IP over WDMpaginas.fe.up.pt/~hsalgado/pstfc/Projecto_IP_WDM_Final.pdfFigure 15 - Mach-Zehnder Interferometer; b) Three Mach-Zehnder Chain .....22 Figure 16 - Basic acousto-optic tunable

INESC Porto, July 2002

Licenciatura em Engenharia Electrotécnica e de Computadores Ramo de Telecomunicações, Electrónica e Computadores

Graduation project – DEEC

IP over WDM

Designing an Optical IP Router

Supervisors Students

Henrique Salgado, PhD Manuel Ricardo, PhD

Bruno Leite Fernando Pinto Igor Terroso Joel Carvalho

Acknowledgements

We would like to thank some people,

without which, this work would have been difficult

to accomplish:

Orlando Frazão

Filipe Sousa

Our families

Our friends

Abstract

Expanding Internet-based services are driving the need for evermore bandwidth in the

network backbone. Besides that communications needs are continuously growing.

Communication networks must evolve to sustain this growing need for bandwidth.

Today, the only technology that can effectively meet such a demand is WDM. This technology

allows the increment of the available bandwidth in the fiber by the use of different wavelengths.

At present, this technology is used in a point-to-point manner in the network core using heavy

multi-layered protocol stacks.

In parallel the Internet transport infrastructure is moving towards a model of high-speed

routers interconnected by optical core networks.

Large-scale efforts are being made to develop standards and products that will eliminate

some of those issues: reduce one or more of the intermediate layers of the protocol stack, allow

automatic routing and provisioning inside the transport network, an allow fast and cheap optical

switching.

Development of these networks demands the use of new optical nodes. Nodes based in

new high performance optical technologies that are able to control the communications channels

and route them correctly. The purpose of this project is to build one of such nodes.

This reports covers the development of an Optical Router, implementing a control plane,

similar to the ones in the IP world, as defined in GMPLS research (IETF). The work can be split

into two parts:

Part I – Development of the underlying optical fabrics (Optical cross connect – OXC).

Part II – Definition of the structure of the protocol stack, needed to establish the control

plane at the OXC.

Different groups in different research units develop each of these parts. This report covers

the work done in part II.

PART I

INESC - UOSE Porto, July 2002



IP over WDM


Part I

Supervisors Students

Henrique Salgado, PhD Manuel Ricardo, PhD

Igor Terroso Joel Carvalho

2 of 90 - IP over WDM I

Table of contents

IP over WDM I - 3 of 90

Table of contents Table of figures................................................................................................................................5 Chapter 1 - Introduction ..................................................................................................................7

1.1. Abstract .................................................................................................................................................................7 1.2. Background and Motivation .................................................................................................................................7 1.3. Report Organization..............................................................................................................................................7 1.4. Contributors...........................................................................................................................................................8

Chapter 2 - Optical Networks..........................................................................................................9 2.1. Introduction ...........................................................................................................................................................9 2.2. The three networks generations ............................................................................................................................9 2.3. OTDM .................................................................................................................................................................10 2.4. OCDM.................................................................................................................................................................11 2.5. WDM...................................................................................................................................................................11

2.5.1. WDM Link ..................................................................................................................................................13 2.5.2. Passive Optical Network (PON).................................................................................................................13 2.5.3. Broadcast and Select Networks ..................................................................................................................14

2.5.3.1. Single-Hop Networks.....................................................................................................................................................................14 2.5.3.2. Multihop Networks ........................................................................................................................................................................15

2.5.4. Wavelength Routing Networks...................................................................................................................15 2.6. DWDM................................................................................................................................................................16 2.7. Summary .............................................................................................................................................................16

Chapter 3 - Optical WDM Components ........................................................................................17 3.1. Introduction .........................................................................................................................................................17 3.2. Optical passive components................................................................................................................................17

3.2.1. Fiber.............................................................................................................................................................17 3.2.2. Couplers.......................................................................................................................................................18 3.2.3. Isolators and Circulators .............................................................................................................................19 3.2.4. Filters...........................................................................................................................................................19

3.2.4.1. Tunable 2 x 2 directional couplers.................................................................................................................................................20 3.2.4.2. Gratings ..........................................................................................................................................................................................21 3.2.4.3. Arrayed waveguide grating (AWG) ..............................................................................................................................................21 3.2.4.4. Fabry-Perot Tunable Filters (FPF).................................................................................................................................................21 3.2.4.5. Mach-Zehnder Tunable Filters (MZF) ..........................................................................................................................................22 3.2.4.6. Liquid-crystal Fabry-Perot filters ..................................................................................................................................................23 3.2.4.7. Acousto-optic tunable filters..........................................................................................................................................................23

3.2.5. Multiplexers ................................................................................................................................................23 3.2.5.1. WDM Multiplexers and Demultiplexers .......................................................................................................................................23 3.2.5.2. Add/Drop Multiplexers (OADMs) ................................................................................................................................................24 3.2.5.3. Optical Cross-Connects (OXCs)....................................................................................................................................................24

3.3. Optical active components..................................................................................................................................24 3.3.1. Amplifiers....................................................................................................................................................24 3.3.2. Erbium-doped fiber amplifiers (EDFAs)....................................................................................................24

3.3.2.1. Semiconductor Optical Amplifiers ................................................................................................................................................25 3.3.2.2. Raman Effect Amplifiers ...............................................................................................................................................................25

3.3.3. Transmitters.................................................................................................................................................26 3.3.3.1. Light-Emitting Diodes ...................................................................................................................................................................26 3.3.3.2. Lasers .............................................................................................................................................................................................26

3.3.4. Receivers .....................................................................................................................................................28 3.3.4.1. PIN diodes......................................................................................................................................................................................28 3.3.4.2. Avalanche photodiodes (APDs).....................................................................................................................................................28

3.3.5. Wavelength Converters...............................................................................................................................29 3.3.6. Optical Gating .............................................................................................................................................29 3.3.7. Wave Mixing...............................................................................................................................................30

3.4. Summary .............................................................................................................................................................30 Chapter 4 - Signal Degradation on Optical Networks...................................................................31

4.1. Introduction .........................................................................................................................................................31 4.2. Attenuation..........................................................................................................................................................31 4.3. Dispersion ...........................................................................................................................................................32 4.4. Nonlinearities ......................................................................................................................................................33 4.5. Kerr effects..........................................................................................................................................................33

4.5.1. Self-phase modulation (SPM).....................................................................................................................33 4.5.2. Crossphase modulation (XPM)...................................................................................................................34

Table of figures


4.5.3. Four-Wave Mixing (FWM) ........................................................................................................................34 4.6. Scattering effects.................................................................................................................................................35

4.6.1. Stimulated Raman Scattering (SRS)...........................................................................................................35 4.6.2. Stimulated Brillouin Scattering (SBS) .......................................................................................................35 4.6.3. Conclusions on nonlinear effects................................................................................................................36

4.7. Crosstalk..............................................................................................................................................................36 4.7.1. Heterodyne crosstalk...................................................................................................................................36 4.7.2. Homodyne crosstalk....................................................................................................................................37 4.7.3. Conclusions on crosstalk.............................................................................................................................37

4.8. Summary .............................................................................................................................................................38 Chapter 5 - Physical devices used in this project – Fiber Bragg Grating’s ..................................39

5.1. Introduction .........................................................................................................................................................39 5.2. History of photosensitivity and fiber Bragg gratings.........................................................................................39 5.3. Photosensitivity in optical fibers ........................................................................................................................40 5.4. Fiber Bragg grating properties............................................................................................................................41

5.4.1. Physical properties ......................................................................................................................................41 5.4.2. Spectral response of fiber Bragg gratings ..................................................................................................43 5.4.3. Coupled Mode Theory ................................................................................................................................45 5.4.4. Apodization of the spectral response of Bragg gratings ............................................................................47

5.5. Fabrication processes ..........................................................................................................................................47 5.5.1. Interferometric technique............................................................................................................................48 5.5.2. Point-by-point technique.............................................................................................................................49

5.6. Applications of Fiber Bragg Gratings ................................................................................................................51 5.6.1. Laser stabilization .......................................................................................................................................51 5.6.2. Fiber lasers ..................................................................................................................................................51 5.6.3. Reflectors in fiber amplifiers ......................................................................................................................51 5.6.4. Raman-Shifted Lasers and Raman Amplifiers ...........................................................................................52 5.6.5. Sensors.........................................................................................................................................................52 5.6.6. Isolation Filters in Bidirectional Lightwave Transmission........................................................................52 5.6.7. WDM Demultiplexers.................................................................................................................................52 5.6.8. Add/Drop Multiplexers and Optical Cross-Connects ................................................................................52 5.6.9. Dispersion Compensators and Wavelength Converters .............................................................................52

5.7. Summary .............................................................................................................................................................53 Chapter 6 – Wavelength Routers...................................................................................................54

6.1. Introduction .........................................................................................................................................................54 6.2. Optical Add-Drop Multiplexers – OADMs........................................................................................................54

6.2.1. Comparison of common OADM structures ...............................................................................................55 6.3. Optical Cross-Connects - OXCs.........................................................................................................................58

6.3.1. SWITCHING TECHNOLOGIES ..............................................................................................................59 6.3.1.1. OPTOMECHANICAL ..................................................................................................................................................................59 6.3.1.2. MICRO-ELECTRO-MECHANICAL SYSTEM (MEMS) ..........................................................................................................59 6.3.1.3. THERMO-OPTICAL ....................................................................................................................................................................61 6.3.1.4. LIQUID CRYSTAL.......................................................................................................................................................................62 6.3.1.5. GEL/OIL-BASED – “Bubble” Switching.....................................................................................................................................63 6.3.1.6. ELECTRO-OPTICAL ...................................................................................................................................................................64 6.3.1.7. ACOUSTO-OPTIC........................................................................................................................................................................64 6.3.1.8. ELECTROHOLOGRAPHIC.........................................................................................................................................................65 6.3.1.9. BRAGG GRATING BASED ........................................................................................................................................................66

6.3.2. Comparison of OXC technologies..............................................................................................................67 6.4. Summary .............................................................................................................................................................68

Chapter 7 – Work Developed ........................................................................................................69 7.1. Introduction .........................................................................................................................................................69 7.2. Grating Fabrication .............................................................................................................................................69 7.3. Development of an Optical Add-drop Multiplexer ............................................................................................71

7.3.1. Implementation of the first structure – OADM 1.......................................................................................71 7.3.2. Performance Assessment ............................................................................................................................72 7.3.3. Upgraded OADM Structure – OADM 2 ....................................................................................................73 7.3.4. Performance Assessment ............................................................................................................................73

7.4. Optical Crossconnect Architectures ...................................................................................................................74 7.4.1. Description of first structure and its implementation.................................................................................74 7.4.2. Performance Assessment ............................................................................................................................75

Table of contents


7.4.3. Development of an upgraded OXC ............................................................................................................76 7.4.4. Performance Assessment ............................................................................................................................77

7.5. Summary .............................................................................................................................................................82 Chapter 8 – Concluding remarks ...................................................................................................83 References .....................................................................................................................................85 Appendix A – Peltier Devices Appendix B – Peltier electronic Controller Appendix C – OADM Paper submitted in Física 2002 Appendix D – OXC architecture paper submitted in Física 2002

Table of figures Figure 1 – Three Generations of Networks ...................................................................................10 Figure 2 – Wavelength Division Multiplexing..............................................................................12 Figure 3 - WDM Link....................................................................................................................13 Figure 4 - Passive Optical Network (PON) ...................................................................................14 Figure 5 - Broadcast and Select Network: a) star topology; b) bus topology ...............................14 Figure 6 - Wavelength Routing Network: a) phisical topology; b) virtual topology ....................16 Figure 7 - Optical Fiber .................................................................................................................18 Figure 8 - Optical Couplers: a) equal ; b) non-equal .....................................................................19 Figure 9 - a) Isolator; b) Circulator; c) Logical scheme of a three port circulator ........................19 Figure 10........................................................................................................................................20 Figure 11 – Concept of a tuneable multielectrode asymmetric directional coupler ......................20 Figure 12 - Fiber Bragg Gratings ..................................................................................................21 Figure 13 - Array Waveguide Grating...........................................................................................21 Figure 14 - Operational setup of a Fabry-Perot Tunable Filter .....................................................22 Figure 15 - Mach-Zehnder Interferometer; b) Three Mach-Zehnder Chain .................................22 Figure 16 - Basic acousto-optic tunable filter ...............................................................................23 Figure 17 - Operational Principle of an EDFA with a: a) 1480 nm pump laser; b) 980 nm pump

laser........................................................................................................................................25 Figure 18 – Semiconductor Optical Amplifier (SOA) ..................................................................25 Figure 19 - Raman Amplification..................................................................................................26 Figure 20 – Stimulated Emission...................................................................................................26 Figure 21 - General structure of a laser .........................................................................................26 Figure 22 - a) Fabry Perot Laser b) DFB Laser.............................................................................27 Figure 23 - 3R Regeneration principle ..........................................................................................29 Figure 24 - a) Mach-Zehnder interferometer and b) Michelson Interferometer configurations

using pairs of SOAs for implementing CPM wavelength conversion scheme......................30 Figure 25 – Optical fiber attenuation.............................................................................................31 Figure 26 - Intersymbol Interference (ISI) ....................................................................................32 Figure 27 - Spectral broadening due to SPM ................................................................................34 Figure 28 – FWM: The mixing of f1 and f2 generate sidebands ..................................................34 Figure 29 - SRS transfers optical power from shorter wavelengths to longer wavelengths .........35 Figure 30 - Heterodyne crosstalk in a WDM system ....................................................................37 Figure 31 - Homodyne crosstalk in a WDM system .....................................................................37 Figure 32 -Power penalties from Heterodyne and Homodyne crosstalk for 8 and 16 WDM

channels .................................................................................................................................37

Table of figures


Figure 33 - Schematic representation of a Fiber Bragg Grating....................................................42 Figure 34 - Spectral Response of a diffraction Grating Filter. ......................................................46 Figure 35 - Refractive index profile of an apodized Fiber Bragg Grating. ...................................47 Figure 36 - Setup for interferometric fabrication of Fiber Bragg Gratings. ..................................49 Figure 37 - Setup for Point-by-point Fabrication Technique. .......................................................50 Figure 38 - Setup for Phase-Mask fabrication of Fiber Bragg Gratings. ......................................51 Figure 39 - Dispersion compensation using an aperiodic grating with a length of 6cm associated

with an optical circulator. ......................................................................................................53 Figure 40 - Common OADM structures: FB-a), FB-b) and FB-c) – Interferometric structures

based in Fiber Bragg gratings; FB-d), FB-e) and FB-f) – Fiber Bragg grating and circulator based structures; FA-a), FA-b) and FA-c) – Array Waveguide Grating Mux structures......56

