dwdm planning

Project P709

Planning of Full Optical NetworkDeliverable 1

Considerations on Optical Network Architectures: Functionalities,Configurations and Client Signals

Suggested readers:

• Managers, PNO Optical Network planners

• Experts on Standard Bodies (ITU T SG-13/15 and ETSI TM1 WG2/3)

• Optical systems and equipment manufacturers

For full publication

January 1999

1999 EURESCOM Participants in Project P709

EURESCOM PARTICIPANTS in Project P709 are:

• Finnet Group

• Swisscom AG

• Deutsche Telekom AG

• France Télécom

• MATÁV Hungarian Telecommunications Company

• TELECOM ITALIA S.p.a.

• Portugal Telecom S.A.

• Telefonica S.A.

• Sonera Ltd.

This document contains material which is the copyright of certain EURESCOMPARTICIPANTS, and may not be reproduced or copied without permission

All PARTICIPANTS have agreed to full publication of this document

The commercial use of any information contained in this document may require alicense from the proprietor of that information.

Neither the PARTICIPANTS nor EURESCOM warrant that the informationcontained in the report is capable of use, or that use of the information is free fromrisk, and accept no liability for loss or damage suffered by any person using thisinformation.

This document has been approved by EURESCOM Board of Governors fordistribution to all EURESCOM Shareholders.

Deliverable 1 Considerations on Optical Network Architectures

1999 EURESCOM Participants in Project P709 page i (ix)

Preface

(Prepared by the EURESCOM Permanent Staff)

Network traffic is increasing at an unprecedented rate, driven by the dramatic growthof the Internet and corporate data communications. The evolution of photonics makesthe development of optical switching and routing structures in the core andmetropolitan part of the transport network possible. This brings an increase incapacity and reduces transport costs.

The Wavelength Division Multiplexing (WDM) technique jointly with optical cross-connect (OXC), and Optical Add-Drop Multiplexing (OADM) equipment, will permitthe realisation of a switched optical layer based on wavelength routing of semi-permanent paths and fast protection/restoration mechanisms for the large amount ofinformation flows carried on the optical links.

As a consequence, the development of an optical network infrastructure will enablethe flexible, reliable and transparent provision of transport services for any type oftraditional and innovative services and applications.

Taking into consideration the current trends, the objective of network planning is tofind the best possible balance between network implementation cost, networkflexibility, network availability and survivability, subject to service requirements andtopological constraints.

The aim of the P709 EURESCOM Project is to investigate a number of alternativestrategies for the planning of the optical transport network - with massive deploymentof WDM, OADM, and small size OXC - that will be used in a middle term future.

This is the first Deliverable (D1) of P709. D1 provides an overview over networkarchitectures, which potentially may be used in the future. It also summarises therequirements on optical networks as well as maturity and availability of opticalfunctionalities. It should be noted that the Deliverable could not include all newfunctionalities of optical devices since it is an ongoing technology and due to thelimited study period, this was not possible.

P709 is a logical continuation of the P615 Project (Evolution towards an opticalnetwork layer) and some input from this Project was used in D1.

D1 is a very useful study for Managers, Optical Network planners, and experts onStandard Bodies of ITU-T SG15 and ETSI TM1 (WG2 & WG3).

Considerations on Optical Network Architectures Deliverable 1

page ii (ix) 1999 EURESCOM Participants in Project P709


1999 EURESCOM Participants in Project P709 page iii (ix)

Executive Summary

Optical WDM network is gaining more and more attention and is being implementedin a number of field trials. Several commercial products are appearing on the marketwith certain maturity. In USA, Europe and Japan, most of PNOs are planning toincrease the capacity of their transport network with massive deployment of WDMpoint-to-point system as well as fixed OADM and small size OXC. The aim ofEURESCOM Project P709 ‘Planning of full optical network’ is to investigate anumber of alternative strategies for the planning of optical transport network.

This Deliverable D1, the first one of P709 Project, concludes the results of Task 2‘Considerations on Optical Network Architectures’ activities. This document is aimedat those people who work on Network Planning for PNOs, Experts on StandardBodies related to the optical technologies, systems and networks, or manufacturersbuilding equipment for WDM networks.

The fundamental idea of the document is to show how the WDM technique couldbring new network architectures through the use of novel optical functions, and howthe latter optical network layer could transport multi-client signals. It could providecompanies with an overview of the state-of-the-art of optical functions, and usefulconsiderations on optical network architectures impacting the network planningprocess.

The first part of this Deliverable discusses the general characteristics of opticalfunctions as they are available now or will be in the near future. Commercial WDMpoint-to-point systems are also described and compared.

Different classes of network architecture, from the simple topologies to more complexstructures are presented in Section 2 in order to select reference network architectures.The possible combinations of basic optical network architectures are collected, inrelation with work carried out in P615 Project. The resulting selection of referencetwo-level network architectures is the following:

• CS-Ring architecture

• OMS-SP Ring architecture

• mesh-ring architecture

• ring-mesh architecture

The document goes on to discuss important network parameters which characterisethe WDM networks in terms of architecture, demand, physical limitation, topologyand survivability.

In the last Section, the possibility to plan an optical network using a non SDH clientsignal is proposed. After a brief investigation into ATM and IP client signalsperformance and functionalities, multi-layer network configurations are proposedusing IP, ATM, SDH and WDM network functionalities. A first evaluation of ATMover WDM, IP over SDH, IP over ATM and IP directly over WDM configurations isdiscussed.

Main achievements of the Deliverable are:

• identification of new optical functions considered necessary in order to enablethe migration to WDM future optical networks or desirable to enhance offerednetwork functionality


page iv (ix) 1999 EURESCOM Participants in Project P709

• selection of optical network architectures

• contribution to determine the physical limitation and typical values of networkcharacteristic parameters

• contribution to determine the ability of planning an optical layer carrying nonSDH signals.

While addressing considerations on optical network architectures, the elementaryoptical functionalities, network configurations and the possibility of carrying nonSDH client signals are discussed. The selected optical network architectures will beused in Task 3 and Task 4 in comparative studies of planning methodologies.


1999 EURESCOM Participants in Project P709 page v (ix)

List of Authors

Jamil CHAWKI France Télécom BD-CNET Task 2 & PIR 2.4 Leader

António Jaime Ramos: Portugal Telecom/ CPRM-Marconi PIR 2.3 Leader

Hélder Gaspar: Portugal Telecom/ CPRM-Marconi

Eduardo Sampaio: Portugal Telecom/ CPRM-Marconi

Reinald Ries: Deutsche Telekom AG PIR 2.2 leader

Ralf Herber: Deutsche Telekom AG

Paulette Gavignet: France Télécom BD-CNET

André Hamel: France Télécom BD-CNET

François Tillerot: France Télécom BD-CNET

Géza Paksy: Hungarian Telecom MATAV

Teresa Almeida: Portugal Talcum

António Jaime Ramos: Internal Reviewer

Dag Roar Hjelme (Sintef, Norway): External Reviewer

Martjin Luyten (NL): External Reviewer


page vi (ix) 1999 EURESCOM Participants in Project P709

Table of Contents

Preface .............................................................................................................................i

Executive Summary...................................................................................................... iii

List of Authors................................................................................................................v

Table of Contents ..........................................................................................................vi

Abbreviations ............................................................................................................. viii

Introduction ....................................................................................................................1

1 Assessment of optical functionalities and WDM point-to-point systems ...................31.1 Description of available functions....................................................................3

1.1.1 Signal Transport (Single mode fibre)..................................................31.1.2 Transmitter ..........................................................................................31.1.3 Receiver...............................................................................................41.1.4 Transponder.........................................................................................41.1.5 Dispersion compensation ....................................................................41.1.6 Optical Amplifier OA, 1R (EDFA) .....................................................51.1.7 Filters...................................................................................................51.1.8 Optical Add Drop Multiplexer OADM...............................................61.1.9 Space switch (matrix) ..........................................................................6

1.2 WDM point-to-point Systems...........................................................................71.2.1 Description of a WDM point-to-point link..........................................71.2.2 N x 2.5Gbit/s systems..........................................................................8

1.3 Identification of new/desirable optical functions ..........................................101.3.1 Wavelength conversion .....................................................................111.3.2 Optical signal monitoring functions (QoS, optical spectrum,

and Failure detection) ....................................................................111.3.3 Optical 3R regeneration.....................................................................111.3.4 Network survivability (protection, restoration).................................111.3.5 Management functions ......................................................................121.3.6 Optical time domain multiplexing OTDM (Long term

function) ........................................................................................121.3.7 Optical packet switching (Long term function) ................................12

1.4 Conclusion ......................................................................................................12

2 Assessment of optical network architectures ............................................................132.1 Complex topologies ........................................................................................13

2.1.1 Connected rings .................................................................................132.1.2 Meshed domains interconnected by a ring trunk...............................132.1.3 Ring domains interconnected by a meshed trunk..............................13

2.2 Characteristic parameters ...............................................................................132.2.1 Specific characteristics of optical network .......................................142.2.2 Parameters related to topology ..........................................................152.2.3 Parameters related to physical limitations.........................................162.2.4 Parameters related to demands ..........................................................172.2.5 Parameters related to architecture .....................................................182.2.6 Parameters related to the survivability approach ..............................19

2.3 Selection of reference network architectures .................................................192.3.1 Two-level CS-Ring architecture........................................................20


1999 EURESCOM Participants in Project P709 page vii (ix)

2.3.2 Two-level OMS-SP Ring architecture .............................................. 202.3.3 Two-level mesh-ring architecture ..................................................... 212.3.4 Two-level ring-mesh architecture ..................................................... 222.3.5 Characteristics of the selected optical network architectures ........... 22

2.4 Identification of physical network parameters limitation .............................. 232.4.1 Identification of mechanisms originating limitations [4, 5].............. 232.4.2 Identification of systems/components which introduce

limitations...................................................................................... 242.4.3 Processes to overcome limitations at present and solve them in

the future ....................................................................................... 252.5 Identification of Ranges of values.................................................................. 25

2.5.1 Functional layer characteristics ........................................................ 262.5.2 Ranges of values ............................................................................... 26

2.6 Conclusions .................................................................................................... 28

3 Potential of WDM routing for different client signals.............................................. 293.1 ATM client signal........................................................................................... 29

3.1.1 ATM Network functionalities and physical layer............................. 293.1.2 ATM Services ................................................................................... 293.1.3 ATM Performance Parameters.......................................................... 29

3.2 IP client signal ................................................................................................ 303.2.1 Internet network layers and services ................................................. 303.2.2 IP protocols : IP v4/v6, RTP and RSVP ........................................... 31

3.3 Network configurations required by ATM/IP client signals.......................... 323.4 Impact of non SDH client signal on the planning of optical network ........... 33

3.4.1 ATM over SDH over WDM : SDH protection vs. WDMprotection....................................................................................... 33

3.4.2 Configuration ATM over WDM....................................................... 353.4.3 Configuration IP over ATM [9] ........................................................ 363.4.4 Configuration IP over SDH [10, 11, 12] ........................................... 373.4.5 Configuration IP over WDM ............................................................ 37

3.5 Conclusion...................................................................................................... 38

4 Conclusion ................................................................................................................ 39

References.................................................................................................................... 40

