eu/project/rcn/93075_en.html

[54] The object management goup specification BPMN 2.0, Business

Process Modelling Notation (BPMN), 2011,
http://www.omg.org/spec/BPMN/2.0

[55] C. Mas Machuca, S. Krauß, K. Casier, “Fault Management and

Service Provisioning Process Model of Next Generation Access
References 68

Networks” International Conference on Network and Service

Management, France, 2011.

[56] K. Casier, K., S. Verbrugge, R. Meersman, D. Colle, M. Pickavet,

P. Demeester, ”A clear and balanced view on FTTH deployment
costs”, Journal of the Institute of Telecommunications
Professionals, vol. 2, no. 3, pp. 27-30, 2008.

[57] S. Chatzi, J. A. Lazaro, J. Prat and I. Tomkos, "Techno-economic

comparison of current and next generation long reach optical
access networks," 9th Conference on Telecommunications
Internet and Media Techno Economics (CTTE), Belgium, 2010.

[58] R. Hülsermann, K. Grobe, and D. Breuer, "Cost and Performance

Evaluation of WDM-based Access Networks," in Optical Fiber
Communication Conference/National Fiber Optic Engineers
Conference, USA, 2013.

[59] J. Zhang, “Tutorial on Small Cell and Heterogeneous Network

(HetNet) Deployment”, Industry Tutorials at IEEE GLOBECOM
2012.

[60] Small cell forum, “About small cell: small cell definition”,
http://www.smallcellforum.org/about/about-small-cells/small-
cell-definition/

[61] China Mobile Research Institute, ‘C-RAN the road towards green
RAN’, white paper, 2011.

[62] Common Public Radio Interface, “Specification version 7”, 2015.

[63] IEEE 1914 working group, task force 1914.1, “Standard for
Packet-based Fronthaul Transport Networks,” available:
http://sites.ieee.org/sagroups-1914/p1914-1/

[64] IEEE 1914 working group, task force 1914.3, “Standard for Radio
Over Ethernet Encapsulations and Mappings,” available:
http://sites.ieee.org/sagroups-1914/p1914-3/

[65] Common Public Radio Interface, “Industry leaders agree to

develop new CPRI Specification for 5G”, 2016,
http://www.cpri.info/press.html
References 69

[66] T. L. Marzetta, "Massive MIMO: An Introduction," in Bell Labs

Technical Journal, vol. 20, pp. 11-22, 2015.

[67] 3GPP TR 36.819, “Coordinated multi-point operation for LTE

physical layer aspects,” 2011.

[68] B. Skubic, G. Bottari, A. Rostami, F. Cavaliere and P. Öhlén,

"Rethinking Optical Transport to Pave the Way for 5G and the
Networked Society," in Journal of Lightwave Technology, vol. 33,
no. 5, pp. 1084-1091, 2015.

[69] P. Öhlén, B. Skubic, Z. Ghebretensaé, W. John and M.

Shirazipour, "Software-defined networking in a multi-purpose
DWDM-centric metro/aggregation network," IEEE Globecom
Workshops, 2013

[70] A. Autenrieth et al., "Evaluation of virtualization models for

optical connectivity service providers," 2014 International
Conference on Optical Network Design and Modeling, 2014.

[71] OpenDaylight, "Linux Foundation Collaborative Project."

Available: http://www.opendaylight.org

[72] Gartner, "Gartner Says Cloud Computing Will Be As Influential

As E-business," 2010, available:
http://www.gartner.com/newsroom/id/707508

[73] I. Drago, M. Mellia, M. M. Munafo, A. Sperotto, R. Sadre, and A.

Pras, "Inside dropbox: understanding personal cloud storage
services." In Proceedings of the 2012 ACM conference on Internet
measurement conference, pp. 481-494, 2012.

[74] Google Drive, available: https://www.google.com/drive/

[75] Microsoft SharePoint Online, available:

https://products.office.com/en-us/sharepoint/sharepoint-online-
collaboration-software