Figure 41 - OADMs based in Multiport Optical Circulators (MOC’s) Configurations. ...............58 Figure 42 - Toshiyoshi and Fujita’s 2×2 MEMS optical switch . .................................................60 Figure 43 - 8x8 2-D Optical switch ...............................................................................................60 Figure 44 - Schematic of 3-D MEMS switching...........................................................................60 Figure 45 - 3-D micromachined mirrors........................................................................................61 Figure 46 - Scratch Drive actuator switching 1x2. ........................................................................61 Figure 47 - Thermooptical switching - a)Schematic diagram of a Mach Zehnder switch ; b)

Light path in one of the switching. ........................................................................................62 Figure 48 - Total internal reflection switching – Agilent’s Champagne Bubble Switch. .............64 Figure 49 - Acoustooptic Tunable filter ........................................................................................65 Figure 50 - Electroholographic Matrix with ferro-electric crystals...............................................66 Figure 51 - A 4x4 rearrangeable nonblocking OXC using: a) 12 three-port Optical Circulators

and b) Four five-port Multiport Optical Circulators (MOC’s). .............................................67 Figure 52 - Transmission and reflection grating spectra. ..............................................................70 Figure 53 – First Optical Add-drop structure implemented - OADM 1.......................................71 Figure 54 – Optical WDM source obtained through slicing of a Broadband optical source’s

(LED) spectra. .......................................................................................................................71 Figure 55 – A – Input signal composed of three WDM channels (l1, l2,l3); B– Output Signal

composed of signals l1 and l3 ; C – Dropped Channel l2 . ...................................................72 Figure 56 – Heterodyne Crosstalk caused by imperfect FBG filtering. ........................................72 Figure 57 - Second Optical Add-drop structure implemented - OADM 2....................................73 Figure 58 - C – Dropped Channel λ2 in OADM 1; D – Dropped Channel λ2 in OADM 2. ........74 Figure 59 – Optical Crossconnect 1 ..............................................................................................74 Figure 60 – Optical Crossconnect 1 performance test...................................................................75 Figure 61 - Power Spectral Response of the OXC in a detuned state. ..........................................76 Figure 62 - OXC 2. ........................................................................................................................77 Figure 63 - Multiwavelength Fiber Ring Laser Source using Fiber Bragg Gratings. ...................77 Figure 64 - Input Signals ...............................................................................................................78 Figure 65 - Output port signals with optical filters detuned..........................................................78 Figure 66 - Outputs when Channel 1 is switched from Input 1 to Output 2. ................................79 Figure 67 - Tuning and detuning speed. ........................................................................................80 Figure 68 - Tuning and detuning of the optical filter (FBG).........................................................81 Figure 69 - Wavelength tuning and detuning of the optical filter (FBG)......................................81

Chapter 1- Introduction


Chapter 1 - Introduction

1.1. Abstract

Optical networks have evolved to sustain the growing need for bandwidth in communication

networks. Wavelength switched networks using Dense Wavelength Division Multiplexing –

DWDM, are a solution to this problem. Development of these networks demands the use of

nodes capable of controlling the communication channels and route them correctly. These nodes

are sometimes referred to as Wavelength Routers.

As explained earlier in this report, the purpose of this project is the development of an

Optical Router and the implementation of a control plane similar to the ones in the IP world, as

defined in GMPLS research (IETF). The work can be split into two parts:

Part 1 – Development of the underlying optical fabrics.

Part 2 – Definition of the structure of the protocol stack, needed to establish the control

plane at the OXC.

Each of these parts was developed by different groups in different research units of INESC

Porto.

1.2. Background and Motivation

This project is currently a hot topic of research. The pursuit of an All-Optical network is

considered the “Quest for the Holy Graal” of Optical Network Research. Several technologies

have tried to reach the same goal and some of them are getting very close to their objective. The

optical technology used is this project is Fiber Bragg Grating based. These devices have proved

to be an essential building block in optical networks nowadays and have many other

applications. Work carried out by some research units like the Optoelectronics unit at INESC and

the Institute of Telecommunications (IT) in Aveiro is used as example1 and architectural

advances are introduced.

1.3. Report Organization

This report is divided in chapters which give readers a possibility to choose the topics of

interest. Following this introductory chapter, where the main purpose and motivation of the



project is presented, the state of the art of the technology at the present time is described. Chapter

(2) gives an overview of Multiwavelength Optical Technology and Networks where fundamental

notions about optical networks and their current state are developed. Since this is a project where

an optical device is actually developed chapter 3describes the state-of-the-art in optical

components for WDM networks. Components ranging from the fiber itself, to active components

and complex amplifiers based in nonlinear effects are described briefly.

Performance tests are an important part of this project and this means that optical networks

and device impairments should be detailed. This is done in Chapter 4.

The key elements in the fabrics of this device are Optical Filters Based in Fiber Bragg

Gratings, their characteristics being fully described in Chapter 5.

The final Chapter, before the analysis of the physical implementation, (Chapter 6) describes

the state-of-the-art in Wavelength Routers, a category in which the device developed in this

project can be classified.

The physical implementation of this work is detailed in Chapter 7 where all the structures

developed are explained and their performance tests are presented.

Concluding remarks are given in Chapter 8. Finally, Appendix A describes the physical

characteristics behind a Peltier Device and the electronic control circuitry is described in

Appendix B.

1.4. Contributors

As a result of this work two articles and a device patent have been submitted for admission

by proper authorities. The following articles where written:

- I. Terroso, J. P. Carvalho, O. Frazão, M. Ricardo, H. M. Salgado, “Avaliação de duas Arquitecturas de OADM Baseadas em Circuladores Ópticos e Redes de Bragg em Fibra Óptica”, Física 2002. - J. P. Carvalho, I. Terroso, O. Frazão, V. Barbosa, M. Ricardo, H. M. Salgado, “Comutador Óptico (OXC) Baseado em Circuladores Ópticos e numa Rede de Bragg em Fibra Óptica”, Física 2002. The device developed was object of a patent described as: - Comutador Óptico (OXC) de 2 × 2 Portas para Sistemas de Multiplexagem em comprimento de Onda e Escalável a N × N Portas.



This invention refers to the area of Optical Fiber Communications and can be defined as an

2x2 Optical Cross-connect for wavelength switched networks (WDM) that can be scaled to NxN

ports.

It should be noted that the main objective of this project was, initially, to develop an Optical

Add-drop Multiplexer but initial encouraging results made us proceed to build an Optical Cross-

connect based on the same technology. Insight into various areas in terms of optical networking

and technology was gained. The “hands-on” approach of this work when dealing with the optical

components and electronic circuits provided us with a thorough practical experience and know-

how that could not be gained through only the theoretical study of these matters.

Chapter 2 - Optical Networks

2.1. Introduction

In the last years we have seen a growing interest in optical networks in order to increase the

capacity of communication networks. The purpose of this chapter is to provide a level set in

about purely optical networks, or All optical networks (AONs)2,3,4 as they are usually called:

What they are good for, how far we have gotten with them, and how far we have yet to go.

2.2. The three networks generations

All optical networks are those that in which the path between the using nodes at the ends

remains entirely optical from end to end. Such paths are termed lightpaths. Each lightpath may

de optically amplified or have its wavelength altered along the way, but it’s a purely optical path.

As the optoelectronics technology to build optical networks is gotten closer to functional and

economic feasibility, more and more groups worldwide are studying them as a possible base

upon which to build the networks of the future, both within the wide-area backbone and for

metropolitan and local area distribution facilities. In light of the potential and recent advances,

all optical networks are very often considered to be the main candidate for constituting the

backbone that will carry global data traffic whose volume has been growing at impressive rates

that are not expected to slow down in the near future.

According to the physical technology employed, one can identify three generations of

networks (Figure 1). Networks built before the emergence of optical fiber technology are the first

Chapter 2- Optical Networks


generation networks (i.e. networks based on copper wire or radio). The second generation

networks employ fibers in traditional architectures. The choice of fiber is due to its large

bandwidth, low error rate, reliability, availability, and maintainability. Although some

performance improvements can be achieved by employing fibers, the performance for this

generation is limited by the maximum speed of electronics (a few Gbps) employed in switches

and end-nodes. This phenomenon is called the electronics bottleneck. In order to satisfy the

increasing bandwidth requirements of emerging applications, totally new approaches are

employed to exploit vast bandwidth (approximately 30THz in the low loss region of single mode

fiber in the neighborhood of 1500nm) available in fibers. Therefore, the third generation

networks are designed as all-optical to avoid the electronics bottleneck. That is, information is

conveyed in the optical domain (without facing any electro-optical conversions) through the

network until it reaches its final destination. The emergence of single mode fibers, all-optical

wide-band amplifiers, optical couplers, tunable lasers (transmitters)/filters (receivers), optical

add-drop multiplexers and all-optical crossconnects5,6 make third generation networks a reality.

Figure 1 – Three Generations of Networks

In order to make use of the vast bandwidth available without experiencing electronics

bottleneck, concurrency among multiple user transmissions can be introduced. In all-optical

networks, concurrency can be supplied through time slots (OTDM - Optical time division

multiplexing), wave shape (OCDM - Optical code division multiplexing) or wavelength (WDM -

Wavelength division multiplexing)7.

2.3. OTDM

In optical time division multiplexing (OTDM)8, many low-speed channels, each transmitted

in the form of ultra-short pulses, are time interleaved to form a single high-speed channel. By

this method, the information carrying capacity of the network can be improved to 100 Gbps or

higher without experiencing electronics bottleneck. In order to avoid interference between



channels, transmitters should be capable of generating ultra-short pulses, which are perfectly

synchronized to the desired channel (time slot), and receivers should have a perfect

synchronization to desired channel (time slot).

OTDM is a promising multiplexing technique, but it is still on research level. It is a very

similar to electronic TDM, the only difference is that OTDM is faster, and the devices used in

OTDM networks are optical. The main advantages of this multiplexing technique are the facts

that only one source is required and the node equipment is simpler in the single channel

architecture. The problems concerning OTDM are associated with the immaturity of all-optical

devices. Very fast lasers are needed and timing synchronization and alignment are still problems.

Additionally, although systems were technically feasible, it would not be economically viable.

Another weakness is that the OTDM is not a transparent multiplexing technique.

2.4. OCDM

In Optical code division multiplexing (OCDM)9, each channel is assigned a unique code

sequence (very short pulse sequence), which is used to encode low-speed data. The channels are

combined and transmitted in a single fiber without interfering with each other. This is possible

since the code sequence of each channel is chosen such that its cross-correlation between the

other channels' code sequences is small, and the spectrum of the code sequence is much larger

than the signal bandwidth. Therefore, it is possible to have an aggregate network capacity

beyond the speed limits of electronics. Like OTDM, CDM requires short pulse technology, and

synchronization to one chip time for detection.

Summarily, in OCDM each optical code (channel) is assigned to its own, independent path

(OCP). Cells with different OCPs can be transmitted in the same fiber at the same time. OCDM

is transparent and no synchronization between different channels is needed. Using OCDM in

addition to WDM can remarkably improve the communication capacity of the network. OCDM

can also be used to improve the communication security.

2.5. WDM

In WDM10, the optical spectrum (low loss region of fibers) is carved up into a number of

smaller capacity channels (Figure 2). Users can transmit and receive from these channels at peak

electronic rates, and many users can use the different channels simultaneously. In this way, the

aggregate network capacity can reach the number of channels times the rate of each channel. In



order to develop an effective WDM network, each user may be able to transmit and receive from

multiple channels. That is why, the tunable transmitter (laser) / tunable receivers (filter) and/or

multitude of fixed transmitters/receivers are employed at end-nodes.

Figure 2 – Wavelength Division Multiplexing

WDM is an optical version of FDM. As we see, the idea is that several signals are

transmitted at the same time in the same fiber at different wavelengths. This way WDM provides

many virtual fibers on a single physical fiber. Today WDM is the most popular alternative to

multiplex signals in the optical domain. WDM is also the most mature all-optical multiplexing

technique. Its main advantages are the signal transparency, scalability and flexibility; the existing

fiber lines can be upgraded by implementing WDM. The main problems are the need for flat

gain amplifiers, increasing noise when the number of channels increases, and the limits in

channel spacing and in the number of channels caused by wave mixing and cross-phase

modulation (see Chapter 4 for further information).

The key parameters of any multiplexing system are the total capacity of the system, number

of channels, the spectral efficiency and the transmission distance.

WDM is the favorite choice over OTDM, and CDM. This is due to the complex hardware

requirements, and synchronization requirements of OTDM and CDM (synchronization within

one time slot time and one chip time respectively). OTDM and CDM are viewed as a long-term

network solution, since they rely on different and immature technology. Whereas it is possible to

implement WDM systems using components that are already (or will be, in a short time),

available commercially. Moreover, WDM has the inherent property of transparency. Since there

is no electronic processing involved in the network, channels act like independent fibers

(transparent pipes) between the end nodes provided that channel bandwidths are not exceeded.

Once a connection is established between the end-nodes on a WDM channel, the communicating

parties have the freedom to choose the bit rate, signaling and framing conventions, etc. (even

analog communication is possible). This transparency property makes it possible to support

various data formats and services simultaneously on the same network. In addition to this great



flexibility, transparency protects the investments against future developments. Once deployed,

WDM networks will support a variety of future protocols and bit rates without making any

changes to the network. The most commonly used architectural forms for WDM networks are

WDM Link, Passive Optical Network (PON), Broadcast and Select Networks, and Wavelength

Routing Networks.

2.5.1. WDM Link

To increase information carrying capacity, second generation networks employ parallel fibers for

individual channels. In the WDM Link approach, parallel fibers are replaced by wavelength

channels on a single fiber (Figure 3). In long haul WDM links, all channels are amplified

together by a single wideband optical amplifier (no separate amplifier for each channel), and

existing fibers are used efficiently by integrating more than one channel in a single fiber.

Therefore, WDM link offers a very cost-effective system. The other factors that make WDM

links very popular are the maturity of this technology and its simplicity of integration with

legacy equipment.7

Figure 3 - WDM Link

2.5.2. Passive Optical Network (PON)

The main feature of a PON is to share fiber between the Central Office and Optical Network

Units (ONU) (Figure 4). The PON establishes a tree structure that enables bi-directional

communication between a server (central office) and multiple customers (ONUs) with

centralized control and routing at the central office.