Appendix 1: Recent Progress in the Performance of Optical TransmissionSystem Components ............................................................................................. 41


page viii (ix) 1999 EURESCOM Participants in Project P709

Abbreviations

AAL ATM Adaptation Layer

ABR Available Bit Rate

ACK ACKnowledgement

APS Automatic Protection Switching

ATM Asynchronous Transfer Mode

AWG Arrayed Waveguide Grating

BER Bit Error Rate

CBR Constant Bit rate

CBFG Chirped Bragg Fibre Grating

CDV Cell Delay Variation

CER Cell Error Ratio

CLR Cell Loss Ratio

CMR Cell Miss-insertion Rate

CS Ring Coloured Section Ring

DA Dispersion Accommodation

DCF Dispersion Compensating Fibre

DFF Dispersion Flattened Fibre

DSF Dispersion Shifted Fibre

EDFA Erbium Doped Fibre Amplifier

FTP File Transfer Protocol

FWM Four Wave Mixing

HDLC High level Data Link Control

IP Internet Protocol

IPv4 / v6 Internet Protocol version 4 / version 6

LAN Local Area network

LLC Logical Link Control

MAPOS Multiple Access Protocol Over SDH

MCTD Mean Cell Transfer Delay

MS Multiplex Section

MSP Multiplex Section Protection

MS-SP Ring Multiplex Section Shared Protection Ring

OA Optical Amplifier

OADM Optical Add Drop Multiplexer


1999 EURESCOM Participants in Project P709 page ix (ix)

OC or OCH Optical Channel

OC-DP Ring Optical Channel Dedicated Protection Ring

O/E Opto-Electronic

OMS Optical Multiplex Section

OM-SDP Ring Optical Multiplex Section Dedicated Protection Ring

OMS-SP Ring Optical Multiplex Section Shared Protection Ring

OPS Optical Protection Switching

OSC Optical Supervision Channel

OTDM Optical Time Division Multiplexing

OTS Optical Transmission Section

OXC Optical Cross Connect

PDU Protocol Data Unit

POH Path Over Head

PPP Point-to-point Protocol

QoS Quality of Signal / Service

RSVP Resource Reservation Protocol

RTP Real Time Protocol

SDH Synchronous Digital Hierarchy

SDXC Digital Cross Connect

SHR Self-Healing Ring

SMF Single Mode Fibre

SNAP Sub Network Attachment Point

STM Synchronous Transport Module

TCP Transfer Control Protocol

UBR Unspecified Bit Rate

UDP User Datagram Protocol

VC Virtual Circuit of ATM or Virtual Container of SDH

VP Virtual Path of ATM

WDM Wavelength Division Multiplexing

WWW World Wide Web


1999 EURESCOM Participants in Project P709 page 1 (42)

Introduction

In order to cope with the increasing network traffic driven by the dramatic growth ofthe Internet and corporate data communications, the evolution of the opticaltechnologies makes the development of optical switching and routing structures in thecore and metropolitan part of the transport networks possible. The WavelengthDivision Multiplexing (WDM) technique jointly with optical nodes will permit thewavelength routing of semi-permanent paths and fast protection/restorationmechanisms in the optical layer.

The purpose of this document is to show how the WDM technique could bring newnetwork architectures through the use of available and forthcoming optical functions,and how the latter optical network layer could be compatible with the transport ofsignals with various formats.

As a logical continuation of the P615 Project (Evolution towards an optical networklayer), some input from this Project was used in this document. Beyond the scope ofthe P615 Project, this document provides results from the investigation of thecommercially available and the desirable optical functionalities especially concerningoptical amplifiers, advanced functions such as wavelength conversion and opticalnodes. It also provides more complex network configurations, based on networkinterconnections, and identifies characteristic parameters.

The first part of this Deliverable discusses the general characteristics of opticalfunctions as they are available now or will be in the near future. Commercial WDMpoint-to-point systems are described and compared. The optical functions needed forthe future WDM networks are progressing rapidly. A set of new optical functionsconsidered necessary in order to enable the migration to WDM future opticalnetworks or desirable to enhance offered network functionality, is also identified.

From these possible network functionalities described in Section 1, different classesof network architecture, from the simple topologies to more complex structures, arepresented in Section 2 in order to select reference network architectures. The resultingselection of reference two-level network architectures is the following :

• CS-Ring architecture

• OMS-SP Ring architecture

• mesh-ring architecture

• ring-mesh architecture

The selected optical network architectures will be used in Task 3 and Task 4 incomparative studies of planning methodologies.

In Section 2, the most important network parameters which characterise the WDMnetworks from the point of view of architecture, demand, physical limitation,topology and survivability, are summarised. Physical limitation and typical values ofnetwork characteristic parameters are presented in the last part of this section.

Finally, in Section 3, the possibility to plan an optical network using a non SDH clientsignal is proposed. After a brief investigation into ATM and IP client signalsperformance and functionalities, multi-layer network configurations are proposedusing IP, ATM, SDH and WDM network functionalities. A first evaluation of ATM


page 2 (42) 1999 EURESCOM Participants in Project P709

over WDM, IP over SDH, IP over ATM and IP directly over WDM configurations isdiscussed.



1 Assessment of optical functionalities and WDM point-to-point systems

Optical network planning activities have to reflect a variety of physical as well aspractical conditions and constraints in order to produce useful results. Among these isthe set of the available optical functions which is to be used for the construction of thenetwork under consideration.

In this first section of the Deliverable, the state-of-the-art of optical functions, as theyare available now or will be in the near future, will be presented. The performance ofavailable WDM systems (point-to-point) will be compared. In this part such resultsare already available from EURESCOM P615 Project. The state-of-the-art of opticalcomponents and of realised functions which can be used in WDM optical networks,however, is progressing very rapidly. Thus it is necessary to update the informationon the presently available optical functions. Finally, a list of desirable opticalfunctions will be discussed in the last paragraph.

1.1 Description of available functions

1.1.1 Signal Transport (Single mode fibre)

All-optical networks are based on a passive fibre infrastructure which serves as thephysical transport medium between the network nodes. The most relevant propertiesof transmission fibres are attenuation, dispersion and non-linearity. Standard singlemode fibres (SMF) as well as dispersion shifted or flattened fibres (DSF, DFF) arecommercially available with standardised properties according to ITU-Trecommendations G.652 ...655. TrueWave

fibre also is a kind of dispersionmanipulated fibre. With attenuation values close to that of SMF its dispersion,however, is kept non zero at a value optimised in order to produce minimumdistortions due to the combined effects of non-linearity and dispersion. In table 1 thecharacteristic data of various fibre types are summarised [1], [2] .

Fibre type SMF DFF DSF True wave

Zero dispersion wavelength (µm) 1.312 1.535 – 1.565 ? >1.56 1.518

Dispersion coefficient at 1.55µm(ps/(nm.km)

17 <2.7 0.1 – 3.5 -1 – 5.5

Table 1: Characteristic data for some types of single mode fibres

1.1.2 Transmitter

Laser diode with direct and integrated modulation

The standard optical transmitter element in WDM systems is a laser diode. Integratedlaser modulation, ILM, offers a high dispersion tolerance through the use of electro-absorption modulator on the same laser chip. Key features of these devices are :

• optical output power 0 to +10 dBm

• modulation bandwidth 10 GHz

• dispersion tolerance 1000 to 10000 ps @ STM-16



• stability & accuracy of optical frequency 0.05 nm

Laser diode with external modulation

External modulation is used in order to reduce the chirp of the optical transmitter andthus increase its dispersion tolerance. Transmitter modules are available for 10 Gbit/stransmission and in laboratory systems modulation bandwidth of 100 GHz has beendemonstrated. Apart from reduced width of the optical spectrum the other propertiesare the same as for directly modulated laser diodes.

1.1.3 Receiver

Optical receivers are found in optical line terminations and in transponders wherethey convert the signal from the optical into the electrical domain. State-of-the-artoptical receivers reach sensitivities close to –30 dBm for 2.5 Gbit/s and around –20dBm for 10 Gbit/s (BER 10E-10).

1.1.4 Transponder

Transponders are opto-electronic frequency converters which basically consist of anoptical receiver and transmitter. The receiver converts the optical input signal into theelectrical domain where it is amplified and sometimes even reshaped and re-timed.This signal is used to modulate a laser diode optical transmitter which produces therequired optical carrier frequency. Most WDM system manufacturers rely ontransponders as input interface into the WDM system. They are available now for bitrates up to 10 Gbit/s. The input sensitivity varies considerably (-5 dBm ... –20 dBm)for devices from different manufacturers. Transponders accept 1.3 µm- as well as1.55 µm input signals and their output powers are around 0dBm. Some manufacturersstill offer devices with output frequencies not matching the ITU-T recommendationsconcerning the WDM channel frequencies.

Transponders which do not regenerate the input signal will work with any type ofintensity modulated digital (binary) client signal independent of signal format (e.g.SDH, ATM...) and bit rate (155 Mbit/s, 622 Mbit/s, 2.5 Gbit/s, 10 Gbit/s) within thelimits of the specifications. Frequency- or phase modulated optical input signalscannot be used with transponders.

1.1.5 Dispersion compensation

Besides fibre attenuation it is the effect of fibre chromatic dispersion which mainlylimits the achievable repeater spacing in optical links. The origin of the latter effect isthe variation of the group delay as a function of the optical frequency. In fibre opticaltransmission lines the dispersion effect increases linearly with fibre length and widthof the optical spectrum and causes pulse distortion and bit interference. As chromaticdispersion is a linear effect it can be compensated by inserting additional appropriateoptical elements into the transmission link.

Dispersion compensating fibre

Dispersion compensating fibre (DCF) is a special type of fibre which for light in the1.55 µm wavelength region has negative dispersion coefficient in the order of –80ps/(nm.km). Thus 1km of DCF is needed to compensate the dispersion of about 5 kmof SMF as the corresponding value is 17 ps/(nm.km) in SMF.



The value of the dispersion coefficient varies as a function of optical frequency inDCF as well as in SMF. Therefore it is not possible to achieve perfect dispersioncompensation in a large frequency range. The attenuation of DCF with typical valuesaround 0.6dB/km is still considerably larger than that of SMF.

Chirped Bragg fibre grating

Using chirped Bragg fibre gratings (CBFG) is another option for dispersioncompensation. These devices provide a low loss solution. However, they work inreflection mode and therefore optical circulators or fibre couplers are necessary toseparate input and output signal. Presently compensation bandwidth of only somehundreds of GHz achievable with one CBFG is more limited compared to DCF. Awider bandwidth can be achieved through the use of longer gratings or cascadedgratings. But the requirement of additional circulators or couplers may be regarded asa drawback. On the other hand, their non-linearity is practically zero which may beparticularly important in very high bit rate systems with 10 Gbit/s and more.

1.1.6 Optical Amplifier OA, 1R (EDFA)

Erbium doped fibre amplifiers (EDFAs) are one of the key building blocks of WDMsystems. They allow the economical power amplification of all the signals in thedifferent WDM channels. The system relevant optical properties of EDFAsare: power gain, saturated output power, noise figure, optical bandwidth andpolarisation mode dispersion. The power gain is calculated as the ratio of output toinput signal power of the amplifier. This value directly determines the maximum linksegment attenuation between consecutive EDFAs. It depends on the number ofchannels and on total link length. In practical links this value varies from below 20 dBto 30 dB. The saturated output power is the upper limit of the total output power fromthe amplifier for high input power. Typical values range from 13 dBm to 17 dBmwhile EDFAs with output powers of up to 30 dBm are commercially available.