This architecture is a good network choice for regional communication providers. The main

technological problem for the PONs is to design cheap, simple, and durable equipment for the

ONUs. 11



Figure 4 - Passive Optical Network (PON)

2.5.3. Broadcast and Select Networks

Broadcast and Select Networks offer an optical equivalent to radio systems. In these

networks, each transmitter broadcasts its signal on a different channel, and receivers can tune to

receive the desired signal. In other words, the data is broadcasted at a special wavelength to all

nodes and the receivers accept only certain wavelengths, i.e., data channels, therefore the data is

rejected in those nodes that it does not belong to. Generally, broadcast and select networks are

based on a passive star coupler (Figure 5a)). This device is connected to the nodes by fibers in a

star topology. The signals received at the input ports are evenly distributed to the output ports.

The main networking problem for these networks is the coordination of pairs of stations in order

to agree and tune their systems to transmit and receive on the same channel. The most important

disadvantages of these networks are splitting loss and lack of wavelength reuse. Therefore,

broadcast and select networks are suitable for local area networks, but are not scalable to wide

area networks. 12,13,14

Figure 5 - Broadcast and Select Network: a) star topology; b) bus topology

2.5.3.1. Single-Hop Networks

In single-hop networks data is converted to electrical form only at the end of the path. These

networks can be either circuit or packet switched, and packet switched networks furthermore

connectionless or connection oriented. Single-hop networks can function as packet networks if



fast tunable receivers and/or transmitters are used. Actually, the first implemented optical packet

networks were single-hop broadcast-and-select networks. 15

2.5.3.2. Multihop Networks

In multihop networks every node has only a few fixed transmitters and receivers. Therefore,

a signal cannot always be directly transmitted from source node to destination node. Instead

signals have to be received at some intermediate nodes along the way, converted to electronic

form and retransmitted. 12,16,17

2.5.4. Wavelength Routing Networks

Wavelength-routed networks currently14 represent the most promising technology for optical

backbone networks. They can be either single-hop or multihop. Wavelength Routing Networks

are composed of one or more wavelength selective nodes called wavelength routers and fibers

interconnecting these nodes. Each wavelength router has a number of input and output ports.

These ports are connected to either end-nodes or other wavelength routers. Each wavelength

router makes its routing decision according to the port and wavelength of the input signal.

Signals routed to the same output port should be on different wavelengths. As long as any two

channels do not share the same fiber link anywhere on the network, they can use the same

wavelength in wavelength routing networks. This wavelength reuse feature results in a

tremendous reduction in the number of wavelengths required for building wide networks.

Depending on design and components in use, a wavelength router may have a variety of

capabilities. For example, its routing matrix may be static or re-configurable, and it may provide

wavelength conversion or not. These features have a direct influence on the operation and

scalability of the network. Therefore, wavelength routing networks are the primary choice for

wide-area all-optical networks.

The network topology (Figure 6) is typically a mesh. In these networks the routing and

switching functions are done on a lower layer called the optical layer. The data transfer is done

using lightpaths. The network is transparent and protocol insensitive and component expenses

are saved because less high layer logic is needed. This network is ideal for circuit switching but

not suitable for packet switching. However, hybrid networks, i.e., networks consisting of both

circuit and packet switching can be implemented.12,18,19,20,21,22



Figure 6 - Wavelength Routing Network: a) phisical topology; b) virtual topology

2.6. DWDM

The literature often uses the term DWDM (Dense Wavelength Division Multiplexing) in

contrast to regular WDM. This term doesn’t denote a precise operation region or implementation

condition, but, instead, is a historically derived designation. The original use of the WDM was to

upgrade the capacity of installed point-to point-transmission links. Typically, this was achieved

by adding wavelengths that were separated by several tens, or even hundreds, of nanometers, so

that strict requirements would not be imposed on the different laser sources and the receiving

optical wavelength splitters. In the late 1980’s, with the advent of tunable lasers that have

extremely narrow linewidths, one could then have closely spaced signal bands. This is the basis

of DWDM.23

2.7. Summary

The emergence of new optical technologies enables the realization of third generation

networks. That type of computer networks completely avoids the electronics bottleneck

appealing to all optical technology. Between the three techniques of concurrency among multiple

user transmissions introduced, WDM was presented as the favorite choice over OTDM, and

OCDM and the reasons for that have been presented. The most commonly used architectural

forms for WDM networks where also mentioned with particular evidence given to Wavelength

Routing Networks. Finally, a small definition of the new standard – DWDM - is made.

Chapter 3- Optical WDM Components


Chapter 3 - Optical WDM Components

3.1. Introduction

Over the past two decades, the telecommunications industry has witnessed an unprecedented

growth in data traffic and the need for networking. The exploitation of WDM as a networking

mechanism where signals are routed, switched, and selected based on wavelength marks the

dawn of a new era in optical communications.

To enable the exploitation of the new, vast, and versatile wavelength dimension, new

technologies had to be developed and demonstrated. Optical sources and receivers operating at a

fixed wavelength were no longer adequate, and tunable as well as switched sources together with

tunable filters were thus developed. New components that can exercise selection, switching, and

routing based on wavelength were also needed, and as a result, components such as Fiber Bragg

Gratings, arrayed waveguide gratings (AWGs), tunable filters, and wavelength multiplexers and

demultiplexers.

3.2. Optical passive components

3.2.1. Fiber

As a transmission medium fiber has several benefits over copper. The attenuation is lower,

the transfer rates can be higher and there is no electromagnetic interference. Additionally, the

fiber is lighter and stronger than the copper. Naturally the fiber also has some weaknesses, such

as dispersion, nonlinear refraction or attenuation. Yet, compared to other transmission media the

fiber is an attractive alternative. In fact, as early as in the 1980's it was clear that the optical fiber

would be necessary to support high capacity systems.

An optical fiber consists of a core and a cladding surrounding it (Figure 7). Both are made of

pure silica glass, but their refractive indices are different. The use of the optical fiber as a

transmission medium is based on this refractive index difference. Depending on the difference

between refraction indexes and the angle at which the light strikes the interface of two different

transmission media, it either reflects or refracts. If the refraction indexes differ sufficiently, a part

of light confronting the interface is reflected. By controlling the angle at which the light waves

are transmitted and encounter the interface of core and cladding, the proportion of the reflected



light can be increased and the signal can be contained within the core and guided to the other end

of the fiber.23, 31

Figure 7 - Optical Fiber

The signal traveling in the fiber has an infinite number of possible paths. These paths are

called modes. The optical fiber can be either multimode or single-mode. Multimode fibers have a

core approximately five times larger. Because the propagation time along different paths is

different, multimode fibers suffer from Intermodal Dispersion, i.e. differences in the propagation

times of waves in the fiber. The multimode fiber is simpler to manufacture making it cheaper,

but the single-mode fiber has several advantages. Today multimode fibers are used if a cheap

alternative is needed, whereas the single-mode fibers are used over the longer distances. The

most remarkable problem in using optical fiber is dispersion. The problems resulting from

dispersion increase as the bit rates increase, which make this problem significant. There are three

kinds of dispersion. Chromatic dispersion follows from the fact the color of the light has an

effect on its propagation speed. If the core of the single mode fiber is asymmetric, and the light

beams traveling along different sides of it have therefore different speeds, the polarization mode

dispersion occurs. Slope mismatch dispersion occurs in single-mode fibers, because dispersion

varies with wavelength.

3.2.2. Couplers

Couplers are simple passive optical components which are used to split or combine signals.

A coupler consists of n input and m output ports. A 1 x n coupler is called a splitter and an n x 1

coupler is called a combiner. Figure 8 a) describes a 2 x 2 coupler. In 2 x 2 coupler a part of

input signal 1 is directed to output port 1 and the rest to the output port 2. In a similar way a part

of input signal 2 is guided to both output ports. The fractions directed to output ports can be

either equal or non-equal (Figure 8b)). Couplers function as building blocks of other

components. A coupler can also be used for measurements by separating a small fraction of

signal for this purpose.



Figure 8 - Optical Couplers: a) equal ; b) non-equal

3.2.3. Isolators and Circulators

Isolators (Figure 9a)) are devices that allow transmission only in one direction and block the

transmissions in the opposite (reverse) direction. This way reflections from, e.g., amplifiers or

lasers can be prevented. Typically the insertion loss, i.e. the loss in the forward direction is

around 1 dB and the isolation, i.e. the loss in the reverse direction, is approximately 40 to 50 dB.

A circulator is a device similar to an isolator, but with multiple ports. Figure 9 b) shows a

circulator with 3 input and output ports. A signal from each port is directed to the next adjoining

port and blocked in all the other ports, as in Figure 9 c). Circulators can be used as a component

in optical add/drop multiplexers and optical cross-connects.34

Figure 9 - a) Isolator; b) Circulator; c) Logical scheme of a three port circulator

3.2.4. Filters

To filter or multiplex channels which are based on wavelengths, one has to separate different

wavelengths from the signal. There are many different ways to do this, but in principle they are

all based on the same idea: some wavelengths are delayed in phase compared to other

wavelengths. This is done by directing them through a longer path.

The key parameters of filters are insertion loss and passband flatness. Insertion losses should be

low and independent of polarization and temperature. Passband should be flat and passband

skirts should be as sharp as possible. Figure 10 shows a signal that illustrates this situation. The



flatter the passband is and sharper the passband skirts are, the smaller crosstalk energy passing

through the adjacent channels is.33

Figure 10

3.2.4.1. Tunable 2 x 2 directional couplers

Tunable 2 x 2 directional couplers have multiple control electrodes placed on the coupling

waveguides. Figure 11 illustrates a multielectrode asymmetric directional coupler fabricated on a

LiNbO crystal, where one arm is thinner than the other. For a wavelength-dropping application

in this device, M wavelengths enter input port 1. Applying a specific voltage to the electrodes

changes the refractive index of the waveguides, thereby selecting one of the wavelengths, say λi,

to be coupled to the second waveguide, so that it exits through port 4. The remaining M - 1

wavelengths pass along the device and leave through port 3. To insert a wavelength and combine

it with an input stream entering port 1, one inserts λi into port 2, so that it couples across to the

top waveguide. Thus, it exits port 3 along with the other wavelengths λ1, … , λi-1, λi, λi+1, λM that

entered port 1.24

Figure 11 – Concept of a tuneable multielectrode asymmetric directional coupler



3.2.4.2. Gratings

Gratings work based on the principle of diffraction where a signal that is guided to a small

hole spreads into many directions. Because one path the wavelength travels is longer than the

other path traveled, it is added in phase. Fiber Bragg gratings (Figure 12) are modified pieces

of fiber that reflect one wavelength and pass through all the others. They are simple to fabricate

and use and they have low insertion losses.34 (See Chapter 5 for further information about Fiber

Bragg Gratings)

Figure 12 - Fiber Bragg Gratings

3.2.4.3. Arrayed waveguide grating (AWG)

In arrayed waveguide grating (AWG) (Figure 13) the idea is that signals with many

wavelengths is copied to several fibers with different lengths and on each fiber all but one

wavelength are rejected. In principle the idea is the same as in gratings: different wavelengths

get different phases, by being delayed differently. AWGs are promising devices. They can be

used in integrated optical circuits, while they can be easily combined with other functions.

Additionally, they can be used as both a multiplexer and a demultiplexer. The disadvantage is

their high temperature coefficient.25,34

Figure 13 - Array Waveguide Grating

3.2.4.4. Fabry-Perot Tunable Filters (FPF)

Fabry-Perot interferometer Filters (Figure 14) are micro-optic cavities in which two mirrors

are placed parallel to each other. When a signal meets a mirror, a part of it continues through and

the other part is reflected. The reflected signal travels straight to the other mirror and is guided

back. It then travels through the same mirror wavelength 1 has already gone and will be added in



phase. The advantage of Fabry-Perot filters compared to many other filter types is their ability to

be tuned to filter different wavelengths.25,34

Figure 14 - Operational setup of a Fabry-Perot Tunable Filter

3.2.4.5. Mach-Zehnder Tunable Filters (MZF)

Whereas the Fabry-Perot interferometer Filters involves light interference by many repeated

reflections, a single Mach-Zehnder Interferometer (MZF) (

Figure 15 a)) involves interference by only two versions of the same light transversing paths

of slightly different length. The multichannel input signal is split into two equal parts by a 3 dB

coupler. The two versions of the same signal traverse paths of slightly different lengths and

merge together in another 3 dB coupler at the output.

Figure 15 - Mach-Zehnder Interferometer; b) Three Mach-Zehnder Chain

The real advantage of the MZF, which usually are used in MZF chains (

Figure 15b)), is that the filter can be realized using lithographic technology, leading to

potentially low fabrication costs. Also, by designing a square waveguide cross-section, these

filters can be made polarization insensitive. The main disadvantages are the slow tunning speed

due to thermal inertia (a few miliseconds) and the complexity of the multistage tunning

control.,25,34



3.2.4.6. Liquid-crystal Fabry-Perot filters

Liquid-crystal Fabry-Perot filters are based on the use of high-speed electro clinic liquid

crystals inside a Fabry-Perot cavity. In this case, the liquid crystal is positioned between the two

fiber end faces, and thus becomes part of the Fabry Perot cavity. These filters can be widely

tuned by appiying a voltage across the crystal, which changes the refractive index, and hence the

optical path length, in the cavity material.26, 27

3.2.4.7. Acousto-optic tunable filters

Acousto-optic tunable filters (AOTFs) operate through the interaction of photons and

acoustic waves in a solid such as lithium niobate. Figure 16 shows the basic operation. Here, an

acoustic transducer, which is modulated by radio-frequency (RF) signal, produces a surface

acoustic wave in the LiNbO crystal. This wave sets up au artificial grating in the solid, the

grating period being determined by the frequency of the driving RF signal. More than one

grating can be produced simultaneously by using a number of different driving frequencies.

Input wavelengths that match the Bragg condition of the gratings are coupled to the second

branch of the AOTF, while the other wave lengths continue on through.28,29,30

Figure 16 - Basic acousto-optic tunable filter

3.2.5. Multiplexers

3.2.5.1. WDM Multiplexers and Demultiplexers

Multiplexer is a device which combines several signals with different wavelengths to one

fiber. Respectively a demultiplexer gets one signal as an input and by separating the different

wavelengths from the fiber, assorts each wavelength to its own output fiber. The purpose of these

devices is to increase the capacity of a fiber by increasing the number of channels per fiber from

one to hundreds.



3.2.5.2. Add/Drop Multiplexers (OADMs)

Add/drop multiplexers are devices that are used for adding or removing single wavelengths,

i.e. channels, from a fiber without disturbing the transmission of other signals. Optical add/drop

multiplexers are widely used in WDM networks. 25 (See Chapter 6 for further information)

3.2.5.3. Optical Cross-Connects (OXCs)

An Optical Cross-connect is a device used in optical switching. An optical channel at one of

the input pots of the OXC could be sent to any of the output ports according to the network

switching requirements. As the OADMs, OXCs are a key component in the AONs.25 (see

Chapter 6 for further information)

3.3. Optical active components

3.3.1. Amplifiers

A signal attenuates while it travels through an optical fiber or through optical components. In

order to travel over long distances the signal has to be amplified. Earlier the signals were

converted to electrical form and regenerated. Today there are optical amplifiers and the signal

can be transmitted over longer distances without conversion to electric form. Compared to

regenerators, optical amplifiers are more flexible to changes in the bit rate. Additionally, they

can be used to amplify several wavelengths at the same time.31

3.3.2. Erbium-doped fiber amplifiers (EDFAs)

Erbium-doped fiber amplifiers (EDFAs) consist of a strand of fiber doped with Erbium

atoms, pumping devices and simple optical components. They are all optical devices. The

operation of doped amplifiers is based on the stimulated emission (Figure 17). The signal gets

more energy if there are more transitions from a higher energy level to a lower than from the

lower to the higher. The energy can be pumped to fiber by giving Erbium atoms more energy

and lifting electrons to higher energy state. When electrons fall to lower level, energy is released

and the signal is amplified.



Figure 17 - Operational Principle of an EDFA with a: a) 1480 nm pump laser; b) 980 nm pump laser

For a long time the major problem with amplifying was the fact that different wavelengths

were amplified with different gains. With EDFAs this is not a remarkable problem as the gain is

relatively flat. Erbium-doped amplifiers were the first doped amplifiers constructed.31,34

3.3.2.1. Semiconductor Optical Amplifiers

Semiconductor optical amplifiers (SOAs) (Figure 18) are based on stimulated emission

similarly as EDFAs. The difference is that instead of energy levels of dopant atoms, the process

is based on electrons and electron holes in the semiconductor. The amplifier consists of two

semiconductors separated with a band gap.

Figure 18 – Semiconductor Optical Amplifier (SOA)

Actually, SOAs are not as good amplifiers as EDFAs. SOAs have wider bandwidth, i.e. they

can be used to amplify more wavelengths. However, their output power is weaker, there is more

polarization and coupling losses and they suffer from crosstalk. Still, there is a lot of interest for

SOAs at the moment. SOAs are small compared to EDFAs which makes them suitable for

building blocks for other devices, e.g. switches and wavelength converters.34

3.3.2.2. Raman Effect Amplifiers

A Raman amplifier is not a compact device in the same way as the other amplifiers described

previously. The amplification process (Figure 19) happens slowly in the transmission fiber over

several kilometers. The main idea is that a light beam with lower wavelength and thus higher

energy is guided to the same fiber with the signal to be amplified. The light beam with higher

energy then delivers energy to the signal and signal is therefore amplified. Gain flatness is a



critical issue with Raman Effect Amplifiers and several pumping bandwidths are needed to get

the desired result.32,34

Figure 19 - Raman Amplification

3.3.3. Transmitters

3.3.3.1. Light-Emitting Diodes

If the distance is short and the data transmission rate is low a cheap device called a light-

emitting diode (LED) can be used as a light source instead of laser. LEDs have broad passband

and low output power. A typical output power of LED is -20 dBm, which is low compared to 0 -

10 dBm output power of lasers.33

3.3.3.2. Lasers

Figure 20 illustrates the stimulated emission. A photon interacts with the atom that is in the

higher energy state. Then the electron drops down to the lower energy state and an equal amount

of energy is released. Because of the stimulating photon the new photon produced has exactly

the same direction, phase and wavelength as the first one. The stimulated emission is a necessity

for the operation of a laser.

Figure 20 – Stimulated Emission

Figure 21 - General structure of a laser



Additionally, a source of energy, a cavity filled with suitable material for the emission and

two mirrors are needed (Figure 21). The material can be solid, liquid or gas. It is suitable for a

particular laser if the energy difference between the lower and the higher state of the atom is

correct such that a photon with desired wavelength could be produced by stimulated emission.

To produce a coherent light beam with the laser, energy is guided to cavity filled with the

particular material. The electrons are then elevated to the higher energy state and photons are

emitted spontaneously. Most photons emitted exit the cavity through the walls of the cavity.

Some of them, however, confront the partial mirror with certain angle and are reflected back to

the cavity. When this kind of reflected photon encounters an electron in the higher energy state

stimulated emission happens and a new photon with exactly the same direction, phase and

wavelength is produced. These photons continue toward the other mirror, stimulate new photons

and are reflected back. When the photons meet the partial mirror, a part of them continue

through and a part is reflected and stimulates new photons. Soon there are plenty of photons

traveling in the cavity, and a great amount of light is also directed through the partial mirror.

In addition to the mentioned conditions, the length of the cavity has to be a multiple of half

the wavelength of the output beam. Therefore the output light consists of only a limited number

of wavelengths. However, the output beam of the lasers should be as narrow as possible and to

improve the laser all these wavelengths but one should be rejected. This can be done by filtering

or by using external cavity. Another alternative is to use a very short cavity. Today simple

Fabry-Perot lasers (Figure 22 a)), as the one discussed previously, are often replaced with

distributed feedback (DFB) lasers (Figure 22b)). In these lasers the width of the cavity has

periodic changes and the beam is reflected several times back and forth over short distances.

This device can work as designed to be a narrow band single mode laser.

Figure 22 - a) Fabry Perot Laser b) DFB Laser



Another remarkable improvement to Fabry-Perot laser is the tunability. External cavity lasers

can function as slow tunable lasers, if the length of the other cavity can be changed. Usually this

is done by moving the mirror or grating which forms the external cavity. A faster way is to

chance the bias current, which chances the refractive index. The disadvantage of this approach is

variation in the output power. Another possibility is to use two independent currents, one for

controlling the output power and the other for controlling the wavelength of the laser.34, 35

3.3.4. Receivers

There are many ways of converting incident light into a current or a voltage, and the general

class of devices that do so are photodetectors. In all different types of photodetectors, only the

semiconductor photodiode subclass has proved to have the right combination of sensitivity, low

noise, small size, low cost, and high speed of response to serve satisfactorily in a fiber optic

communication system. The internal processes by which they convert a photon flux into a

current are exactly the reverse of the processes of a light semiconductor LED or a Laser diode.

There are three types of semiconductor photodiodes: PN diodes, PIN diodes and avalanche

photodiodes (APDs).33,3

3.3.4.1. PIN diodes

At one time PN photodiodes were the most prevalent detection device used in lightwave

systems, but PIN devices have now superseded these. These photodiodes are called pin

photodiodes because the material is made up of p-type, intrinsic (lightly doped intrinsic

semiconductor material), and n-type material. In such a device, for every single photon incident,

a single electron will rise to its excited state. This will be satisfactory for most short-range and

low bit-rate systems. However, if a signal has been weakened significantly, then a more

advanced type of detector may be required to detect it.34,36

3.3.4.2. Avalanche photodiodes (APDs)

One option would be to use an “avalanche photodiode” (APD), which is a discrete

semiconductor device like the PIN photodiode. It differs, however, in that for every incident

photon, it can generate several excited electrons – as many as 100. Therefore the signal is

boosted many times over and so lower optical powers can be detected successfully. An APD

operates at a much higher voltage than a PIN and is designed so that a photon moves an electron

to its excited state with enough energy to cause further electrons to be excited also. These extra

electrons can themselves cause further electrons to rise to their excited states, and so there is a

chain reaction process – or an “avalanche.”



Unfortunately, APDs are not without their problems. First, the amount of amplification is

limited. If too much amplification is provided, the device will “run away” and simply provide an

essentially continuous (large) current. Second, APDs are noisy because thermal action can

randomly promote an electron to the conduction band. The current resulting from this thermal

action is noise. And finally, the avalanche action takes time to occur, meaning that the device has

a certain response time. APDs have been used a great deal in the network but at the highest

speeds, the equipment providers tend to use pin diodes with optical preamplifiers.34,36

3.3.5. Wavelength Converters

Wavelength converters are used for converting the wavelength of an incoming signal to a

different wavelength. They can be used for instance as a part of a switch or cross-connect or in

3R regenerators. 3R regenerators are devices that regenerate signals amplitude and regenerate the

signal also in time and frequency domains. The 3R regeneration principle is shown in the Figure

23.

Figure 23 - 3R Regeneration principle

Wavelength converters can be optoelectronic, or based on optical gating or wave mixing.

Optoelectronic converters are devices which convert signal to electronic form, regenerate and

retransmit it. The problem with optoelectronic converters is that they require fixed data rate and

format. The two other possibilities are discussed here in more detail.25

3.3.6. Optical Gating

There are two possibilities to use optical gating: cross-gain modulation and cross-phase

modulation (CPM). The first is based on the fact that the gain of a SOA dependents on input

power. A low power probe signal with desired wavelength is sent into the SOA. Because the

probe signal has a low gain compared to the input signal, it will experience high gain when the

gain of the input signal is high (state 1) and low gain when gain to the input signal is zero (state

0).



In cross-phase modulation (Figure 7) the phase of the probe signal is changed on basis of the

input signal. The phase modulation is then converted into intensity modulations by using

interferometer. Cross phase modulation is the more attractive alternative of these two. It requires

less power and it has a better extinction ratio. Additionally, pulse distortion, from which cross-

gain modulation suffers, is not a problem but this feature is the operational principle of the

device.37,38

Figure 24 - a) Mach-Zehnder interferometer and b) Michelson Interferometer configurations using pairs of SOAs for implementing CPM wavelength conversion scheme

3.3.7. Wave Mixing

In wave mixing the idea is to construct a desired signal by using probe signals with such

wavelengths that together with the input signal they form another signal with desired

wavelength. The advantage of this approach is the transparency, and the disadvantage is the fact

that there are additional signals in the output and these signals have to be filtered out. [39,40,41] (for

further information in four wave mixing see Chapter 4)

3.4. Summary

In this chapter an overview in optical WDM components has been made. The main

characteristics and functioning principles of the active and passive components had been

described briefly. From all, the special evidence went for optical fiber, gratings, circulators,

isolators, OADMs, OXCs, EDFAs, Leds and Lasers because those have been used in this project

as will be described in Chapter 7.

Chapter 4- Signal Degradation on Optical Networks


Chapter 4 - Signal Degradation on Optical Networks

4.1. Introduction

The attractiveness of lightwave communications is the ability of silica-optical fibers to carry

large amounts of information over long repeaterless spans. To utilize the available bandwidth,

numerous channels at different wavelengths can be multiplexed on the same fiber. To increase

system margins, higher transmitter powers or lower fiber losses are required. All these attempts

to fully utilize the capabilities of silica fibers will ultimately be limited by attenuation, dispersion

and nonlinear interactions between the information-bearing lightwaves and the transmission

medium. The optical nonlinearities can lead to interference, distortion, and excess attenuation of

the optical signals, resulting in system degradations. At the system level, and with the dense

packing of channels, attention was given to another WDM impairments such as channel

crosstalk.

4.2. Attenuation

Attenuation, also known as fiber loss or signal loss, is one of the most important properties of

an optical fiber, because it leads to a reduction of the signal power as the signal propagates over

some distance. When determining the maximum distance that a signal can propagate for a given

transmitter power and receiver sensitivity, one must consider attenuation. So, attenuation largely

determines the maximum unamplified or repeaterless separation between a transmitter or a

receiver. Since amplifiers and repeaters are expensive to fabricate, install and maintain, the

degree of attenuation in a fiber has large influence on the system cost.23,42,43

Figure 25 – Optical fiber attenuation



4.3. Dispersion

Dispersion is the widening of a pulse’s duration as it travels through a fiber. As a pulse

widens, it can broaden enough to interfere with neighboring pulses (bits) on the fiber, leading to

intersymbol interference (ISI), thereby create errors in the receiver output. Dispersion thus limits

the bit spacing and the maximum transmission rate on a fiber-optic channel.42,43

Figure 26 - Intersymbol Interference (ISI)

One form of dispersion is intermodal dispersion. This is caused when multiple modes of the

same signal propagate at different velocities along the fiber. Intermodal dispersion does not

occur in a single-mode fiber.

Another form of dispersion is material or chromatic dispersion. In a dispersive medium, the

index of refraction is a function of the wavelength. Thus, if the transmitted signal consists of

more than one wavelength, certain wavelengths will propagate faster than other wavelengths.

Since no laser can create a signal consisting of an exact single wavelength, material dispersion

will occur in most systems.

A third type of dispersion is waveguide dispersion. Waveguide dispersion is caused because

the propagation of different wavelengths depends on waveguide characteristics such as the

indexes and shape of the fiber core and cladding.

At 1300 nm, material dispersion in a conventional single mode fiber is near zero. Luckily,

this is also a low attenuation window (although loss is lower at 1550 nm). Through advanced

techniques such as dispersion shifting, fibers with zero dispersion at a wavelength between

1300–1700 nm can be manufactured. In a dispersion shifted fiber, the core and cladding are

designed such that the waveguide dispersion is negative with respect to the material dispersion,

thus canceling the total dispersion. The dispersion will only be zero, however, for a single

wavelength.44



4.4. Nonlinearities

Nonlinear effects in fiber may potentially have a significant impact on the performance of

WDM optical communications systems. Nonlinearities in fiber may lead to attenuation,

distortion, and cross-channel interference. In a WDM system, these effects place constraints on

the spacing between adjacent wavelength channels, limit the maximum power on any channel,

and may also limit the maximum bit rate.

There are two categories of nonlinear effects: Kerr effects and scattering effects. The first

consists of three phenomena. In an optical fiber the core in which the optical signals travel has a

specific refractive index that determines how light travels through it. However, depending upon

the intensity of light traveling in the core, this refractive index can change. This intensity-

dependence of refractive index is called the Kerr effect. It can cause Self-phase modulation

(SPM) of a signal, whereby a wavelength can spread out onto adjacent wavelengths by itself. It

can also cause cross-phase modulation (XPM) whereby several different wavelengths in a WDM

system can cause each other to spread out. Finally, it can result in Four-wave mixing (FWM) in

which two or more signal wavelengths can interact to create a new wavelength.

There are two nonlinear scattering effects. “stimulated Raman scattering” involves light

losing energy to molecules in the fiber and being re-emitted at a longer wavelength (due to the

loss of energy). In “stimulated Brillouin scattering” light in the fiber can create acoustic waves,

which then scatter light to different wavelengths.42,43,45,46

4.5. Kerr effects

4.5.1. Self-phase modulation (SPM)

SPM is caused by variations in the power of an optical signal and results in variations in the

phase of the signal. In phase-shift-keying (PSK) systems, SPM may lead to a degradation of the

system performance since the receiver relies on the phase information. SPM also leads to the

spectral broadening of pulses, as explained below. Instantaneous variations in a signal’s phase

caused by changes in the signal’s intensity will result in instantaneous variations of frequency

around the signal’s central frequency. For very short pulses, the additional frequency

components generated by SPM combined with the effects of material dispersion will also lead to

spreading or compression of the pulse in the time domain, affecting the maximum bit rate and

the BER.42,43,45



Figure 27 - Spectral broadening due to SPM

4.5.2. Crossphase modulation (XPM)

XPM is a shift in the phase of a signal caused by the change in intensity of a signal

propagating at a different wavelength. XPM can lead to asymmetric spectral broadening, and

combined with SPM and dispersion may also affect the pulse shape in the time domain.