1.1.7 Filters

Fibre Bragg Grating Filters

These filters are based on photosensitivity in Ge-doped core optical fibres; reflectiongratings are written by illuminating the fibre with a standing wave interferencepattern. In recent years, fibre Bragg Gratings have proven successful as in line filters.This device has the advantages of being low loss, with a narrow pass-bandcharacteristic (0.5 nm) and potentially low cost. This is a very promising technologyfor fixed filtering with a channel spacing in the nm range. Tuneable filtering isobtained by stretching the fibre where the Bragg filter is deposited. They arecommonly used with optical circulators to obtain the OADM functionality.

Diffraction Gratings and Phased Arrayed Gratings

Grating devices are suited to address several wavelengths simultaneously becausethey pass a discrete set of predefined wavelengths. Two types are available; the firstone is a micro-optic diffraction grating. The typical insertion loss per channel of adiffraction grating device is 3 dB with a -30 dB adjacent crosstalk level.

The other type is an integrated optic device (SiO2 or InP) called Arrayed WaveguideGrating; for an AWG, the typical insertion loss is 5 dB and the crosstalk is 25 dB(figure 1). Components are commercially available for 8, 16 or 32 channels with 100



or 200 Ghz channel spacing and the insertion losses for a transit wavelength are 6 to10 dB.

Figure 1: Spectral response of a wide pass-band AWG

1.1.8 Optical Add Drop Multiplexer OADM

Concerning optical add and drop facilities, two main suppliers are offering them :Ciena and Pirelli. Indeed 8 channels can be added or dropped at each amplifier site inthe Ciena Multiwave 4000 and 12 in the Pirelli Wavemux. Figure 2 indicates wherethis function can be introduced in a WDM point-to-point system. It shows theexample of insertion and extraction of channels at an amplifier site.

The available add/drop functions are fixed but selectable add/drop facilities havealready been announced as well as reconfigurable OADMs and OXCs. This isplanned for the next years but to our knowledge no exact dates have been given.

am plifie rs

W D M

term inal

W D M

term inal

d rop add

transit

Figure 2: Possible position of an add/drop function in a WDM point-to-pointsystem

1.1.9 Space switch (matrix)

Switching matrices are available which are suited for realising of flexible OADMsand OXCs. Various approaches have been followed to perform the switchingfunction. Devices relying on mechanical operation contain actuators, e.g. motors,electro-statically or piezo-electrically deflected micro mirrors for the switching of theoptical signal. Due to the required mechanical movements of part the switching times



achieved so far range from 30ms to 500ms. Wave-guide devices which make use ofthermal or electro-optic effects are considerably faster as can be seen in table 2 whichcontains data of various switching matrices. With respect to insertion losses andchannel crosstalk wave-guide devices do not perform as well as mechanical switchingmatrices.

Technology mechanical thermal Electro- optic

actuator motor mirror Electro-static polymer thermal

size 64 x 64100 x 100

8 x 864 x 64

16 x 16 8 x 827 x 27

8 x 8 8 x 8

insertion loss (dB) 2 2 4 8 8 8

switching time (ms) 500 30 40 2 2 <2

channel isolation (dB) 60 60 60 50 40 30

Table 2: Characteristic features of optical switching matrices

1.2 WDM point-to-point Systems

First, we will describe what is generally called a WDM point-to-point system, thenthe performances of the WDM products will be compared and finally we will givesome elements concerning the future functionalities (OADMs and OXCs).

1.2.1 Description of a WDM point-to-point link

Figure 3 shows what is commonly called a WDM point-to-point system (in thebroken-line rectangle). We have considered a system which processes a bi-directionaltransmission over two fibres (which is more often the case) with in-line amplification.The non-amplified systems will not be considered in this document. Each of theoptical signals coming from SDH (or SONET) terminals (or ADMs) is sent into atransponder (optional or not) used to provide a wavelength compatible signal to theWDM terminal. Moreover, the signals which feed the WDM terminal must be able tobe transmitted on dispersive fibre (standard fibre G.652) over several hundreds of kms(typically 500). The WDM terminal contains a multiplexer (or a coupler) and abooster for one direction and a preamplifier and a demultiplexer in the other direction.Then, the optical multiplex is launched into the transmission fibre and regularlyamplified (each 80 to 120 km) by optical amplifiers (so without opto-electronicconversion). Another terminal, similar to the first one is placed at the end of the lineand the signals are either directly received by the SDH terminal or via a transponder.

SDHinter-faces

(tra

ns-

pond

ers)

WDM terminal

amplifiers

Multi-wavelength system (WDM)

SDHinter-faces

SDHinter-faces

SDHinter-faces

(tra

ns-

pond

ers)

WDM terminal

Figure 3: WDM point-to-point system description



The management of the system is made using an Optical Supervisory Channel (OSC)at a wavelength (very often outside the multiplex) which is dedicated to transportingfault, configuration and performance information. This channel is processed in theWDM terminal sites and also in each amplifier site (demultiplexing, detection,processing, emission, multiplexing). The systems with transponders in the WDMterminal are sometimes called ‘open systems’ (because they can theoretically acceptvarious input signal formats). Systems without this feature are called ‘embeddedsystems’ (they are less flexible but the cost can be reduced compared to the opensystems).

Today, the maximum bit rate per channel at the input of the WDM terminal (or thetransponder) is 10 Gbit/s. This is the maximum bit rate available with existing SDHTDM equipment (STM-64). Moreover, the transmission of 10 Gbit/s optical signal isdifficult, due, essentially, to chromatic dispersion and polarisation mode dispersion(PMD) limitations. In fact, many WDM systems use the multiplexing of 2.5 Gbit/ssignals. Some systems are already announced for a total capacity of 200 Gbit/s andmore with a number of wavelengths reaching 80 (and even 96). Some of the systemsaccept both 2.5 and 10 Gbit/s input rates (they will be indicated hereafter) and alsolower rates (155 and 622 Mbit/s). Concerning the Nx2.5 Gbit/s WDM systems within-line amplification, the ITU-T recommendation G.692 addresses 4 or 8 channelsmultiplexed together (with a possible extension to 16 or 32). The target transmissiondistance over G.652, G.653 or G.655 fibre is 640 km with nominal span lengths of 80(with 8 spans), and 120 km (5 spans). The channel grid is defined with 100 GHz inter-channel spacing (193.1 THz is the reference frequency).

1.2.2 N x 2.5Gbit/s systems

Table 3 shows the performance of the WDM systems on which information isavailable. It indicates the supplier and the name of the product, the availability, themaximum bit rate capacity, the distance between amplifiers and the maximumdistance that can be reached (it depends on the number of spans that can be cascaded).The systems are designed for G.652 standard fibre but some suppliers like Lucent alsorecommend G.655 fibre. The channel spacing is also indicated, giving an idea of theoptical bandwidth that is used for the transmission. We can see that the increase in thenumber of wavelengths has led to the reduction of the channel spacing from 200 GHzto 50 GHz (0.4 nm).



P r o d u c t n a m eA v a i l a -

b i l i t yC a p a c i t y ( G b i t / s )

M a x . d i s t a n c e

( k m )

λ n u m b e r

( N )

C h a n n e l s p a c i n g

( n m )S p a n ( k m )

A l c a t e l1 6 8 6 W M *

y e s 4 0 6 4 0 1 6 1 , 6 8 0

B o s c hF l e x P l e x T M O - 4

y e s 1 0 1 6 0 4 5 ? 4 0

C i e n aM u l t i w a v e 1 6 0 0

y e s 4 0 6 0 0 1 6 0 , 8 1 2 0

C i e n aM u l t i w a v e 4 0 0 0

y e s1 0 0 /

( 2 4 0 )4 0 0 4 0 ( 9 6 ) 0 , 4 1 0 0

D S C C o m .D S C F o c u s W D M

y e s ?1 0

( t o 4 0 )5 0 0

4( t o 1 6 )

_ 7 0 t o 8 0 ?

E r i c s s o n E R I O N , A X D 8 W

y e s 2 0 6 0 0 8 1 , 6 _

E r i c s s o n E R I O N , A X D 1 6 W

2 n d S e m . 9 8 ?

4 0 _ 1 6 0 , 8 ? _

L u c e n t / W a v e S t a rO L S 8 0 G

y e s 4 0 6 4 0 1 6 0 , 8 8 0

L u c e n t / W a v e S t a r O L S 4 0 0 G *

4 è Q . 9 8 2 0 0 6 4 0 8 0 0 , 4 _

N E C S p e c t r a l w a v e y e s ? 8 0 4 0 0 3 2 _ 8 0

P i r e l l iT 3 1

y e s 2 0 5 0 0 8 1 0 0

P i r e l l iW a v e M u x

2 n d S e m . 9 8 ?

8 0 / ( 1 6 0 )

6 0 03 2

/ ( 6 4 )0 , 8 / ( 0 , 4 ) 1 2 0

* 10 Gbit/s allowed inputs

Table 3: Description of the main point-to-point WDM systems

The management of these systems is done using software interfaces which allowaccess to the system state (alarms, configuration,...), the OSC channel being used totransport the required data. Each terminal or amplifier site can be accessed eitherdirectly (craft terminal) or remotely. Today the supervision data of the WDM system(in the optical layer) are processed independently of those of the SDH (or SONET)equipment (in the SDH layer). The new generation of WDM systems proposeevolution concerning management in order to have a global management of the twolayers. For that purpose, Ciena transports a few overhead bytes of the SDH layer inthe OSC channel.

The protection is implemented at the multiplex level (MSP 1+1) but by duplication ofthe whole WDM system (terminals and amplifiers), the switching being done by theSDH interfaces (in the SDH layer). This is the case for all the equipment suppliersexcept Ericsson, which proposes a switching from the working transmission line tothe protection one in the WDM terminal (in the optical layer). Moreover Ericsson hasintroduced the "Flexing Bus" concept which applies to a ring network topology withan unused section dedicated to the protection. It allows a rapid protection in theoptical layer.



manufacturer no. of channelsavailable (plan)

optical ADM fixedavailable (plan)

optical ADM flexibleavailable (plan)

OXC available(plan)

Alcatel 16 λ (40 λ 4Q98) - (?/40 4Q98) - (?) - (?)

Bosch 8 λ (16 λ 1Q99) - (4/16 4Q99) - (?) - (16 x 16 ?)

Ciena 40 λ (80 λ 4Q98) 8/16 (16/40 4Q98) - (? 4Q99) - (? 4Q99)

DSC 8 λ (32 λ 2Q99) ?/8 (?/32 4Q98) - (?/32 2Q99) - (?)

ECI - (8 λ 3Q98) ? (?) - (?) - (?)

Ericsson 16 λ (32 λ 1Q99) - (4/16 3Q98) 16/16 (32/32 1Q99) - (?)

Fujitsu 8 λ (? λ ?) ? (?) - (?) - (?)

Lucent 16 λ (80 λ 2Q99) - (?/16 4Q98) - (40/80 2Q99) - (? 2Q99)

NEC 32 λ (64 λ 4Q98) ? (?) - (?) - (8 x 8 ?)