Although XPM may limit the performance of fiber-optic systems, it may also have

advantageous applications. XPM can be used to modulate a pump signal at one wavelength from

a modulated signal on a different wavelength. 47,48

4.5.3. Four-Wave Mixing (FWM)

FWM occurs when two wavelengths operating at frequencies f1 and f2, respectively, mix to

cause signals at 2f1-f2 and 2f2-f1. These extra signals, called sidebands, can cause interference if

they overlap with frequencies used for data transmission. Likewise, mixing can occur between

combinations of three or more wavelengths. Using unequally spaced channels can reduce the

effect of FWM in WDM systems. FWM can be used to provide wavelength conversion. 49,50,51

Figure 28 – FWM: The mixing of f1 and f2 generate sidebands



4.6. Scattering effects

4.6.1. Stimulated Raman Scattering (SRS)

SRS is caused by the interaction of light with molecular vibrations. Light incident on the

molecules creates scattered light at a longer wavelength than that of the incident light. A portion

of the light traveling at each frequency in a Raman-active fiber is downshifted across a region of

lower frequencies. The light generated at the lower frequencies is called the Stokes wave. The

range of frequencies occupied by the Stokes wave is determined by the Raman gain spectrum,

which covers a range of around 40 THz below the frequency of the input light. In silica fiber, the

Stokes wave has a maximum gain at a frequency of around 13.2 THz less than the input signal.

The fraction of power transferred to the Stokes wave grows rapidly as the power of the input

signal is increased. Under very high input power, SRS will cause almost all of the power in the

input signal to be transferred to the Stokes wave.

In multiwavelength systems, the channels of shorter wavelength will lose some power to

each of the higher wavelength channels within the Raman gain spectrum. To reduce the amount

of loss, the power on each channel needs to be below a certain level.42,43,45,46

Figure 29 - SRS transfers optical power from shorter wavelengths to longer wavelengths

4.6.2. Stimulated Brillouin Scattering (SBS)

SBS is similar to SRS except that the frequency shift is cause by sound waves rather than

molecular vibrations. Other characteristics of SBS are that the Stokes wave propagates in the

opposite direction of the input light, and SBS occurs at relatively low input powers for wide

pulses (greater than 1µs) but has negligible effect for short pulses (less than 10 ns). The intensity

of the scattered light is much greater in SBS than in SRS but the frequency range of SBS, on the

order of 10 GHz, is much lower than that of SRS. Also, the gain bandwidth of SBS is only on the

order of 100 MHz.



To counter the effects of SBS, one must ensure that the input power is below a certain

threshold. Also, in multiwavelength systems, SBS may induce cross talk between channels.

Cross talk will occur when two counterpropagating channels differ in frequency by the Brillouin

shift, which is around 11 GHz for wavelengths at 1550 nm. The narrow gain bandwidth of SBS,

however, makes SBS cross talk fairly easy to avoid.42,43,45,46

4.6.3. Conclusions on nonlinear effects

Nonlinear effects in optical fibers may potentially limit the performance of WDM optical

networks. Such nonlinearities may limit the optical power on each channel, limit the maximum

number of channels, limit the maximum transmission rate, and constrain the spacing between

different channels.

The details of optical nonlinearities are very complex and beyond the scope of this report.

They are a major limiting factor in the available number channels in a WDM system, however,

especially those operating over distances greater than a few dozen kilometers. The existence of

these nonlinearities suggests that WDM protocols that limit the number of nodes to the number

of channels do not scale well.

4.7. Crosstalk

The narrow channel spacing in dense WDM links give rise to crosstalk, which is defined as

the feedthrough of one channel’s signal into another channel. Crosstalk can be introduced by

almost any WDM component, including optical filters, wavelength multiplexers and

demultiplexers, optical switches, optical amplifiers and by the fiber itself. Crosstalk from

neighbouring inputs is a fundamental difficulty of wavelength routing which cause severe

degradation in system performance. Nonlinear crosstalk is that induced by fiber nonlinearities

refereed in previous section. Linear crosstalk can be classified into two categories, depending on

its origin.42,43,52,53,54,55

4.7.1. Heterodyne crosstalk

Optical filters and demultiplexers often leak a fraction of the signal power from neighbour

channel operating at a different wavelength that interferes with the detection process, resulting in

noise addiction on the detector. Such incoherent crosstalk is called heterodyne crosstalk or

interchannel crosstalk. Figure X shows an example of the origin of heterodyne crosstalk in a

WDM component, case of a demultiplexer.56



Figure 30 - Heterodyne crosstalk in a WDM system

4.7.2. Homodyne crosstalk

Another case of crosstalk is the Homodyne crosstalk or intrachannel crosstalk, due to is

coherent nature is far more penalizing that the Heterodyne crosstalk. This kind of crosstalk

usually occurs on wavelength routing where already exists a leakage signal at the same

wavelength due to incomplete filtering. Figure X shows an example of the origin of homodyne

crosstalk in a WDM component, case of an optical switch.56

Figure 31 - Homodyne crosstalk in a WDM system

4.7.3. Conclusions on crosstalk

The power penalties from Heterodyne and Homodyne crosstalk for WDM channels are

function or the individual crosstalk level. Figure 32 illustrates that fact for 8 and 16 WDM

channels. Here each channel contributes an equal amount of crosstalk power. This shows that the

Homodyne crosstalk effect is more severe, since it falls completely within the receiver passband.

Figure 32 -Power penalties from Heterodyne and Homodyne crosstalk for 8 and 16 WDM channels



4.8. Summary

Impairments in optical networks such as attenuation, dispersion, non-linearities due to Kerr

and scattering effects, and crosstalk have been discussed. The attenuation, is the most important

property in a optical fiber, in fact it is in the low attenuation region (near from 1550 nm) that the

WDM techniques are actually been used. Three dispersion types were also analyzed ant it was

said that it is the widening cause of an optical pulse when it travels through a fiber. Nonlinear

effects in fibers, such as the three Kerr effects and the two scattering effects also potentially limit

the performance of WDM networks. Finally a discussion in the two crosstalk types evidences the

Homodyne crosstalk effect as the worst of the impairments in WDM system devices.

Chapter 5 - Physical devices used in this project – Fiber Bragg Grating’s


Chapter 5 - Physical devices used in this project – Fiber Bragg

Grating’s

5.1. Introduction

The fiber optic filter based on Bragg gratings is the main component in this project and it

would be appropriate to introduce some of the fundamental concepts of these devices. The

following is a description of the photosensitivity effect responsible for the formation in the fiber

of permanent diffraction structures. These structures are what make this device such an effective

wavelength filter.

5.2. History of photosensitivity and fiber Bragg gratings

The first observations of refractive index changes were observed in germanosilica fibers and

were reported by Hill and co-workers at the CRC – “Communication Research Center – Canada”

in 197857. These observations were made during a test to study the non-linearities of a special

type of germanosilica fiber. While injecting visible light from a 488 nm Argon laser into the core

of the fiber the attenuation kept increasing and as a consequence of this the intensity of reflected

light from the fiber also increased. This increase in the reflectivity was a consequence of an

optical diffraction pattern that had been formed in the fiber’s core due to the photosensitivity

characteristics of the fiber.

Discovery of this characteristic of optical fibers brought forth a whole new field in optical

fiber investigation. A new era in terms of fiber optics based devices had begun. In combination

with other types of diffraction gratings, Bragg gratings permit the design of complex in-fiber

devices like resonant cavities, pass-band filters and wavelength multiplexing devices. New and

exciting structures based in fiber Bragg gratings such as optical sensors, Fabry-Perot type Bragg

gratings used in pass-band filters, non-uniform diffraction pattern gratings used in dispersion

compensation have been studied and developed58. Many more structures based on the same

principle are always appearing as a consequence of the intense investigation being developed in

this field.

Although this discovery was extremely important, after the initial discovery, the advances

were slow for almost 10 years. Diffraction gratings were initially all self-induced. This means

that they were formed without human intervention or control. This led to a lack of applicability



of these components because the only wavelength that could be reflected was that of the laser

that created the diffraction pattern initially. Lasers used at that time were Argon ion lasers at 488

nm. Some adjustments can be made through mechanical stress during the fabrication process but

this adjustment does not bring enough flexibility to the device because adjustments up to the

infra-red band (used in telecommunications applications) were not possible. Moreover, in the

blue-green wavelength regions, these gratings are intrinsically unstable owing to the continuing

photosensitivity of the fiber. This causes the grating to continually evolve during its use as a

Bragg reflector. The grating can even disappear completely if it is exposed to light of a different

blue-green wavelength. The instabilities reported and other problems with self-induced gratings

made the use of these devices impossible in optical communications or sensing applications.

5.3. Photosensitivity in optical fibers

Photosensitivity in optical fibers can be defined as the maximum refraction index change that

can be induced in an optical fiber through exposure to UV (ultra-violet) light59. A well known

characteristic of optical fibers is the high absorption verified for wavelengths in the UV band

(<300 nm). In fact, at these wavelengths the photon energy is considerable and electronic links

resonance is seen. UV exposure is a single photon process and as such, the induced index

alteration is ~6 orders of magnitude higher than the one verified in exposure with visible light.

Fibers with high germanium doping proved to be highly photosensitive. Theoretical models (still

not proven) that explain this phenomenon state that the defects introduced by the germanium in

the crystalline structure of the fiber are responsible for the achieved photosensitivity. Recently,

new techniques were developed that increase the photosensitivity of any type of fiber. This is

important because a standard fiber with germanium doping does not have a sufficient doping

profile as to obtain enough photosensitivity to inscribe the diffraction pattern correctly. Among

these techniques we have hydrogenation or hydrogen loading, flame brushing, and boron

codoping. The first technique (hydrogenation) is the most widely implemented. Hydrogen

molecules are diffused into the fiber core and penetrate the crystalline structure at high pressure

and temperatures. The presence of these molecules makes the fiber much more sensitive to

refractive index changes due to the exposure to UV light. It should be noted that the increased

fiber/waveguide photosensitivity as a result of hydrogen loading is not a permanent effect, and as

the hydrogen diffuses out, the photosensitivity decreases.

The other processes are also valid as means of increasing the photosensitivity of optical

fibers and have their advantages but hydrogenation is considered the best in terms of final results



for telecommunications applications. There are several advantages of enhancing fiber

photosensitivity though hydrogenation. The first and foremost is that it allows strong Bragg

gratings to be fabricated in any type of germanosilica fiber, including the standard

telecommunications fibers that typically have low germanium concentration, and hence, low

intrinsic photosensitivity. Second, permanent changes occur only in regions that are UV

irradiated. Finally, unreacted hydrogen in other sections of the fiber slowly diffuses out. Thus

leaving negligible absorption losses at the optical communication windows.

The physical mechanisms through which fibers get their refractive index changed are not

completely explained yet. Three models have been reported in the literature that try to describe

this phenomenon but it would become too cumbersome to describe them here in light of this

project’s objectives, but suitable references are provided for those that wish to analyze this

subject further60,61.

5.4. Fiber Bragg grating properties

5.4.1. Physical properties

In its simplest form, a fiber Bragg grating consists of a periodic modulation of the refractive

index in the core of a single-mode optical fiber (See Figure 33). These are uniform fiber gratings,

where the phase fronts are perpendicular to the fiber longitudinal axis and the grating planes are

of a constant period. They considered the fundamental building blocks for most for most Bragg

grating structures. Light guided along the core of an optical fiber will be scattered by each

grating plane; if the Bragg condition is not satisfied, the reflected light from each of the

subsequent planes becomes progressively out of phase and will eventually cancel out. Where the

Bragg condition is satisfied, the contributions of reflected light from each grating plane add

constructively in the backward direction to form a back-reflected peak with a center wavelength

defined by the grating parameters.



Figure 33 - Schematic representation of a Fiber Bragg Grating.

The reflected wavelength is given by the following equation:

Λ= 02nBλ Equation .1

The Bragg relationship in its differential form is given by:

Λ

∆Λ+∆

=∆0

0

nn

BB λλ ; Equation .2

These equations state that any measurable quantity applied to the grating that causes a

refractive index change or period change, induces a deviation in the resonant wavelength. This is

one of the key features of these devices that will be further explained later on in this report. The

sensitivity of the gratings with temperature is a consequence of thermal expansion of the silica

matrix and thermal dependence of the refractive index. Thus, for a temperature deviation T∆ ,

the correspondent deviation in wavelength is given by:

( ) TTTn

nT BBB ∆+=∆

∂∂+

∂Λ∂

Λ=∆ ξαλλλ 11 Equation .3

Where α and ξ are respectively, the thermal expansion coefficient and the thermo-optical

coefficient. In the case of silica, the thermal expansion coefficient has an absolute value of 161055.0 −−× K and the thermo-optical coefficient a value of 16100.8 −−× K . This means that, the

change in the reflected wavelength as a result of temperature variations is dominated by the



change in the refractive index. On the other hand, the mechanical stress sensitivity comes

simultaneously from deformation of the silica matrix and alteration of the refractive index due to

the photo-elastic effect. The resulting change in the resonant wavelength through mechanical

strain, for a longitudinal deformation ε∆ , is given by:

( ) ελεεε

λλ ∆−=∆

∂∂+

∂Λ∂

Λ=∆ eBBB pn

n111 Equation .4

where pe represents the photo-elastic constant of the fiber’s material. In the case of silica, this

constant has an approximate value of 0.22.

5.4.2. Spectral response of fiber Bragg gratings

As scientists strive to find a suitable model to describe the physical phenomenon behind the

formation of diffraction gratings, the flexibility of a fiber Bragg grating in terms of filtering

dictated the need to model their spectral characteristics.

Theoretical models have been established that relate the spectral dependence of a fiber

grating and the corresponding grating structure. The need to establish a model for the physical

process that relates the variation of refractive index in the fiber’s core as it is exposed to UV

radiation is associated to the mathematical description of the spatial distribution of that same

variation. The theory presently accepted describes the refractive index spatial variation in terms

of modulation of the refractive index, ),,( zyxn∆ in the fiber’s core. This theory is incomplete

and makes use of spatial coordinates. Besides this theoretical inability, the experimental

verification of the fringes contrast over the interference pattern is very difficult, particularly

when using a pulsed source. So, obtaining ),,( zyxn∆ from the spectral response of a diffraction

grating can’t be done without imposing some hypothesis and additional approximations that

introduce some limitations in the model. As such, it is supposed that the interference pattern is

perfectly sinusoidal and the fiber’s reaction to UV exposure is linear in behavior. Additionally,

the germanium concentration is considered uniform and the fringes are equally spaced. The

intensity profile of these fringes and their visibility are constant over the grating’s length.

Considering all of these restrictions, the refractive index profile is given by:

)cos()( 0 zKnnzn ⋅∆+= Equation .5

Where n0 is the mean value of the refractive index and K is the grating’s associated vector,

which is orthogonal to the index modulation planes. The amplitude is given by Λ/2π , where

Λ is the distance between consecutive planes.



Incoming light with a propagation vector Ki is deviated towards the diffraction wave vector

Kd=Ki-K. If the diffracted propagation vector matches the one of the incoming wave (forward

propagating wave), a strong Bragg diffraction occurs following Kd; otherwise, the efficiency of

the diffraction is reduced.

The wavelength for guided light within the core of a single mode fiber, that verifies this

resonance condition is given by the first order Bragg condition.

Λ= 02nBλ . Equation .6

The reflectivity of the diffraction grating is given by:

∆=B

VnLRλ

ηπ )(tanh 2 . Equation .7

In the previous equation Bλ is the resonant wavelength of the grating, L is the grating’s

length and )(Vη is the overlapping coefficient between the LP01 guided mode and the index

modulation. This equation is obtained through simplification of the equations explained later on

when the Coupled Mode Theory is explained. The reflectivity depends on two important factors:

the number of modulation planes Λ= /LN , and the modulation magnitude of the n∆ index. In

practice, the magnitude and modulation period of the refractive index are not rigorously constant

over the grating’s length. This means that the obtained value for n∆ is, in fact, the mean value of

)(zn∆ over the grating’s total length. Incoming light is partially reflected in each of the planes of

the diffraction grating. If the Bragg condition is not satisfied, the fractions of light reflected

become more and more out of phase. Balance between the total length and the exact length that

verifies the phase condition determines the width of the grating’s spectral response. The

expression that is used to calculate the Full Half Width Maximum (FHWM) of the spectral

response with these values is a first order approximation, given by:

22

0

12

+

∆=∆Nn

nsBλλ . Equation .8

The parameter s tends to unity in case of reflectivity values near 100% and tends to 5.0≈ for

low reflectivity gratings.



5.4.3. Coupled Mode Theory

The spectral response of a diffraction grating can be described using the coupled mode

theory62. It is an accurate model for obtaining quantitative information about the diffraction

efficiency and spectral dependence of fiber Bragg gratings. This is not the only theory available

but it is the most widely used and most effective.

Analysis with this theory involves the calculation of the eigen-modes of the fiber and

consequently it is used to represent a disturbance induced in the field by the refractive index

modulation. This study demands the characterization of optical fibers with small index difference

between core and cladding. As such, it is possible to use linearly polarized fields. Besides this,

two approximations are made: the absorption losses are neglected and the propagation modes are

not significantly coupled by the radiative modes. In these conditions, it can be demonstrated that

the solution to the wave equation is a pair of coupled differential equations:

ziv

v eCidz

dC β∆−+ Ω= 2 Equation .9

ziv

v eCidz

dC β∆−+− Ω= 2 Equation .10

Where the symbols + and – signal the forward-going propagation path and counter

propagation path, and

Λ−=∆ πββ Equation .11

Where β is the propagation constant and Ω is the transverse coupling coefficient, given by:

B

nλ

ηπ∆=Ω . Equation .12

Solving the set of coupled differential equations, using the normalized boundary conditions

1)0( =+C and 0)( =− LC , the final equation for the reflectivity is obtained63:

∆<ΩΩ−∆

Ω

∆>Ω+∆

Ω

=,

)(cos)(sin

,)(cosh)(sinh

)(sinh

),(22

222

22

222222

22

ββ

ββ

λpara

QLSL

paraSLSSL

SL

LR Equation .13

Where



=),( LR λ Reflectance or reflectivity as a function of Wavelength and filter length;

=λ Wavelength;

L = Filter Length;

=Ω Coupling coefficient;

;

Λ−=∆ πββ

=β Eigen propagation constant;

=Λ Perturbation period;

22 β∆−Ω=S and

iSQ =Ω−∆= 22β ;

This equation is an accurate definition of the spectral response of an uniform Fiber Bragg

Grating (FBG), but in practice, the equation used most frequently is Equation .7.

Figure 34 shows the spectral response of a diffraction grating with the following

characteristics: L = 10 mm, mµ078.1=Λ , 4102.1 −×=∆n .

Figure 34 - Spectral Response of a diffraction Grating Filter.

Reflectivity has its maximum value R= 97% for 1563 nm. This is the wavelength that obeys

the Bragg resonance condition. Multiple reflections between opposite sides of the grating lead to

the formation of sidelobes in the spectral response outside the resonance condition. These



sidelobes compromise the performance of the grating when acting as a filter and a process

known as apodization is used to eliminate these sidelobes.

5.4.4. Apodization of the spectral response of Bragg gratings

The reflection spectrum of a finite-length Bragg grating with uniform modulation of the

index of refraction is accompanied by a series of sidelobes at adjacent wavelengths as referred in

the previous section. It is very important to minimize and, if possible, eliminate the reflectivity

of these sidelobes (or apodize the reflection spectrum of the grating) in devices where high

rejection of the nonresonant light is required, as is the case with the filters used for this project.

In practice, apodization is accomplished by varying the amplitude of the coupling coefficient

along the length of the grating64. This can be seen in Figure 35. A method used to apodize the

response consists in exposing the optical fiber with the interference pattern formed by two non-

uniform UV light beams. In the phase mask technique, apodization can also be achieved by

varying the exposure time along the length of the grating, either from a double exposure or by

scanning a small writing beam or using a diffraction efficiency phase mask.

Figure 35 - Refractive index profile of an apodized Fiber Bragg Grating.

5.5. Fabrication processes

In the following, the fabrication processes through which fiber Bragg gratings are inscribed

in optical fibers will be described. Special interest will be given to the process used to inscribe

the gratings used in this project.

Inscribing diffraction patterns in a fiber is a complicated and high precision process. As

previously stated, inscription of these patterns requires a high power UV light source, centered at

244 nm, in the Germanium-Oxygen (GeO) defects or wrong bonds. Energy densities by unit area



used to efficiently write diffraction gratings have a threshold of approximately 2/150 cmmJ≈ .

These values demand the use of a UV laser source. Although there are several possibilities to

obtain laser emission at 244 nm, like the Árgon or Nd-YAG laser with, respectively, doubled and

quadrupled frequency, the choice fell in the use of a KrF excimer laser. This laser’s wavelength

is perfectly centered in the 244 nm absorption peak, meaning that the writing of gratings can be

performed directly without the need for frequency multiplication crystals. This laser emits

optical pulses with a duration of 20 ns with an energy of 100 mJ to 1.4 J per impulse. The energy

density over the transversal section of the beam is 213 cm×≈ and it is non-uniform having a

high divergence profile. This means that a set of slits and lenses must be used to obtain a

uniformly distributed energy area. In this process 70% of the initial energy is lost and the beam

must be focused over a 22.010 mm× area, corresponding to a maximum fluency of 15 J/cm2

corresponds. Writing of diffraction patterns with fluency levels of 500 mJ/cm2 demands

exposure for min5≈ . This corresponds to a total radiation dosis of 7.5 kJ/cm2.

Up to date there are only a few externally written fabrication techniques, namely, the

interferometric technique, the point-by-point technique and the phase mask technique. The latter

was the technique used in this project. Next, a brief description of these fabrication processes

will be made with particular emphasis to the phase mask technique.

5.5.1. Interferometric technique

The interferometric technique was the first to be developed. It was demonstrated by Meltz65.

It uses an interferometer that splits UV light from a source into two beams and then recombines

them in order to form an interference pattern. This pattern is used to expose a photosensitive

optical fiber thus creating a refractive index change in the fiber’s core. Several types of

interferometers are used with this technique each one having its specific characteristic giving the

process different capabilities. This interferometric process is very flexible because it allows the

creation of gratings with any central resonant wavelength without the need for different sources.

It also allows making Bragg gratings with any desired length.

The main disadvantage of the amplitude-splitting interferometric technique is it’s

susceptibility to mechanical vibrations. Displacements as small as submicrons in the position of

mirrors, beam splitter, or mounts in the interferometer can cause the fringe pattern to drift,

washing out the grating. In addition to the above short-coming, quality gratings can only be



produced with a laser source that has good spatial and temporal coherence with excellent output

power stability.

Figure 36 - Setup for interferometric fabrication of Fiber Bragg Gratings.

5.5.2. Point-by-point technique

Another technique used is the point-by-point technique that is accomplished by inducing a

change in the refractive index a step at a time along the core of the fiber66. Each grating plane is

produced separately by a focused single pulse from an excimer laser. A single pulse of UV light

from the laser passes through a mask containing a slit. A focusing lens images the slit onto the

core of the optical fiber from the side, as shown in Figure 37, and the refractive index of the core

in the irradiated fiber section increases locally.

The fiber is then translated though a distance Λ corresponding to the grating pitch in a

direction parallel to the fiber axis and the process is repeated to form the grating structure in the

fiber core. Essential to the point-by-point fabrication process is a very stable and precise

submicron translational system. The main advantage of this technique lies in it’s flexibility to

alter the Bragg grating parameters. Because the grating structure is built up a point at a time,

variations in grating length, grating pitch and spectral response can easily be incorporated. One

disadvantage of this technique is that it is a tedious process due to its step by step nature. Error in

the grating spacing due to thermal effects and/or small variations in the fiber strain can occur.

This limits the gratings to a very short length making the fabrication of high reflectivity gratings



difficult. As such, this is not an adequate process to fabricate the grating to be used in this

project.

SubMicron translation device

Pulsed UV beam

Point-by-point fabricated grating

SubMicron translation device

Pulsed UV beam

Point-by-point fabricated grating

Figure 37 - Setup for Point-by-point Fabrication Technique.

5.5.3 – Phase mask technique The last technique to be addressed is, most certainly, the most used and one of the most

effective methods for inscribing gratings in photosensitive fiber. This technique employs a

diffractive optical element (phase mask) to spatially modulate the UV writing beam (See Figure

38)67. Phase masks may be formed holographically or by electron-beam lithography. The mask

has an interference pattern written on it and the UV light passing through the mask and getting to

the fiber core will photoimprint a refractive index modulation with fringe period one-half that of

the mask.

Use of this process greatly reduces the complexity of the fiber grating fabrication system.

The simplicity of using only one optical element provides a robust and an inherently stable

method for reproducing Fiber Bragg Gratings. Since the fiber is usually placed directly behind

the phase mask in the near field of the diffracting UV beams, sensitivity to mechanical vibrations

and, therefore, stability problems are minimized. Low temporal coherence does not affect the

writing capability (as opposed to the interferometric technique) due to the geometry of the

problem.



Figure 38 - Setup for Phase-Mask fabrication of Fiber Bragg Gratings.

5.6. Applications of Fiber Bragg Gratings

A brief reference should be made to the other applications where Bragg gratings are used

5.6.1. Laser stabilization

The wavelength of laser diodes is sensitive to temperature fluctuations. Fiber Bragg gratings

can be applied to stabilize the diode wavelength with respect to temperature. By inserting a

length of fiber with a fiber Bragg grating, the length of the cavity can be made to dwarf the scale

of the temperature fluctuations, rendering the diode immune to such changes68.

5.6.2. Fiber lasers

Historically it has been very difficult to produce high quality high power lasers at 1550 nm.

The design of grating based fiber lasers falls into two basic categories, end-pumped and ring

lasers. The first uses a 980nm laser coupled to a length of fiber doped with erbium. The pump

laser excites the erbium ions in a manor similar to an EDFA. A cavity is defined in the doped

fiber with Fiber Bragg Gratings. The ring laser schematically consists of a loop of fiber

connected by a pump laser and containing a length of erbium doped fiber. The cavity is defined

though the use of a coupler and a single grating. This second structure will be used in our work

and will be described in Chapter 769.

5.6.3. Reflectors in fiber amplifiers

Bragg gratings can be used to flatten the gain profile of the EDFA. By using special gratings

called Side Tap Gratings, the excess gain can be pushed down. Side Tap Gratings do not reflect



light back down the core of the fiber, like normal gratings, but instead couple light into the

cladding of the fiber, where it is dispersed and lost.

5.6.4. Raman-Shifted Lasers and Raman Amplifiers

Raman-shifted lasers and Raman amplifiers enable efficient conversion of short-wavelength

light into longer wavelengths suitable for long-distance fiber transmission. Raman gain is

obtained through energy transfer from pump light to the laser output or amplified signal as

mediated by molecular vibrations in the silica fiber70.

5.6.5. Sensors

In a non-communications related example, Fiber Bragg Gratings can be used in sensors. The

area of embedded sensors in composite materials to detect strain in static structures is

extensively studied and has a lot of commercial applications.71,72

5.6.6. Isolation Filters in Bidirectional Lightwave Transmission

With a very effective reflection property, fiber grating filters can be used as isolation filters.

They block adjacent channels and far-end crosstalk.

5.6.7. WDM Demultiplexers

The grating’s property of deflecting light incident upon it with an angle that depends on the

wavelength of that light can be used to demultiplex WDM signal into their different channels.

5.6.8. Add/Drop Multiplexers and Optical Cross-Connects

These structures are fundamental in communications systems using WDM and are fully

described in Chapter 673.

5.6.9. Dispersion Compensators and Wavelength Converters

Using chirped fiber Bragg gratings with a special refractive index profile, the chromatic

dispersion that induces significant distortion on optical pulses can be reduced74.



Figure 39 - Dispersion compensation using an aperiodic grating with a length of 6cm associated with an

optical circulator.

5.7. Summary

In this chapter, a main component of WDM networks has been presented – Fiber Bragg

Gratings. Their physical characteristics and spectral response have been presented.

Fiber Bragg Gratings are vastly used in various types of applications. The main applications,

fundamentally telecommunications applications where presented. There where many more to

present but we kept to the most important in the context of our work.

Gratings are a fundamental component in wavelength selective networks; they are, in some

technologies integrated in Wavelength Routers. These devices are the core of Optical

Wavelength Selective Networks and will be described in the next chapter.