Pirelli-Quante 32 λ (64 λ 1Q99) 12/32 (?) - (?/32 4Q99) - (16 x 16 4Q00)

Siemens 8 λ (32 λ 3Q99) - (32/32 3Q99) - (?/32 4Q99) - (16 x 16 4Q99)

Tellium 32 λ (64 λ 1Q99) 4/16 (?) (64/64 3Q99) - (128 x 128 3Q99)

Table 4: Specifications of WDM systems presently available and underdevelopment

Table 4 presents data of available systems as well as future plans of variousmanufacturers concerning number of WDM channels, fixed and flexible OADM andOXC. The first entry in each column is the value characteristic for the systems as theyare presently available from the corresponding manufacturer. The second entry inbrackets in the column indicates the manufacturers future plans. The plannedperformance as well as the estimated time of realisation are written wheneverpossible. A minus sign indicates that the function is not available at present and a (?)indicates that manufacturer information is not available.

1.3 Identification of new/desirable optical functions

In this part of the Deliverable, WDM optical functions considered either strictlynecessary or just desirable for enabling or enhancing optical networking, areidentified. The following functions were identified:

• OXC nodes

• Flexible OADMs

• Wavelength conversion

• Optical signal monitoring functions (QoS, optical spectrum, and Failuredetection)

• Optical 3R regeneration

• Wide band (80 nm) Optical Amplifiers

• Network survivability (protection, restoration)



• Management functions

• Dense WDM systems (64 and 128 λ)

• OTDM

• Optical packet switching

1.3.1 Wavelength conversion

Definition: to convert the optical frequency of an optical channel on a WDM combfrom its original position in the input signal comb to another position in the outputsignal comb; to convert a full comb from its original position in the fibre window toanother position in the same window .

Application: necessary function in order to improve flexibility in the optical networkby the use of wavelength domain switching on OXCs and OADMs. Allows opticalspectrum dynamic allocation that can be used for re-routing and survivability.Function that could be used to implement transponder function. Functiontransparency desirable, regarding signal origin, format and bit rate - all-opticalwavelength conversion. Some realisations provide 2R regeneration properties.

Availability: from 2000 onwards

1.3.2 Optical signal monitoring functions (QoS, optical spectrum, and Failuredetection)

Definition: monitor the optical signal, in order to determine BER, frequency accuracyand stability, and detect failures. Integration with existing network managementsystems desirable

Application: necessary function to assure QoS and activate survivability mechanismsin the network.


1.3.3 Optical 3R regeneration

Definition: perform optically, Amplification, Reshaping and Retiming of the opticalsignal

Application: necessary function to overcome “opacity” of optical systems, allowinghigher cascadability numbers.


1.3.4 Network survivability (protection, restoration)

Definition: ensures network fast recovery from a state of failure, using optical layermechanisms

Application: necessary function to assure provision of service with QoS, by re-routingtraffic on the network via alternative available routes, either pre-allocated(survivability dedicated resources -protection) or dynamically allocated (restoration)

Availability: protection: from 1998 onwards; restoration: from 2000 onwards



1.3.5 Management functions

Definition: allow management of optical network elements by an integrated networkmanagement system, via standardised management interfaces whenever possible.

Application: management of optical network elements by existing integrated networkmanagement systems.


1.3.6 Optical time domain multiplexing OTDM (Long term function)

Definition: optically multiplex digital optical signals in the time domain. This opticaldigital signal will be characterised by the optical wavelength, by technology (e.g.soliton, soliton +WDM), and by its bit rate. May be associated to WDM: N x OTDMsignals on a N optical channel WDM signal.

Application: increase the transmission bit rate beyond electronic limit (currently40GHz). Allow optical packet transmission and switching. Needs improving ofoptical signal processing techniques (clock recovery, all-optical regeneration) andnecessary components (short pulse optical sources, optical timing systems, OTDM-OADMs, OTDM-WDM converts, optical memories)

Availability: > 5 years

1.3.7 Optical packet switching (Long term function)

Definition: performs dynamic routing (switching) of optical packets

Application: enable very high-speed digital optical packet networks. Needsimprovement of optical signal processing techniques like optical addressing, opticalheader procession... and components such as fast optical switches, optical memories...

Availability: 5 years

1.4 Conclusion

After a presentation of available optical functions and WDM point-to-point systems, aset of new optical functions considered either strictly necessary in order to enable theintroduction/ migration to WDM future optical networks or desirable to enhanceoffered network functionality, was identified. A possible definition and theapplication context were analysed. Most of these functions have already beendeveloped and tested either in laboratory or in field trials.

The use of these functions can however not be considered independently of thearchitectural network context. The next Section proposes reference networkarchitectures for the optical layer, where the above described optical functions couldfind their place.



2 Assessment of optical network architectures

In this section we provide an overview of classes of network architectures, from thesimple topologies to more complex structures which potentially may be used in futureoptical networks. It points out most important network parameters, which characterisethe WDM networks in terms of architecture, topology and survivability. The aim ofthis section is to select and propose complex optical network structures for furtherstudy of Task3 and Task4. Typical values of parameters, related to topology, physicallimitations, are listed at the end of this section.

2.1 Complex topologies

A network characterised by a complex topology is composed of sub-networks whichare directly interconnected by sharing nodes. In such a network topology the aim is tooptimise the capacity of the network by mixing the different types of traffic on as fewnetwork elements as possible. The sub-networks will have basic topologies.

2.1.1 Connected rings

The combined advantages of good protection performance, low cabling costs andefficient use of the network elements can be achieved by a network structurecomposed of connected rings.

2.1.2 Meshed domains interconnected by a ring trunk

The ring trunk network gives excellent protection capabilities with a minimum ofinterconnections. However, the requirements of transport capacity between neighbouring nodesare high compared to the mesh, and therefore the trunk ring is mainly advantageous in areaswhere the cost of installing cables is high.

2.1.3 Ring domains interconnected by a meshed trunk

The meshed trunk network has the advantage of providing excellent node-to-nodephysical connectivity and, thereby, provides many alternative routes for traffic. Thetraffic on the cables in a meshed network can, to a large extent, be considered point-to-point and, therefore, the requirements to the transmission capacity on the links areeasy to predict.

2.2 Characteristic parameters

The purpose of this section is to present and discuss various parameters to be takeninto account to allow the future routing and dimensioning of the optical networks.Techniques and principles commonly applicable to existing networks do notnecessarily apply to the dimensioning of networks based on wavelength routing, andthis justifies the need for a specific analysis of network requirements.



2.2.1 Specific characteristics of optical network

• High capacity: the core network has to deal with many applications and servicesenvisaged for the future, which will probably have different bandwidths. Theoverall traffic volume is expected to be large and to increase as applications andservices become cheaper and easier to use. Hence, the network has to have alarge capacity and to be able to handle the granularity of optical channels.

• Transparency: in order to take into account most of the assets of opticalfunctions and to reduce the complexity of equipment, the signal should not beconverted to the electrical domain wherever it is possible. Several levels oftransparency could be specified such as signal format, bit-rate, transfer mode andservice. Full transparency usually does not exist, as physical constraints alwayscause transparency limitations.

• Flexibility : refers to the ability of the network to accommodate changes in trafficpatterns. This could be easier in optical networks since the granularity of handledsignals is higher.

• Connectivity is the network ability to establish connections independently of theactual state of the network. Full connectivity means that any connection betweenany two points of the network can be established at any time. In optical networksthe availability of wavelengths, OXC blocking or limited number of OADMwavelengths are the main barriers to full optical connectivity.

• Scaleability: is the possibility of capacity or functionality upgrade of a networkby adding new facilities in uniform steps. In optical networks the gradualincrease of available wavelengths without changing the whole WDM terminal isa key scaleability feature.

The relationship between characteristic parameters of different network layers andother network related requirements is shown in figure 4. During a network planningprocess topological, architectural and general requirements are to be considered andfulfilled.



Input data

Network planning

Geographical Topologicaltopology characterisics

Basic Architecturalnetwork characteristics

structuresNetwork characteristics

Service Functionalavailability layer

requirements characteristics CapacityTransparencyConnectivity

FexibilityGranularity

Transmission Functional Scalabilityperformances layer

characteristics

Available Networkequipment implementation

hw,sw

Network engineering

Operation

andMaintenance

Traffic demands

Figure 4: relationships of network characteristics and network requirements

2.2.2 Parameters related to topology

The definition of network design rules for WDM optical networks, the evaluation oftopology optimisation algorithms and network planning tools have to be done usingnumerical simulations. However, the demonstration of general results can be achievedin some cases, which provide simple design rules that could be used for guidelines innetwork planning. Some of the most important network parameters are:

• The number of nodes (N), a node being either a source of traffic (optical channel)or a pure transit node

• The node degree (D), defined as the mean number of nodes directly (i.e. withoutany transit) connected to a node via one or more fibres.

• The link length (LF) normalised to the node spacing.

• The number of fibre per link (F)

• The shape of the network (to be defined)

• The network density



From those parameters, it has been demonstrated [3] that a good estimate of W,number of wavelengths required to meet the traffic demand (T being the number ofchannels per connection) can be expressed as :

W

N

N T

FDLF

≈−

1

2 1 2

3 2

π

However, those parameters may not be fully sufficient for describing a network, sincetopological particularities can be noticed.

The network density d is defined by the formula: dD

N=

− 1 .

This parameter reflects the depth of the mesh in the network. Given a fixed number ofnodes in the network, extreme values are obtained with full mesh (d=1) and ring(d=2/(n-1)). The figure below illustrates various situations for network topologieswith a given link density.

Full mesh Mesh Ring Mesh

D = 4; N = 5 D = 3; N = 6 D = 2; N = 6 D = 2.5; N = 8

d = 1 d = 0.6 d = 0.4 d = 0.357

Figure 5: Examples of various graph densities

2.2.3 Parameters related to physical limitations

The quality of the signal across the network dictates engineering rules for networkplanning. In an all-optical network, the transmitted data remain as optical signals allalong the path in the optical layer. However, each path cannot be fully considered as apoint-to-point WDM link because : signals on different paths may travel through adifferent number of optical devices, the number of wavelengths and signalcharacteristics on fibres can differ with links. The physical limitations lead todegradation of signal quality through cross-talk, signal distortion and noiseaccumulation. The design of optical cross-connects and the definition of thearchitecture of the nodes should ensure the nodes provide the necessary functionality.However, there is the issue of how many nodes can be cascaded along one opticalpath while keeping the signal quality to an acceptable level. As long as theregeneration of the signal has to be performed by the electrical layer, a limitation onthe optical path length should be addressed as well. Another issue is related to thelimited number of wavelengths per fibre due to cross-talk problems and limitedamplifier bandwidth. The problem can be smoothed by wavelength re-use, whichcomplicates the wavelength allocation and routing problem. In summary, thefollowing parameters should be considered, which are closely related to the signalquality:

• the average and maximum link length in the network



• the maximum number of optical nodes that can be crossed without regeneration

• the maximum optical path length allowed without any regeneration

• the number of wavelength conversions along the path

• the number of wavelengths, optical channels, per fibre being used

• BER Degradation and system bandwidth

• optical channel individual integrity (Optical power level, Wavelength stability...)

• reconfiguration time (re-routing and switching time)

2.2.4 Parameters related to demands

The demand matrix (or matrices) has an important role in the planning process,because it can affect the selection of the architecture, the grooming and routingpolicy, the transport characteristics of the optical layer, and so on. In principle thedemand matrix for the optical layer should be expressed in wavelengths, but often it isexpressed in the typical unit of the client layer that leaves to the optical layer plannera higher degree of freedom in the optical layer design .