Chapter 6– Wavelength Routers


Chapter 6 – Wavelength Routers

State of the art in Optical Switching and Routing

6.1. Introduction

A Wavelength Router is a device used in WDM networks. It has one or more input and

output ports through which it is connected to other wavelength routers and/or end-nodes by using

one or more fiber links for each neighbor. Wavelength routers should be able to route signals on

different wavelengths at different input ports to (possibly) different output ports independent of

the signals on other input ports and on other wavelengths. According to the routing matrix

present there are four major types of wavelength router architectures: Add-Drop Multiplexers,

Optical Cross-Connects, Static Wavelength Routers and Reconfigurable Wavelength Routers.

In this project the first two architectures are the ones implemented and as such, a description

of the state of the art in terms of these devices is suitable at this point. These devices, in the

context of optical networks in general, have been briefly described above in Chapter 3. Here a

description of some of the technologies used in terms of optical switching and routing is made.

6.2. Optical Add-Drop Multiplexers – OADMs

These devices perform a simple but very important task in terms of optical network operation

as described in Chapter 2. Optical add–drop multiplexers (OADMs) will be required in future

wavelength-division multiplexed (WDM) ring and bus networks to link the network with local

transmitters and receivers. OADMs can also provide interconnection between network

structures75.

These devices are generally evaluated in terms of performance through crosstalk

measurements. Although there are other problems related with these devices like losses, non-

linear effects, etc… Crosstalk is the most limiting factor in terms of performance. It arises in

OADMs through component imperfections. Optical crosstalk at the same wavelength as the

transmitted signal is generally referred to as homodyne or in-band crosstalk. It is particularly

serious because it cannot be removed by filtering76,77, and has been shown to severely limit

network performance. Within homodyne crosstalk, incoherent crosstalk causes rapid power

fluctuations, while coherent crosstalk changes the optical power of the signal78. Incoherent

crosstalk occurs in ring and bus networks when the signal and interferer are from different



optical sources. It causes power fluctuations at the receiver, resulting in bit error rate (BER)

degradation and a power penalty. Coherent crosstalk in these networks is due to multiple paths

between ports within the OADMs. It causes variable attenuation levels between OADM ports. So

long as the level variation is slower than the decision threshold at the receiver, the effect of

coherent crosstalk is generally to reduce the size of the eye, without increasing eye closure.

There is no accompanying power penalty, because the BER is measured against the optical

power at the receiver. Incoherent and coherent crosstalk together give a range of possible power

penalties, because coherent crosstalk can cause variation in both signal and incoherent crosstalk

powers at the receiver79. Combination of coherent and incoherent crosstalk leads to a range of

possible BERs and power penalties for OADMs deployed in a network link.

6.2.1. Comparison of common OADM structures

A brief description of common OADM structures will be made here. In Figure 40 some

structures are presented. Only the most common OADM structures are referred, Bragg gratings

and arrayed waveguide grating multiplexers (AWGMs). These two technologies have been

selected because they are widely published, cover a range of proposed OADM structures, and are

comparatively mature. Among the other technologies, we have ring resonators80, acousto-optic

tunable filters81, and micromirror arrays82,83 and finally, the technology used in this project:

tunable gratings84,85. This last technology will not be analyzed here because it will be described

thoroughly in Chapter 1.



Figure 40 - Common OADM structures: FB-a), FB-b) and FB-c) – Interferometric structures based in Fiber Bragg gratings; FB-d), FB-e) and FB-f) – Fiber Bragg grating and circulator based structures; FA-a), FA-b) and FA-c) – Array Waveguide Grating Mux structures

The structures in the figure where initially developed as Fixed-Wavelength structures, this

means that they deal only with wavelengths for which they are initially tuned; there is no

possibility of tunability during functioning.

The first set of structures (from a) to c)), uses interferometric processes as Bragg gratings to

select the path that the optical signal must follow. Although having good crosstalk

characteristics, these structures use couplers which is a big problem because it causes a lot of

attenuation especially to dropped channels. Besides this, interferometric systems incur in a

specific problem called back-reflection that can severely harm the light sources if they are placed

directly at the input of these devices. Slow tunability (as referred earlier in this report about

Mach-Zehnder interferometers)

In the second set of structures (d), e) and f)) we have approaches using gratings, circulators

and couplers. Structure d) is very simple but very poor in performance too. Is exhibits high

attenuation due to the use of the coupler, and high levels of crosstalk. Structure e) is a refinement

of the previous structure where attenuation problems are solved by introducing a new circulator.

Crosstalk levels are still relatively high but are almost completely solved using structure f) in

which an isolator is used to prevent crosstalk problems inside the diffraction gratings. Our

OADM structure, described later in this report is very similar to this one but adds tunability

though temperature control of the gratings. The remaining three structures, FA-a to FA-c, are



based on AWGMs. These OADMs operate quite differently to the Bragg grating-based

structures. Rather than separating out the channel to be dropped, the entire WDM signal is

demultiplexed by the AWGM(s). Each channel traverses a separate path through the OADM

before being remultiplexed. The path for the channel be added and dropped is broken to provide

ADD and DROP ports. Structures FA-b and FA-c are low-crosstalk variations of structure FA-

a86.

Comparing the fixed-wavelength OADMs it can be said that structures using Bragg gratings

generally have lower insertion losses than those based on AWGMs. They also generally have

fewer coherent crosstalk terms, although the coherent crosstalk levels are low for all structures.

The incoherent crosstalk levels show no consistent differences by technology.

The four low-crosstalk design variations do all offer significantly better incoherent crosstalk

performance as expected, at the cost of a higher component count, or more complex components.

Combining multiplexing and demultiplexing functions in single AWGMs leads to more

leakage paths and greater coherent crosstalk.

There are two general trends in the crosstalk performance of the OADMs briefly described

here. The first is that low crosstalk levels (incoherent and coherent) are generally associated with

either a higher component count, or more complex components.

It should be referred that introducing tunability in these structures will increase the crosstalk

levels but overall performance in terms of flexibility is definitely increased.

Bragg-grating and AWGM-based structures can offer excellent homodyne crosstalk

performance. However, the Bragg grating based structures are superior overall, because of their

reduced insertion loss and better filtering characteristics.

The AWGM-based structures optically filter all channels at each OADM, whereas those

using Bragg gratings only add filtering at the OADMs in the network where the channel is added

and dropped. Filtering from AWGM cascades has been shown to cause signal distortion and eye

closure87.

An upcoming technology that uses gratings is the Multiport approach. In these structures

(Figure 41)88, optical circulators with multiple ports are used in order to reduce the insertion loss

due to the use of several circulators. The reduction of component count helps normalize the

attenuations imposed on different channels. But there are also setbacks. The multiport circulators



are yet to be proven effective when many ports become necessary, (i.e.), they do not support

scalability.

Figure 41 - OADMs based in Multiport Optical Circulators (MOC’s) Configurations.

As previously denoted, there are other solutions to implement these devices. It would be

impossible to describe them all in this report and maintain an introductory structure, but suitable

references are provided and further study can be conducted through those articles.

6.3. Optical Cross-Connects - OXCs

An optical cross-connect (OXC), operates directly in the optical domain, just like the

described OADMs, but, instead of just adding or dropping channels, it switches the input

channels into different output fibers.

Just as electrical switching replaced mechanical relays of the past, optical switching is on the

verge of replacing some of today’s electrical switching functions in telecommunications

networks. Although new approaches to optical switching are constantly being developed, optical

switch designs can be roughly classified into seven categories: optomechanical, thermo-optical,

liquid crystal, micro-electrical mechanical, gel/oil based, electro-optical, and others such as

acoustooptic, semiconductor optical amplifier (SOA) and ferro-magnetic.

In evaluating the performance of the different optical switches, the following individual

technology appraisals include assessments of reliability, energy usage, port configurations and

scalability, optical insertion loss, cross-talk, temperature resistance, and polarization dependent

loss characteristics.



6.3.1. SWITCHING TECHNOLOGIES

6.3.1.1. OPTOMECHANICAL

Optomechanical switches employ electromechanical actuators to redirect a light beam. One

type of optomechanical switch inserts and retracts a reflective surface into a light stream to

redirect it to another port. Another architecture redirects the light stream by bending a grating-

written fiber.

In terms of optical insertion loss and switching speed, performance characteristics of

optomechanical switches vary according to architecture; performance can range from low to high

loss and slow to fast speed. However, the universal drawback for optomechanical switches is the

durability and cycle limitation of the mechanical actuators. This type of structure found its

success in MEMS that are described bellow.

6.3.1.2. MICRO-ELECTRO-MECHANICAL SYSTEM (MEMS)

MEMS can be considered a subcategory of optomechanical switches; however, because of

the fabrication processes and miniature natures, they have different characteristics, performance

and reliability concerns. MEMS use tiny reflective surfaces to redirect the light beams to a

desired port by either ricocheting the light off of neighboring surfaces to a port, or by steering

the light beam directly to a port89. Analog-type, or 3-D, MEMS mirror arrays have reflecting

surfaces that pivot about axes to guide the light. Digital type, or 2-D, MEMS have reflective

surfaces that “pop up” and “lay down” to redirect the light beam propagating parallel to the

surface of the substrate. The reflective surfaces’ actuators may be electrostatically-driven or

electromagnetically-driven with hinges or torsion bars that bend and straighten the miniature

mirrors.

MEMS devices easily scale to large port counts because of miniature sizes and

semiconductor fabrication processes, but due to the density and microscopic size of the light

paths entering the substrate, MEMS can be a challenge to package.

Although highly accepted by the industry, these MEMs also have some problems. In order to

switch the different wavelengths, an effective spatial division of all the wavelength channels

must be made and each wavelength must have its own switching plane (a set of mirrors that

make it possible for it to be routed across the device).



Figure 42 - Toshiyoshi and Fujita’s 2×2 MEMS optical switch .

Figure 43 - 8x8 2-D Optical switch

Figure 44 - Schematic of 3-D MEMS switching



Figure 45 - 3-D micromachined mirrors.

Figure 46 - Scratch Drive actuator switching 1x2.

6.3.1.3. THERMO-OPTICAL

Planar lightwave circuit thermo-optical switches are usually polymer-based or silica on

silicon substrates. Thermo-optical switches use temperature control to change index of refraction

properties of Mach-Zehnder interferometer-based waveguide arms on the substrate. The light is

processed by waveguide interaction and is guided through the appropriate path to the desired

port90.

Thermo-optical switches are small in size but have a drawback of having high driving-power

characteristics and issues in optical performance.

A typical 2X2 thermal-optic switch has an insertion loss of <2.5 dB, an extinction ratio of

>35 dB (for two cascaded switches), and a switching speed of 1 – 3 ms. The switch can be

constructed using a Mach-Zehnder configuration (Fig. 1), which consists of two, 3 dB couplers

connected by two waveguides serving as phase shifters. Thin film resistors are deposited on the

Mach-Zehnder arms, so that one of the arms can be heated to change its refractive index, and the

accumulated phase difference of light propagating through the two arms can be modulated.

When light is launched into one of the input ports, it is split into two MZ arms by the 3 dB

coupler with equal optical power and p/2 phase difference. As light travels through the MZ arms,

the phase difference can be altered due to the temperature difference between the two

waveguides. After passing through the second 3 dB coupler, the two beams recombine either

constructively or destructively at either of the two output ports, depending upon the exact phase

difference between the two Mach-Zehnder arms controlled by the heater. This modulation of

temperature achieves the purpose of switching the light between the two output ports. The



electrical power needed to switch each path is on the order of a few hundred milliwatts. Switches

can also be cascaded for added extinction ratio without sacrificing much on insertion loss.

Figure 47 - Thermooptical switching - a)Schematic diagram of a Mach Zehnder switch ; b) Light path in one of the switching.

6.3.1.4. LIQUID CRYSTAL

Liquid crystal switches work by processing polarization states of the light. Apply a voltage

and the liquid crystal element allows one polarization state to pass through. Apply no voltage

and the liquid crystal element passes though the orthogonal polarization state. These polarization

states are steered to the desired port, are processed, and are recombined to recover the original’s

signal properties. With no moving parts, liquid crystal is highly reliable and has a good optical

performance, but can be affected by extreme temperatures if not properly designed.

Its drawbacks are essentially three. It is fairly slow (especially at low temperatures, where

switching times can be hundreds of milliseconds); is difficult to integrate with other optical

components; and has relatively high light losses from the liquid crystal itself, the polarization

splitters, and imperfections in the fairly complex optical path.

One of the most challenging aspects of applying liquid crystals to optical switching directly

relates to their use of polarization. The optical polarization of any input signal is completely

uncontrolled. Therefore, the signal must be split into two known orthogonal polarizations using

polarization splitters and switching done separately on each. The results are then recombined to

form the output. This approach is troublesome and costly to implement and could cause

unacceptable polarization mode dispersion (PMD), in which short pulses are spread out in time

because different components of the pulses propagate at different speeds, depending on their



polarization. In addition, compensating for the liquid crystal’s temperature dependence renders it

too costly for all-optical switching needs in the metro and access networks91,92.

6.3.1.5. GEL/OIL-BASED – “Bubble” Switching

Index-matching gel and oil-based optical switches can be classified as a subset of thermo-

optical technology, as the switch substrate needs to heat and cool to operate. However, their

exists an added dimension in that heating a portion of the switch causes an index of refraction

changed atmosphere to form at the waveguide junctions. This index of refraction changed

“bubble” or liquid redirects the light stream through the appropriate waveguide path to the

desired port.

Total internal reflection—known as TIR, the phenomenon that makes light propagate down

an optical fiber—can, with an added twist, also serve as the basis of a switch. The way the

principle works, if light attempts to cross from a medium of higher refractive index (Dielectric 1)

to one of lower refractive index (Dielectric 2) at too shallow an angle, all of the light is reflected

from the interface back into the high-index medium [see top left part of figure below]. The trick

to exploiting the phenomenon in a switch is to turn the effect off (or on) by replacing (or not

replacing) the second medium with one whose index of refraction matches that of the first. The

best-known product based on this phenomenon is the Agilent Champagne switch, in which

sections of waveguide intersect with fluid-filled channels [see bottom part of figure below].

There are inherent losses in this type of structure, and even worse, these losses cause

crosstalk. To minimize these detrimental effects, the intersection should be kept as small as

possible.

Designers of TIR switches are therefore faced with a pair of conflicting requirements: low

loss must be traded off against high isolation. An additional problem in TIR switches of this type

is that the reflected wave undergoes a wavelength-dependent phase shift because of energy

storage in the bubble. This causes amplitude variations and dispersion in the switch’s output,

lowering its usefulness for some applications.

Note that because this is a matrix switch, the number of intersections equals the product of

the number of inputs and the number of outputs. As mentioned above, each intersection traversed

by the light contributes to the loss and crosstalk, limiting the scaling of the matrix to less than

100 ports because the number of intersections to be crossed by a light beam in the worst case

may equal the total number of ports.