Demand distribution

The distribution of demands among optical nodes is an important characteristic of thenetwork. Different distributions can lead the planner towards different networkarchitectures and allow him to use different methodologies and algorithms in theplanning process. Both single wavelength and multi-wavelength demands should beaddressed.

Optical channel granularity

The design of present transport networks is based on static traffic conditions. A largeset of objectives can be taken into account in the optimisation. Entries, generally, aregiven in Mbit/s. The granularity, or channel capacity, defines the correspondencebetween the demand matrix in the optical layer and the wavelengths matrix. Thechoice of a granularity iss of courses dictated by technology. Most of multi-wavelength systems recently available use STM-16 granularity. However,announcements were made towards STM-64 granularity, allowing mixed granularityon the same fibre.

Capacity per link

The physical limitations lead to a limit in the number of optical channels per fibre.The transmitted power per wavelength must be large enough to provide an acceptableSignal to Noise Ratio at the receiver. However, it is not possible to increase signalpower since optical amplifier gain may saturate and non-linear effects like four-wavemixing will degrade signal transmission performance. Thus, there is a compromisebetween the capacity per link, the distance between amplifiers and the number ofcascaded amplifiers (link length).

Demand grooming and consolidation

Depending on the network services offered, various kinds of traffic demands can becarried in the optical layer : voice, video signals, data, leased lines... The bit-rateallocated to each kind of signal may be different. The planning process also includesthe grooming and consolidation of those signals towards the optical layer. Different



grooming alternatives can be taken into account, mainly based on the demanddistribution, their requirements and the target value of utilisation of the optical layer.For that purpose, grooming optimisation should also take into account the impact inthe optical layer.

O p tic a l la y e r

D e m a n d s(lo w e r le v e l)

G r o o m in g

Figure 6: Grooming process

2.2.5 Parameters related to architecture

Flat or hierarchical networks

A single layer network can be structured in two different ways: flat or hierarchical. Aflat network does not impose any constraints on the demand routing, while ahierarchical one allows, in principle, a communication between two peer hierarchicalnodes through one or more nodes of a highest hierarchy. Very often traffic in largetelecommunication networks is allocated to hierarchical levels or tiers. Two, three or,in some special cases, four level networks are designed. The main advantages oftraffic hierarchisation are the clear routing rules and easy manageability. However,sometimes this results in longer paths for individual demands. In the near future, all-optical networks with two hierarchical levels will probably be common for corenetwork applications. Many European projects considered a two-level all-opticalnetwork, which included an upper level meshed configuration based on OXCs, and alower level based on WDM rings.

Number of sub-networks on each hierarchical level

Both a flat and a hierarchical network can be split into several sub-networks.Generally a sub-network is defined by the node connectivities and by theindependence of its own survivability mechanism (that is the reason why these sub-networks are often called Survivable Sub-Networks). Generally, in a hierarchicalnetwork, interconnections of sub-networks in the same network level are allowed onlyvia the next higher hierarchical level. The number of sub-networks on a network levelis not limited, and it depends only on the actual network size.

Types of sub-networks

A sub-network can be any network partition, but usually consists of some kind ofbasic network topology, like rings or small meshes. Candidates for sub-networks arein optical networks WDM rings and optical meshes.

Number of transiting nodes (hubs) per sub-network

Unless the demand distribution does not need to cross more than one sub-network,each sub-network must have at least one special node where the outgoing traffic istransited. These nodes (often called hub nodes) have special functionalities for inter-working with other sub-networks.



2.2.6 Parameters related to the survivability approach

A protection or a restoration mechanism (or both) is generally applied to a network inorder to increase the demand survivability against failures. The recovery mechanismoften adds some new constraints to the network planning activity such as restorationtime, length of paths, routing ... These constraint will be addressed in the DeliverableD2.

2.3 Selection of reference network architectures

Having studied network topologies we now have to select the reference networkarchitectures that will be implemented in those topologies. In practice the PNO’s corenetwork is divided into hierarchical levels from the traffic routing point-of-view. Thishierarchisation can be naturally translated to appropriately interconnected topologicaldomains, resulting in complex network topologies. The different domains can then beimplemented using several network architectures. In the following we investigatethose kinds of hierarchical optical architectures which consist of the basic opticalnetwork domains like:

• CS-Rings,

• OMS-SP Rings

• OXC-based optical mesh

The reasons for choosing this particular set of network architectures were thefollowing:

- the CS-Ring was considered by the previous EURESCOM Project P615 as agood first step in introducing optical functionalities in SDH networks since thisarchitecture combines SDH functionalities of existing equipment (routing andlinear MS protection) with optical routing for logical node ordering.

- the OMS-SP Ring is a very advanced full-optical architecture, where both routingand protection are implemented optically. Therefore, it is expected that the OMS-SP Ring will bring specific problems that must be taken into account in theplanning of optical networks.

- the OXC-based mesh is seen as an advanced optical architecture, where theintroduction of optical routing and restoration might contribute to simplify thecomplexity of the electronic equipment in high-capacity mesh networks.

The hierarchical network configurations selected are seen as interesting possibilitiesfor the partitioning of real networks in more manageable domains. Theseconfigurations are expected to be useful in different network scenarios and will,possibly, be deployed at different stages in time, in the evolution from existingnetworks to networks based on the new optical network architectures.

In Table 5 possible combinations of these domains are collected. The investigation islimited to two-level architectures only because these architectures can be consideredto be a realistic solution for core networks. Complex optical network architecturesconsisting of these domains will have a performance dependent on the differentprotection schemes and re-routing strategies applied over individual sub-networks.Only the dual node interconnecting architectures are selected because in largecapacity core networks the disjoint alternative routing is an essential requirement. In



the following paragraph, we will provide a qualitative comparison since more preciseresults on network planning for the various architectures proposed here should beprovided in Deliverable D3.

Upper level CS-Ring OMS-SP Ring OMS-SP Ring Optical mesh

Lower level CS-Ring OMS-SP Ring Optical Mesh OMS-SP Ring

Selected architecture 1 2 4 3

Study P615 P709 P709 P709

Table 5: Selected reference architectures for comparison of two-level opticalnetworks

2.3.1 Two-level CS-Ring architecture

A two or more level (tier) CS-Ring architecture consists of hierarchicallyinterconnected CS-rings. Interconnection and cross-connections are carried out at theSDH client layer. The main advantages of this architecture are: Wavelength allocationcan be planned for each ring independently. Equipment is available now.

Figure 7: Two-level CS-Ring architecture

2.3.2 Two-level OMS-SP Ring architecture

The main advantage of this architecture is the optical connectivity between the rings.In the hubs the optical OCH level flexibility depends on the optical cross-connectcapability of the applied OADMs. First generation OADMs only have fixed add/dropcapability. In this case wavelength conversion could be necessary for theinterconnections.



Figure 8: Two-level OMS-SP Ring architecture

2.3.3 Two-level mesh-ring architecture

This is probably the most promising architecture for the future. Traffic in lowercapacity rings is collected and transported by the very high capacity upper level, likein a traditional SDH network. In case of large network sizes OXCs probably have tohave wavelength conversion functionality in order to establish large numbers ofoptical paths. Further study is necessary in order to develop suitable optical protectionand restoration mechanisms for this complex architecture. Analogously to similarSDH network architectures, the Optical Sub-network Connection Protection (OSNCP)(1+1 optical path protection) can be a candidate for this purpose.

Figure 9: Two-level Optical Mesh-Optical Ring architecture



2.3.4 Two-level ring-mesh architecture

In some special cases this architecture could be an optimal solution. In this case a veryhigh capacity optical ring is necessary for interconnection of meshed sub-networks onthe lower level. The protection and restoration problems are similar to the mesh-ringarchitecture.

Figure 10: Two-level Optical Ring-Optical Mesh architecture

2.3.5 Characteristics of the selected optical network architectures

The following table characterises the reference networks in terms of some of theparameters discussed above.

Reference network architectures

CharacteristicParameters

CS-Ring -CS-Ring

OMS-SPRing-OMS-SPRing

OMS -SPRing-Mesh

Mesh - OMS-SPRing

Selected architecture 1 2 4 3Generaltransparency no limited yes yesconnectivity full SDH

(VC-4/3/12)limited optical(OC)

optical(OC)

optical (OC)(VC-4-3-12)

restoration on opticallayer

no yes (limited) yes (limited) yes

flexibility SDX DXC limited good goodgranularity VC-4 wavelength wavelength wavelengthscaleability low low good

ArchitecturalHub equipment SDXC OADM OXC OXC (SDXC)No. Of hierarchicallevels

2(3...) 2(3...) 2 2

Flexibility on opticallayer

no limited limited yes

Types of subnetworks rings rings rings/OXCs/links

rings/OXC/links

Equipment availability now by 2000 by 2000 after 2000

Table 6: Some characteristic parameters of the reference network architectures



2.4 Identification of physical network parameters limitation

In this part of section 2, an attempt is made to identify physical network parameterslimitation. By this, we mean the limitations of real existing optical components andsubsystems, on realising complete aimed functionality, and its impact on opticalnetworking. These limitations can be grouped into two sets, regarding its origin:intrinsic limitations and technological immaturity limitations.

The first set comprises the limitations imposed by the optical nature of usedtechnology, which in spite of the degree of perfection of used components or systems,cannot be eliminated (theoretical limits). The second set groups the limitations mainlydue technological immaturity, which result in a non-ideal behaviour or thecomponents; these limitations are expected to be greatly reduced with theimprovement of technological aspects.

2.4.1 Identification of mechanisms originating limitations [4, 5]

Some effects, their respective causes and originated limitations are listed below:

2.4.1.1 Optical channel individual integrity

Crosstalk

Crosstalk in WDM systems arises due to filter/Demux imperfections and due to fibrenon-linear effects. Both can be kept small enough by appropriate system design sothat no significant penalties are expected. The following types of induced crosstalkwere identified:

Non-linear crosstalk: due to non-linear effects in the fibre; Four Wave Mixing FWMis the most relevant non-linear mechanism in the networks considered.

FWM generated cross-talk: depending on the optical power, the wavelength spacingof optical channels, on the fibre dispersion values, and on the transmission distance, itis a limitation on the number of channels for HD-WDM, as it becomes higher forchannels more closely spaced and for lower dispersion values.

SRS generated crosstalk: depending on the optical power, the wavelength spacing ofoptical channels, and on the transmission distance, it is a limitation for the number ofoptical channels and the range of the system.

Linear crosstalk: due to non-perfection of components such as filters and switches;can limit cascadability. Two types of linear crosstalk: inter-band crosstalk and intra-band crosstalk. Inter-band crosstalk is induced by other wavelength optical channelsdue to imperfect filtering. Intra-band crosstalk is due to the presence of residuallevels of optical power in other wavelengths of the used comb (non perfect filtering),which will add to the signal in those wavelengths. Thus it is a consequence ofswitching. Crosstalk induced by this process cannot afterwards be removed as it hasthe same wavelength of the optical signal to which it was added. This constitutestherefore a limitation to cascadability of nodes with crossconnect/ switchingfunctionality.

BER degradation:

Physical mechanisms that produce pulse broadening with travelled distance, lead toBER degradation, and limit system bandwidth and range. The higher the transmittedbit rate the higher importance these effects have.



Dispersion: the first cause of pulse broadening and consequent BER penalty, is fibrechromatic dispersion coefficient. This is due to the dependence of propagatingvelocity on the wavelength. For SMF, in the 3rd window, even for narrow linewidthoptical sources this effect becomes important.

PMD: Polarisation Mode Dispersion, results from the fact that due to non-perfectgeometry of the fibre and induced mechanical stress, the fibre presents birefringenceand therefore the two polarisation states have slightly different propagating velocities.This Birefringence varies randomly along the fibre. Special fibres as DCF and EDF(used in EDFAs) present higher values for PMD than SMF. So for long links usingDCF and EDFAs the total PMD can result in significant BER degradation.

SPM: this is a non-linear mechanism, due to the dependence of fibre refractive indexon the optical propagating field intensity. Intensity modulation results into refractiveindex modulation and therefore phase (and frequency) modulation of the opticalsignal occurs. This causes linewidth broadening and thus BER degradation.

2.4.1.2 Network element cascadability

Optical Amplifier OA

Optical amplifier cascadability limitation due to optical amplifier generated noise; thisnoise accumulates with the number of transverse optical amplifiers therefore limitingcascadability of OAs. Optical amplifier cascadability limitation due to non-flatness ofoptical amplifier gains, which causes different amplification of individual opticalchannel power. In case of a change in the number of channels passing through theOA, power transients may cause degradation.

OXCs, optical Switches

Cascadability limited due to optical power losses, intra-band linear crosstalk andbandwidth reduction by filtering.

Optical Mux/Demux

Cascadability limited due to optical power losses and bandwidth reduction by filtering

2.4.2 Identification of systems/components which introduce limitations

Various elements contribute to the physical limitations mentioned above. A relationbetween the physical nature of the introduced degradation, depending on thecomponent, and the impact on the optical transmission is established in the followinglist.

Optical Fibre

Crosstalk: Non-linear effects, in particular FWM

System bandwidth limitation: Dispersion

BER degradation: Attenuation, Non-linear effects, in particular FWM

OXC, Optical Switches

BER degradation: Insertion losses

Linear crosstalk: optical filtering, optical switching

System bandwidth narrowing: optical filtering; filter misalignment; wavelengthinstability



Optical Filters , Optical Mux/demux

BER degradation: Insertion losses

Linear crosstalk: non-ideal optical filtering

System bandwidth narrowing: optical filtering; wavelength instability

Optical Sources

System bandwidth narrowing: Line width, Wavelength accuracy, Wavelengthstability, Wavelength tuning accuracy (if tuneable sources)

BER degradation: optical power efficiency coupling

Optical inline amplifiers

BER degradation: amplified spontaneous emission noise

BER degradation of less amplified channels: Different channel poweramplification due to gain curve (non–flat)

Limited cascadability: build-up of optical signal degradation as a result of abovementioned mechanisms, with number of transverse elements.

2.4.3 Processes to overcome limitations at present and solve them in the future

Most of these limiting mechanisms arise from the use of high optical power levels andWDM signals; i.e., already known and existing, their influence only became importantin the present context. Some of them cannot be eliminated (however they can becompensated) such as dispersion effect, attenuation, non-linear effects. Others, withtechological improvements on the manufacturing processes of the components, can begreatly reduced, almost eliminated, such as linear crosstalk due to filtering andswitching, maintaining system bandwidth, by improving wavelength accuracy,stability, tuning, filtering and misalignment etc.

2.5 Identification of Ranges of values

In this point, the number of parameters related to the layers of the optical channel, theoptical multiplex section and the optical transport section in WDM networks arespecified. These values are necessary in the planning process of such networks.Optical layer performance of today’s commercially available WDM systems forterrestrial optical networks, however, still keeps improving. In order to take intoaccount the expected technical progress of optical WDM transmission systems asmuch as possible there are up to three values specified for various parameters. Thefirst of these values represents presently available commercial WDM systems. Thesecond parameter value has been derived from manufacturers’ publications oradvertisements regarding product development in the near future. Whenever possiblea third value is presented that reflects results from laboratory experiments ortheoretical calculations and which may indicate what will be achieved in futuresystems.



2.5.1 Functional layer characteristics

According to Rec. G.otn an optical network will consist of the following three opticallayers:

• Optical Channel (OCH) layer

• Optical Multiplex Section (OMS) layer

• Optical Transmission Section (OTS) layer

These functional layers can be characterised in terms of the following parameters:

Layer CharacteristicsOCH Transparency

Maximum number of cascaded opticalnodesOptical cross-connect capacityOptical protection

OMS Number of available wavelengthsWavelength conversionOptical protection

OTS Power budgetDispersion budgetAccumulated noiseCross-talkNumber of cascaded optical amplifiers

Table 7: Characteristics of optical layers

2.5.2 Ranges of values

Values reflecting limitations of present implementation of optical functionality and itsimpact on optical networking, either obtained by calculation, simulation orexperimentation, will be presented in this section.

Power budget

The achievable power budget depends mainly on the optical output power of theEDFAs and on the number of wavelengths used in the network as well as on themaximum number of EDFAs cascaded in any transparent link, which may happen tocome into existence due to optical switching, configuration, or restoration processeswithin the network. Power budget of present systems is limited to a value of 30 dB ofattenuation between EDFAs. This value is optimised with respect to a dispersionlimited system (no dispersion accommodation) at a bit rate of 2.5 Gbit/s on standardsingle mode fibre resulting in a total link length of 500 km to 600 km whichcorresponds to 150 dB of optical power budget. If a larger transparent total link lengthis desired the power budget between neighbouring EDFAs must be reduced.Dispersion accommodation necessary in this case introduces additional opticalattenuation.

Dispersion budget

If standard single mode fibres (G.652) are to be used without dispersionaccommodation (DA) techniques the dispersion budget is limited to values of about12000 ps/nm for 2.5Gbit/s transmission. 10 Gbit/s transmission without DA is too



severely limited in length to be attractive. With DA methods the limitations indispersion budget are relaxed considerably. A values of 75000 ps/nm isexperimentally estimated.

Accumulated noise

The amount of accumulated noise depends on the system design. With a transmissionlink design, using a shorter span length, that takes into account the networkrequirements the accumulated noise effects can be kept small enough so that nosignificant penalties are to be expected

Number of cascaded OAs

Available WDM point-to-point systems are designed to use a maximum number of upto 6 EDFAs in cascade as outlined above. It is possible to increase this numberconsiderably without running into ASE noise accumulation problems if the opticalbudget between succeeding EDFAs is reduced. Transmission of 8x5 Gbit/s over atotal distance of 4500 km of standard single mode fibre using dispersionaccommodation technique has been demonstrated. With dispersion shifted fibre 20x5Gbit/s over 9100 km have been achieved. With soliton techniques transmission of8x10 Gbit/s over 10000 km was shown using more than 300 cascaded EDFAs.Parameters values for OCH, OMS and OTS sections are summarised in table 8.

parameter availableperformance

announcedperformance

limits

Optical channel (OCH) layer characteristicstransparencynumber of optical nodes in cascade 6-7number of wavelength conversions 1-2 5-10Optical Multiplex Section (OMS) layer

number of available wavelengths 16-32 80-100 <200optical cross-connect capacity - 16x16 128x128protection methods ? ? ?supervision channel ? ? ?Optical Transmission Section (OTS) layerpower budget (for link between consecutiveEDFAs) in EDFA cascade

<30dB <30dB <30dB

number of cascaded EDFAs (2.5Gbit/s client) 4-6 Min. 30 at reducedpower budget of 20dB, 300 possible

number of cascaded EDFAs (10Gbit/s client)G.652

4-6 Min. 15 at reducedpower budget of 20dB, 300 possible

dispersion budget (2.5Gbit/s client) nodispersion accommodation (DA)

12000 ps/nm

dispersion budget (10Gbit/s client) DA req. DA req. DA req.accumulated noise ?cross-talkInter-bandIntra-band

2525

3530

5035

Table 8: Parameter values for OCH, OMS and OTS layers



2.6 Conclusions

In this Section, network configurations have been proposed, consisting ofinterconnected basic topologies, either in flat or hierarchical arrangements. Fourreference network architectures have been selected, which will be considered furtherin Task 3 and Task 4 of the Project. Various parameters have been presented, relatedto topology, physical limitations, demands architecture and survivability. Theirrelationship with the network planning and network engineering processes has beenmentioned. A particular attempt has been made to identify physical parameterslimitation and ranges of value for power and dispersion budget, accumulated noiseand number of cascaded optical amplifiers.

However the application of the described architecture and the study of networkplanning problems should also consider the client layers. The next Section isdedicated to the potential of the optical layer for routing various formats, which couldbe promising, when considering the actual growth of different client signals such asATM and IP.



3 Potential of WDM routing for different client signals

The objective of this part is to describe the various client signals, which could becarried in the optical networks. In order to pinpoint the possibilities to plan an opticalnetwork using a non SDH client signal, the requirements of client layers is analysed.

Recent development of ATM and IP physical interfaces could have an impact on thefuture optical network design. After a brief investigation on ATM and IP clientsignals performance and functionalities, multi-layer network configurations areproposed using IP, ATM, SDH and WDM characteristics. A first evaluation of suchconfigurations is discussed.

3.1 ATM client signal

3.1.1 ATM Network functionalities and physical layer

ATM is a connection oriented technology with end-to-end QoS supporting multipleservice. Main advantage of ATM are:

• fast switching

• high-speed interfaces

• efficient use of bandwidth through statistical multiplexing at VP and VC levels

• ATM network functions are : VP or VP-VC switch, VP or VC Cross-Connectand ATM Multiplexer

3.1.2 ATM Services

An ATM Service Category is intended to represent a class of ATM connections thathave homogeneous characteristics in terms of traffic pattern, QoS requirements andpossible use of control mechanisms, making it suitable for a given type of resourceallocation. A first classification of these services/capabilities may be seen from anetwork resource allocation viewpoint. We can identify (ATM Forum), Constant BitRate (CBR), Variable Bit Rate (VBR), Available Bit Rate (ABR) and Unspecified BitRate (UBR) categories.

3.1.3 ATM Performance Parameters

The ATM performance parameters are defined in ITU-T recommendation I.356, andthey are:

• Cell Error Ratio (CER)

• Cell Loss Ratio (CLR)

• Cell Mis-insertion Rate (CMR)

• Mean Cell Transfer Delay (MCTD)

• Cell Delay Variation (CDV).

It is necessary to identify the network degradation sources that have a direct orindirect impact on the ATM performance parameters. These sources have a global



effect on all services that are supported by the network. Table 9 shows the impact ofthe network degradation sources on the ATM performance parameters.

Item CER CLR CMR MCTD CDV

Propagation delay ä

Statistical physical errors ä ä ä

Switching architecture ä ä ä

Buffers capacity ä ä ä

Number of nodes between ends ä ä ä ä ä

Traffic load ä ä ä ä

Resources allocation ä ä ä

Faults ä ä

Table 9: Impact of degradation sources on the ATM parameters

Transmission degradation sources have been highlighted

The referred network degradation sources can be grouped into two main areas: ATMdegradation sources and Transmission degradation sources. The ATM degradationsources are directly dependent on the ATM technology and equipment. Thetransmission degradation sources are directly dependent on the physical medium ortechnology used to transport an ATM signal. In this section only the transmissiondegradation sources will be analysed. They are: Propagation delay, Statistical physicalerrors and Faults.

3.2 IP client signal

3.2.1 Internet network layers and services

The Internet offers global connectivity for data communications between differentcomputers networks. As shown in Figure 11, the Internet is based on a four-layerprotocol. The lowest layer is called the link layer which contains protocols for LANsand sub-network communications such as Serial Line IP and Point-to-point ProtocolPPP [6], [7] and [8].



Application layerSMTP, FTP, SNMP, HTTP,TELNET, MIME, NNTP

Transport LayerTCP, UDP

Internet layerIPv4, IPv6 IP ng

Link layerLAN-Ethernet WAN-PPPX.25, ISDN, Frame Relay ..

• SMTP :Simple Mail Transport Protocol• FTP : File Transport Protocol• SNMP: Simple Network Management Protocol• HTTP : Hypertext Transfer Protocol• TELNET : Remote Terminal Protocol• NNTP : Net News Transfer Protocol• MIME: Multipurpose Internet Mail Extensions

Figure 11: IP network model and services

PPP was designed as a standard method of transmitting IP datagrams over point-to-point links. Initial deployment intended to govern point-to-point link over short locallines, leased lines, ISDN, and old telephone service POTS using standard dial-upmodems.

The second layer is the Internet Layer (IP-v4/v6 RTP, RSVP...) which providesrouting and relaying between LANs and sub-networks. The next layer, TCP-UDPtransport layer, controls end-to-end communication links between different Internetsystems. TCP is used to provide a reliable transportation using end-to-end control. Itis also includes error control detection and recovery for the transmitted packetswithin the Internet network. For short message applications another protocol, the UserDatagram Protocol (UDP), is used. UDP is also connectionless oriented but withoutany overhead for establishing end-to-end control. Finally, application layer providesInternet applications like e-mail, WWW, FTP, Telenet.

3.2.2 IP protocols: IP v4/v6, RTP and RSVP

IPv4 defines datagram format, addressing routing and error reporting. It supports 32bit address space. IPv4 can only offer connectionless, Best Effort packet delivery anddoes not support end-to-end QoS. IPv6 introduces a new modular datagram formatwith the following advantage :

• Flow label: the header includes a flow label. Flow values can be assigned toparticular streams of traffic with special QoS requirements.

• Large network addressing: IPv6 supports 128 bit network addresses.

• Security: IPv6 requires and supports both authentication and confidentiality.

Real Time Protocol RTP is designed especially to support a new type of real timetraffic. The RTP architecture, called application layer framing, replaces TCP with asimple framework for a direct application. For example, audio algorithms can toleratemissing data much better than can lengthy delays. Instead of introducing delays withre-transmissions, RTP applications prefer the transport layer to simply forget aboutmissing data. Resource reservation protocol RSVP is a signalling protocol for IPallowing applications to make network resource reservations for data flow. It can beused for multi-cast application in IP network. The deployment of RSVP in IP networkcan support end-to-end QoS.



3.3 Network configurations required by ATM/IP client signals

First, the main Optical network functionalities considered are:

-Optical Multiplexing / De-multiplexing OMux/Demux

-Optical Cross Connect OXC

-Optical Add/Drop Multiplexing OADM

-Optical Protection Switching OPS

IP v4/v6

•RoutingATM VC/VP

• Switching

• XC

SDH LOP/HOP

• XC• ADM• APS

WDM•OXC•OADM•OPS

Figure 12: IP / ATM / SDH & WDM network functionalities

Figure 12 illustrates network functionalities for WDM optical transport network aswell as for SDH, ATM and IP network. It is clear from this figure that the mainrouting and switching functions are for IP and ATM respectively, but cross-connecting; Add/Drop and protection functions can be obtained using both SDH andWDM technologies. The main difference between SDH and WDM remains theprocessing of SDH in time domain and for WDM in frequency domain. Taking intoaccount the network functionalities of the different client signals the followinglayering structure is proposed for the 2003-2005 period (Figure 13). Differentconfigurations of network architecture are proposed in the next part of the document.Inter-working of IP, ATM, SDH and WDM will be discussed later.



IP v4/v6

PPP(HDLC) ATM VC / VP(Cell Based)

SDH HOP/LOP PDH

WDM Optical NetworkOMS/OCH

?

Figure 13: IP/ATM/SDH and WDM network mapping for 2003-2005

Based on the above network functionalities, 5 configurations were selected to studythe impact of an ATM or IP client signal on the planning of an optical network. Theconfigurations are the following :

• ATM over SDH over WDM

• ATM over WDM

• IP over ATM over SDH over WDM

• IP over SDH over WDM

• IP over WDM

For each configuration 2 cases are presented, according to the implementation ofprotection and routing in the WDM optical network.

3.4 Impact of non SDH client signal on the planning of an opticalnetwork

3.4.1 ATM over SDH over WDM : SDH protection vs. WDM protection

Under normal operation only two degradation sources are active: the propagationdelay and the statistical physical errors. In what concerns the propagation delay, thisis constant and will not vary unless one fault occurs. The statistical physical errors aredirectly related to the physical medium and technology used. In the WDM case, lowervalues for the BER (less than 10E-10) are expected. Another case is the situationwhen a connection problem occurs. The problems could be grouped into three mainareas:

• connection degradation (problems in the transmission cable)

• connection breaks (cable cut-off)

• terminal equipment problems (faults).

In the case of the connection degradation and soft terminal equipment problems,several ATM parameters will be affected, and in that way the several services



supported by the connection. Depending on the degradation factor the services couldbe smoothly or strongly affected. This depends on the CLR and CER values and theirimpact on the services. There are services where low CLR and CER are required andothers where higher values, of the order of 10E-3, can be tolerated for some period oftime. One example of each is the video and the voice service, respectively.Connection breaks and strong terminal equipment problems will severely affect theconnections and render the normal way of signal transmission impossible. In thesesituations the ATM services supported will be affected. The connections will be lost.Within a fault condition several situations could occur. Normally the re-routing andprotection schemes of the network will be activated in order to establish onealternative connection between affected end points. The services and the ATMperformance parameters will be affected. Several ATM cells are lost and others willbe delayed. Another MCTD is established. After the establishment of the alternativeconnection we will have a situation similar to the normal operation. If the fault affectsone end of the system, in what concerns the ATM service, the service will be affectedpermanently until the fault is cleared.

So, taking into account what was expressed before, the re-routing or protectionscheme used in the transmission technology, such as the WDM technology, will havea strong impact on the ATM performance. Several protection schemes should bestudied in order to minimise the impact on the ATM performance, in case of WDMlink failure. This configuration study should be performed taking into account twodifferent scenarios. The first one considers the WDM as a means of transmission andall the protection is provided by the SDH structure. The second considers that all theprotection is provided by the WDM. The SDH is used as traffic encapsulation.

The influence of such scenarios should be analysed taking into account the ATMsignal characteristics. For that, the most critical ATM performance parameters CER,CLR, CTD and CDV are considered. In order to perform this study higher levelsignals such as video and audio signals in interactive applications, are considered.From the user point of view these are the most critical ones. Data applications, such asfile transfer, can be affected by the degradation of the transmission signal for somemilliseconds. The user does not sense this degradation or problem most of the time.One of the most critical points is the switching time or protection action when afailure occurs. As a reference the «APS protocols for OMS SP Ring network»document issued by the EURESCOM P615 Project is used. The main results aresummarised in table 10.

Example MS-SPRing OMS-SPRing

Node failure 4.325 ms 7.325 ms

Bi-directional signal fail 5.190 ms 8.190 ms

Unidirectional signal fail 5.810 ms 8.810 ms

Unidirectional signal degrade 7.655 ms 13.655 ms

Table 10: Protection switching times for 7-node rings with 51 km-long spans

In the above table the worst cases are used. This means that for SDH and WDM timesof 8ms and 14ms will be used respectively as reference times for protection actions.Considering higher level signals, such as video signals, their QoS requirements can bemapped into ATM performance parameters as shown in table 11.



ATM Performance Parameters Value for High video quality Value for Low video quality

CTD 4 ms (99% of time) 400 ms

CDV 500 µs (99% of time) 130 ms

CLR < 1.7 E-9 < 3 E-7

CER --- < 4 E-6

Table 11: Video performance parameters

Also for video signals one question to be considered is the number of transmittedframes. For normal transmission rates one frame is transmitted each 33 ms.

3.4.1.1 Protection at SDH level

According to table 10 the protection scheme takes about 8 ms to reroute the traffic.Each VC-4 takes 125 µs at 155 Mbps (STM-1). So dividing 8 ms (the rerouting time)by 125 µs (the time necessary to transfer one VC-4) we get 64. In each VC-4 we cantransfer 44151 cells. So, multiplying 44151 by 64 we get 2826 cells. ConcerningATM at full SDH link speed, 2826 cells are lost.

3.4.1.2 Protection at WDM level

According to Table 10 the protection scheme takes about 14ms to re-route the traffic.This means that with STM-1 link 112 VC-4 are lost. Concerning ATM at full SDHlink speed, 4945cells are lost. The several ATM services supported by the SDH linkwill be affected. The impact on the higher level service (such as video-conference)depends on fail recovery mechanisms or applications used, as well as on the amountof information lost. As an example, for a video-conference signal transmitted at 30frames/s a fail will affect one or two frames. The loss of one or two frames does notimpact video services. In conclusion, the impact of SDH on the performance is notcritical for this service.

3.4.2 Configuration ATM over WDM

The recent standardisation of ATM cell based interface leads to a direct connectionbetween the ATM VP/VC and the Optical Channel OCH of the WDM opticalnetwork. At present, ATM cell based is proposed for UNI interfaces with 155, 622Mbit/s bit rate for the ITU standard and up to 2.5 Gbit/s for the ATM Forum.

A first scenario of introducing WDM switching layer in an ATM network is the use ofWDM ring architectures (Figure 14). In this configuration, the VP cross-connects aredirectly interfaced with OADM nodes. When a failure occurs in an ATM network, theVP connections are re-routed over the OCH network without the need for anyelectrical back up VPs. The vacant bandwidth can be used for other services. The useof WDM ring architectures provides a high capacity self-healing network for the VPservices.

A second scenario is also proposed. The WDM switching layer is composed of OXCnodes connected to VP cross-connect. The OXCs create a flexible and highly securesystem needed for CBR paths in an ATM backbone network with a data rate from 155Mbit/s to 2.5 Gbit/s.



_

Figure 14: WDM/ATM network scenarios

Considering the values of table 10 for the worst case protection switching time, theprotection scheme takes about 14ms to perform the traffic protection.

3.4.3 Configuration IP over ATM [9]

Different approaches to running IP over ATM are considered for performingconnection-less IP and connection oriented ATM technologies. Classical IP defines IPto ATM address resolution and IP packet encapsulation. It allows the use of directATM connections inside a sub-network, but shares the VC connection with all otherapplications running between the same transmitter and receiver.

• IP encapsulation over ATM Adaptation Layer 5 (ALL5): two methods forcarrying IP over AAL5 are proposed. The first method, called Logical LinkControl (LLC) Encapsulation, allows multiplexing of multiple protocols over asingle VC, whereas the second method, called VC based multiplexing, assumesthat each protocol is carried on a separate VC.

• IP Switching: IP switching provides high-speed IP routing. It combines the high-speed layer 2 switching of ATM with standard IP routing, establishing direct VCconnection whenever a long IP datagram flow is detected by the router (Figure15). The advantage of an ATM switch is that the QoS and multicast are available.The disadvantage of an IP switch is that it is cell-oriented, not packet-oriented,and connection-oriented, unlike the connection-less IP network protocols .

IP Flowcontrol

IPRouter

ATMVC Switch

Figure 15: IP switching



3.4.4 Configuration IP over SDH [10, 11, 12]

SDH is a serious candidate to interconnect high capacity IP routers. IP over SDH is anefficient approach compared to IP over ATM. The additional overheads required forIP over SDH is about 3% compared to 18-25% for IP over ATM. IP The PPPencapsulated IP datagrams are framed into SDH Virtual Container VC (Figure 16).The HDLC provides the delineation of PPP encapsulated IP datagrams [13]. IPdatagram size can be can be in theory up to 64 Kbytes, but for most individualInternet users 1500 bytes is the maximum length which is determined by the Ethernetaccess network. Scrambling of HDLC framed signals is required to provide securityagainst emulation of the SDH set-rest scramble pattern and replication of the STM-Nframe alignment word.

IP• Client datagrams, IPv4, IPv6..

PPP• IP multiprotocol encapsulation• Error checking• Link initialisation control

HDLC• PPP packet delineation

SDH Path• STM-N, VC4-Nc

Figure 16: IP-PPP over SDH layers

IP datagram can be directly encapsulated over SDH, without PPP interface, usingHDLC frame switches. This configuration provides multiple access,broadcast/multicast-capable switched LAN and WAN environment using essentiallypoint-to-point SDH lines transmission media.

3.4.5 Configuration IP over WDM

The importance of running IP over WDM is a new issue for the all-optical network.At the moment there is no standard for direct working of IP over the Optical Channelsof a WDM network. To solve this problem two items are identified:

• encapsulation mechanism to carry IP payloads over OCH, overheads, errorcontrol...

• procedures and protocols between IP routers and WDM network elements OXC,OADM.., for overall network routing and protection mechanism.

For the IP over OCH configuration, various all-optical network architectures,presented in section 2, such as point-to-point, interconnected rings and ring-meshconfigurations, should be considered.

A typical example of inter-working between the IP and the WDM layer is thesignalling protocol for the optical protection mechanism. The signalling protocol mustbe implemented between the IP and WDM layers to provide a fast reconfiguration ofthe optical protection layer or switches (Fig.17). The most important parameter of thesignalling protocol is the connection set-up time, the time from when the host driversends a request, until the configuration from the switch is received. Connection set-up



time depends on signalling transport network. In WDM network the signallingprotocol could be transported over the optical supervisory channel .

TCP , UDP

IP

Signalling Control

OPTICAL SWITCHES

Fig17: Signalling protocol layers for IP/WDM using optical protectionmechanism

3.5 Conclusion

This last section has been focused on analysing an alternative solution to SDH/WDMbased optical networks. First, an overview of ATM and IP client signals (networkfunctions, services and performance) has been presented, focusing on IP, ATM, SDHand WDM optical network functionalities. The following multi-layer networkarchitectures have been selected to study the impact of ATM and IP client signals onthe planning of optical network. The following configurations have been identified:


• ATM over WDM



• IP over WDM

It was concluded that ATM and IP client signals requirements on the performance andthe management of WDM optical network are not the same as SDH client signal.WDM protection mechanism, reconfiguration time and network architectures shouldbe reconsidered for ATM or IP client signals. For the IP over WDM scenario thefollowing topics are identified for a future EURESCOM Project:

• Encapsulation of IP over the Optical Channel

• Signalling protocol for the layer inter-working

• Management and OAM issues.



4 Conclusion

In this Deliverable, results from Task 2 ‘Considerations on Optical NetworkArchitectures’ activities are presented.

Most of the important optical functions used for simple network topologies areconsidered to be available by 1998. A list of desirable functions enabling orenhancing optical networking, have been identified as follows:

• Flexible OADMs and OXCs

• Optical signal monitoring functions (QoS, optical spectrum, and Failuredetection)

• Optical 3R regeneration

• Network survivability (protection, restoration)

• Management functions

From the assessment of optical network architectures, four complex networkarchitectures have been selected as reference configurations. The selectedconfigurations are listed as follows:

• two-level CS-Ring architecture

• two-level OMS-SP Ring architecture

• two-level mesh-ring architecture

• two-level ring- mesh architecture

Characteristic parameters dealing with architectures, topologies, demands andphysical limitation have been identified and ranges of value have been proposed.

The selected configurations will be used for network planning studies in Tasks 3 and4 activities of the Project.

In the last Section of the Deliverable, the potential of optical transport network formulti-client applications is discussed. Multi-layer network configurations areproposed using IP, ATM, SDH and WDM network functionalities. The followingconfigurations are proposed:


• ATM over WDM



• IP over WDM

The cumulative configurations using many layers like ATM over SDH over WDMand IP over ATM over SDH over WDM are considered to be an inefficient solutionfor their resource consuming overhead requirements. Unfortunately, directly basedATM over WDM and IP over WDM approaches require the definition of a newnetwork interface and corresponding management system.



References

[1] Katsusuke Tajima, «Low loss optical fibres realised by reduction of Raleighscattering loss», OFC ’98 Technical Digest, pp 305-306.

[2] Yadlowsky, M.J., Deliso, E.M., Da Silva, V.L., «Optical fibres andAmplifiers for WDM Systems», Proc. of the IEEE, Vol. 85, No. 11, Nov.1997

[3] D.R.Hjelme, A.Royset, D.J.Slagsvold, ‘How many wavelengths does it take tobuild a wavelength-routed optical network ’, ECOC ’96, Sept.1996

[4] Demeester,P. et al, “Photonic transport networks : Why, How and When ? ”,NOC’96, Vol.I pp. 158 –165.

[5] Schiess, M. et al, “ Guidelines for the introduction of Optical NetworkingFunctions”, EURESCOM Project P615, D3, May 98.

[6] Peter Newman et al, ‘IP switching and Gigabit routers’, IEEE communicationmagazine January 1997, pp 64-69.

[7] Gurdeep Singh Hura, ‘The Internet: global information superhighway for thefuture’, Computer Communication 2,1998, pp 1412-1430.

[8] Reuven Cohen et al, ‘Using proxies to enhance TCP performance over hybridcoaxial networks’, Computer Communication 2,1998, pp 1502-1518.

[9] RFC 1483, ‘Multiprotocol encapsulation over ATM ALL 5’,http://www.ietf.org/

[10] C.J. den Hollander, ‘The mapping of HDLC payloads into SDH VirtualContainers’, ITU-T, COM 15-61, December 1997.

[11] RFC 1619, ‘PPP over SONET/SDH’, http://www.ietf.org/

[12] RFC 2171, ‘MAPOS’, http://www.ietf.org/

[13] RFC 1661, ‘Point-to-point Protocol PPP’, http://www.ietf.org/



Appendix 1: Recent Progress in the Performance of OpticalTransmission System Components

Introduction

Even though the state-of-the-art of available optical functions was reviewed at thebeginning of the Project the progress achieved in the performance of some of thefunctions has been considerable since then and it seems appropriate to take that intoaccount.

EDFA properties

The absolute values of EDFA key properties power gain, saturated output power andnoise figure of EDFAs have not changed dramatically. The variations of theseproperties as a function of optical carrier frequency seriously impacts theperformance of WDM transmission systems. These limit e.g. the maximum number ofchannels and the total number of cascaded amplifiers in a WDM link.

In recent work it has been demonstrated that the variation of these parameters as afunction of optical carrier frequency can be reduced considerably and correspondingimprovements in system performance can be achieved. Within the entire gainbandwidth of an EDFA the deviation of actual gain from the nominal gain is specifiede.g. in terms of a dB value as it is often undesirable to take into account the moredetailed information of the optical gain spectrum. The diagram in Fig. a-1 represents again spectrum of a conventional type EDFA. In contrast to that the gain spectrum, Fig.a-2 shows the considerably enlarged gain range and correspondingly lower gainvariation within a certain wavelength band that has been achieved in [2].

Figure a-1 : Gain spectrum of a conventional EDFA Figure a-2: Wide bandEDFA gain spectrum [2]

Temporal behaviour of EDFAs is characterised by very slow gain relaxation effects[1], [3], [4]. Usually these relaxations do not interfere with the very rapid powervariations of the transmitted optical signals. Under special circumstances e.g. if someof the WDM channels fail, it is, however, possible that distinct power transients in thesurviving channels are produced. In WDM links with a large number of cascadedEDFAs these gain relaxations will be even more pronounced. It has been shown [4]that in a cascade of conventional EDFAs these gain relaxations can induce powertransients of nearly 0dBm in the surviving channel under extraordinary conditions.Under the same conditions in a cascade of EDFAs with an advanced gain controlscheme, however, the power transient was suppressed by 10dB.



In today’s WDM point-to-point systems with only SDH clients these gain relaxationsare not likely to pose any problems. For a future network which will extensively makeuse of flexible OADMs, OXCs and optical packet switching appropriate EDFAs withimproved relaxation properties are necessary [5].

In table a-1 values for some of the improved EDFA properties are shown. Theseimproved values have not yet been achieved in one single device.

conventional EDFA features: advanced EDFA features

gain band: 1,54µm.1,565µm

1,54µm ... 1,565µm gain band: 1,528µm ...1,611µm

gain bandwidth: 25nm gain bandwidth: 83nm

in band gain variation: 3dB in band gain variation: 3dB

no. of channels: 32 (100GHz spacing) no. of channels: 100 (100GHz spacing)

200 (50GHz spacing)

overshoot: 0dBm overshoot: -10dBm

Table a-1 : Properties of conventional and advanced type of EDFA

Conclusion

Due to strong competition among the different manufacturers of optical componentsand transmission systems rapid progress in the performance of some devices(according to desired and necessary properties) can be observed.

References

[1] M.F. Krol, Y. Liu, J.J. Watkins, ’Gain variations in optically gain clampedErbium doped fibre amplifiers‘,Proc. ECOC’98, pp 43-44

[2] Y. Sun, J.W. Sulhoff, A.K. Srivastava, A. Abramov, T.A. Strasser, P.F.Wysocki, J.R. Pedrazzani, J.B. Judkins, R.P. Espindola, C. Wolf, J.L Zyskind,A.M. Vengsarkar, J. Zhou, 'A Gain-Flattened Ultra Wide Band EDFA forHigh Capacity WDM Optical Communication Systems', Proc. ECOC’98, pp.53-54

[3] S.H. Lee, S.H. Kim, 'Performance of all Optical Gain-Clamped EDFA in 8Channel x 10Gbps WDM Using Stimulated Brillouin Scattering', Proc.ECOC’98, pp. 47-48

[4] M. Karasek, J.C. van der Plaats, 'Modelling of a pump power loss controlledgain locking system for EDFA application in WDM transmission systems‘,IEE Proc.-Optoelectron., Vol. 145, No. 4, August 1998, pp. 205-210

[5] S.Y. Park, H.K. Kim, S.M. Kang, G.Y. Lyu, H.J. Lee, J.H. Lee, S.Y. Shin, 'Again-flattened two-stage EDFA for WDM optical networks with a fast linkcontrol channel', Optics Communications, 153, (1998), pp.23-26

dwdm planning

Documents