This technology has been compared to proven inkjet printer technology and can achieve good

modular scalability. However, for telecom environments, uncertainty exists about gel/oil-based

long-term reliability, thermal management and optical insertion losses93.

Figure 48 - Total internal reflection switching – Agilent’s Champagne Bubble Switch.

6.3.1.6. ELECTRO-OPTICAL

Electro-optical switches use highly birefringent substrate material and electrical fields to

redirect light from one port to another. A popular material to use in an electro-optical switch is

Lithium Niobate. An electrical signal is fed as the control into the substrate of the device. This

electrical field changes non-isotropically the substrate’s index of refraction. The index of

refraction change manipulates the light through the appropriate waveguide path to the desired

port. Opto-electrical switches are extremely fast and are reliable, but they pay the price of high

insertion loss and possible polarization dependence.

6.3.1.7. ACOUSTO-OPTIC

Acousto-optic optical switches receive acoustic-wave-induced pressure from a RF-fed

piezoelectric transducer to generate fine gratings in optical waveguides. The gratings diffract

lights to the desired port.

The Acoustooptic Tunable Filter (AOTF)94,95 (Figure 49), utilizes TM and TE polarization

modes in a birefringent optical waveguide in LiNbO3. In AOTF, after passing the first polarizing

beam splitter, all signals have the same polarization. For the signals at selected wavelength(s),



corresponding RF drive signal(s) are injected in to a polarization converter device. Finally,

selected signals are dropped by the second polarizing beam splitter device. The AOTF can be

used by a passive combiner device (to add new signals) to construct a multi-wavelength

configurable add-drop multiplexer.

Figure 49 - Acoustooptic Tunable filter

6.3.1.8. ELECTROHOLOGRAPHIC

Electroholography is the newest all-optical switching technology. This method features a

solid-state switch matrix created from rows and columns of ferroelectric crystals such as lithium

niobate or potassium lithium tantalite niobate [Figure 50]. Rows correspond to individual fibers,

and each column is for a different wavelength. Each crystal is laser etched with a Bragg grating

(which causes a quasi-periodic modulation in its dielectric properties) to create a hologram in

which the crystal’s optical properties are changed when it is energized, for example, by the

application of an electric field. In current implementations, such as those by Trellis Photonics,

individual crystals are manually assembled, and thus must be greater than 1 mm on a side. As the

technology evolves, the holographic elements may be able to be written more densely into a

single crystal; then patterning will be required only for the electrodes through which the

energizing electric fields are applied to each crystal or holographic element. When a crystal is

not energized, light goes through it. Energized crystals, on the other hand, deflect a controllable

portion of the incident light to the appropriate fiber. Holographic switches are quite fast and

claim instant signal restoration. They, along with other switches made from electro-optic

materials, will be fast enough for the long-term application of optical packet switching. Because

it is an emerging technology, no data about its long-term reliability is available, but past

holographic applications like high-density storage have shown lifetime issues with the holograms

themselves. On the plus side, electroholographic switches may be easily integrated with other

network functions like equalization and monitoring. Being electrostatically controlled, they

consume negligible power. The technique allows a single crystal to be used for switching and for



variable attenuation, since the fraction of light reflected is controllable by an applied signal. Yet,

from an application viewpoint, the technology is not the ideal solution it is sometimes

represented to be. The approach is that of a wavelength-selective matrix switch. The hologram

blocks are analogous to the mirrors in a 2-D MEMS switch. The number of matrix elements in

an electroholographic switch, therefore, increases as the product of the number of input and

output ports, and will not scale well. As the switch matrix size is increased to the sizes needed

for core network switching, the required optical beam size will expand and optics for collimating

and focusing the beams will be required. Non-energized blocks in the optical path will contribute

to the loss and crosstalk of the switch. Also, holograms are diffractive elements that are

inherently polarization and wavelength dependent, leading to dispersion and polarization-

dependent loss (PDL) issues.

Figure 50 - Electroholographic Matrix with ferro-electric crystals

6.3.1.9. BRAGG GRATING BASED

Although many times not mentioned in literature, this technology has great potential in the

optical switching domain. The main idea is to use optical filters based in fiber Bragg gratings in

conjunction with optical circulators to appropriately direct the light between input and output

fibers. The work done in this report is based in this technology but there are some

implementation studied mainly in academic environments because the industry is more inclined

to the MEMs approach (inspite of its complexity and cost). Bragg gratings, as described in

Chapter 5, are especially suited to filter wavelengths and as such, present good crosstalk

capabilities. As seen with OADM structures, the overall performance of this type of devices is

very good.



It is unfortunately true that commercial versions of OXCs using this technology are not

available and that high port count are only proven to work in theoretical studies, but research is

being made, this project being an excellent example, in order to make this technology proliferate

and have better commercial acceptance.

In Figure 51, an example of a scalable OXC can be seen. An even better solution is presented

in the patent submitted as a result of this project.

Figure 51 - A 4x4 rearrangeable nonblocking OXC using: a) 12 three-port Optical Circulators and b) Four five-port Multiport Optical Circulators (MOC’s).

Some of the problems associated with this specific configuration are stated in the patent and

how our solution can improve overall performance.

Fiber Bragg grating technology is very mature and has very good temperature stability

besides having easy coupling to fiber and low loss96.

6.3.2. Comparison of OXC technologies

Switching Technology CROSSTALK INSERTION LOSS

POWER CONSUMPTION SPEED WAVELENGTH

DEPENDENCEPOLARIZATION DEPENDENCE COST SCALABILITY

MEMS SMALL LOW LOW ms SMALL SMALL HIGH LARGEThermo-Optical LARGE MODERATE HIGH ms SMALL SMALL MEDIUM SMALLLiquid Crystal LARGE HIGH MODERATE < ms LARGE SMALL HIGH MODERATE

Bubble (Gel/Oil) LARGE HIGH MODERATE ms HIGH MODERATE HIGH SMALLElectro-Optical LARGE MODERATE MODERATE ns - ms SMALL or LARGE LARGE HIGH SMALLAcousto-Optic MODERATE HIGH MODERATE micros LARGE SMALL MEDIUM SMALL

ElectroHolographic LARGE HIGH LOW ms LARGE LARGE HIGH SMALLBragg Grating SMALL LOW MODERATE s SMALL NONE SMALL MODERATE

The only conclusion that can be taken from this analysis is that switching technologies are

numerous and suitable for different types of applications. In the long run there won’t be any sole

winners. Each technology will have its own market niche.



6.4. Summary

The most important technologies used in the fabrication of Optical Add-drop Multiplexers

(OADMS) and Optical Cross-connects (OXCS) have been showed and a performance evaluation

has been made. As stated above, none of these technologies is vastly superior to another. They

all have their applicability in different types of networks with different requirements. Our work

focused in one of these technologies, namely, Fiber Bragg Grating Based. In the following

Chapter, the practical work developed in our project will be described and the performance of

some architectures will be examined.

Chapter 7– Work Developed


Chapter 7 – Work Developed

7.1. Introduction

The practical implementation of this project could only be accomplished after careful

planning and study of the possible architectures. In order to make the results systematic, a step-

by-step approach was used. The steps followed where:

Grating fabrication – Fabrication of high reflectivity and low insertion loss gratings

Development of an Add-drop Multiplexer (OADM – 1)

Development of a temperature controller to tune the grating

Structure assembly – OADM 1

Performance evaluation and testing

Upgrade of the OADM developed – OADM 2

Structure assembly – OADM 2


Development of an Optical Crossconnect (OXC)

Development of additional WDM Laser Sources to place at the input ports

Structure assembly – OXC 1

Performance test

Development of an upgraded OXC


7.2. Grating Fabrication

A key component to the success of this work was the fabrication of gratings with good

characteristics such as: high reflectivity, low insertion loss, and other important characteristics.

The gratings used where fabricated using the Phase Mask technique. This allowed the use of

a low spatial and temporal coherence KrF excimer laser at 248 nm. A piece of standard single

mode fiber (SMF) is illuminated with UV light coming from the laser and passing through the



phase mask creating a diffraction pattern in the fiber (This was previously explained in Chapter

5).

The fiber had been previously kept under high-pressure hydrogen atmosphere in order to

enhance its photosensitivity due to hydrogen diffusion into the glass matrix. This process

described in Chapter 5, helps to reduce the writing time necessary to obtain high reflectivity

gratings. The phase mask used had a spacing period of Λ= 1072 nm. The laser was fired with a

power level of 300mW during approximately 15 minutes to fabricate this grating.

The spectral response of the resulting grating is presented in Figure 52.

1548 1549 1550 1551 1552 1553 15540,0

0,2

0,4

0,6

0,8

1,0

Tran

smiss

ion

(dB

)

Ref

lect

ivity

λ (nm)

-16

-14

-12

-10

-8

-6

-4

-2

0

Figure 52 - Transmission and reflection grating spectra.

The resulting spectral response shows approximately 100% reflectivity at the wavelength of

1550.9 nm. The Full Width Half Maximum (FWHM) is 0.2 nm which is very good for a grating

with this reflectivity. This grating is a little wide in terms of bandwidth occupied but it would be

possible to obtain better results if apodized masks where used. There is a performance trade-off

between reflectivity and spectral width. However, considering that the WDM channel spacing is

0.8nm (100GHz) this is not a severe problem in terms of crosstalk.

A further problem is excessive loss in the shorter wavelength side of the main peak in the

transmission spectrum. This is due to coupling from the waveguide mode to the backward-

propagating cladding mode. This may be overcome by suppressing the coupling itself and by

shifting the region in which the excessive loss occurs outside the waveband being used. In either

case it is achieved by modifying the profile of the fiber [97].



7.3. Development of an Optical Add-drop Multiplexer

7.3.1. Implementation of the first structure – OADM 1

The idea behind this part of the project was to develop a fixed OADM with good

performance characteristics. A schematic of the first structure that was developed is presented in

Figure 53.

(λ1,λ2,λ3)

Input Output

λ2

Drop Add

Signal A

Signal C

Signal B (λ1,λ3)

(λ2)

Figure 53 – First Optical Add-drop structure implemented - OADM 1.

This structure selectively filters a channel with wavelength λ2 from the input signal composed

of three channels of wavelengths: λ1=1549.9 nm, λ2=1550.7 nm and λ3=1551.5 nm, separated by

0.8 nm (100 GHz). This input signal was obtained through optical spectral slicing of a LED’s

emission spectra. This source is briefly explained in Figure 54.

PCGPIB

BroadbandOptical Source

Coupler

Optical Spectrum Analyser

λ1 λ2 λ3

50:50

Figure 54 – Optical WDM source obtained through slicing of a Broadband optical source’s (LED) spectra.

It should be noted that the power injected in the fiber by this source is low and reflection in

the gratings followed by the coupling loss, will reduce even further the power levels of the

resulting signal. This is not detrimental in the analysis of our OADM but better results could be

obtained using sources with more significant power levels.



7.3.2. Performance Assessment

The results obtained are shown in Figure 55 where the input signal composed of three WDM

channels is represented together with the output and the dropped channel. We can see that the

Insertion Losses in this first structure are small, in the order of 1dB.

1548 1549 1550 1551 1552 1553 1554-25

-20

-15

-10

-5

0

A B C

Tran

smis

sion

(dB

)

λ (nm)

Figure 55 – A – Input signal composed of three WDM channels (l1, l2,l3); B– Output Signal composed of signals l1 and l3 ; C – Dropped Channel l2 .

Crosstalk is the main performance evaluation parameter for these devices and can be of two

types: Heterodyne or Interchannel Crosstalk (Figure 56) – Derives from interferences of small

power levels that appear outside the channel’s bandwidth, causing an increase in the bit error rate

when detecting the other channels; Homodyne or Intrachannel Crosstalk – Results from

interferences inside the channel’s bandwidth[98].

Figure 56 – Heterodyne Crosstalk caused by imperfect FBG filtering.

In the performance tests on the OADM structures, only Heterodyne Crosstalk has been

tested.



The Crosstalk isolation level, i.e., the difference between the power level of the channel and

the interference from adjacent channels (Crosstalk), is 14.5 dB. This is not a spectacular result

but having in consideration that this is a simple configuration where one does not intend to

reduce the crosstalk this is actually an acceptable value.

In order to increase the isolation level, a new configuration was developed – OADM 2. This

new structure is shown if Figure 57.

7.3.3. Upgraded OADM Structure – OADM 2

(λ1,λ2,λ3)

Input Output

λ1

(λ1,λ3)

λ3

Drop Add

Signal A Signal B λ2 λ2

Optical Isolator

Signal C (λ2)

Figure 57 - Second Optical Add-drop structure implemented - OADM 2

In this structure, new gratings where added in the drop output in order to remove any

crosstalk residues still present in the output signal. The isolator in the central arm between the

circulator prevents any signal residues coming from signals being added from interfering with

the signal being dropped or signal just following the direct path though the OADM.


The results of this new and more complex structure are shown in Figure 58. In this figure we

can see the dropped channel λ2 using the first OADM and that same dropped channel using the

second structure. The Insertion loss has increased by 2 dB but the Crosstalk isolation has

increased 5 dB. The noise seen in the Figure is the noise floor of the equipment and it is not

entirely present in the signal. The fact is that the Optical Spectrum Analyzer (OSA) does not

have enough resolution to read the effective crosstalk level present in the signal which is very

low.



1548 1549 1550 1551 1552 1553 1554-25

-20

-15

-10

-5

0

C D

Tran

smis

sion

(dB

)

λ (nm)

Figure 58 - C – Dropped Channel λ2 in OADM 1; D – Dropped Channel λ2 in OADM 2.

7.4. Optical Crossconnect Architectures

7.4.1. Description of first structure and its implementation

The next step in the project’s objective was to assemble an Optical Crossconnect. The first

structure to be presented is the one shown in Figure 59. This structure is patented [99] and referred

in our own patent has one of the recent developments in this area. It has relatively good

performance and it is a good starting point to study this type of OXCs.

λ1, λ2 , λ3

λM

Output 1

Output 2

Input 1

Input 2

Figure 59 – Optical Crossconnect 1

As described above in Chapter 5, Bragg gratings can be tuned to reflect different

wavelengths strain or temperature change of their physical characteristics. Now, instead of only

stabilizing the grating so that the central wavelength is fixed, the approach was to heat the

grating with a Peltier Device in order to tune it to the desired wavelength thereby selecting

which channel is dropped and which channel is added. This means that the grating is used as a



Tunable Optical Filter. Description of the developed control electronic circuit is given in

Appendix B.

Another Fiber Bragg Grating was fabricated with a central wavelength of 1550.7 nm,

FWHM of 0.2 nm and near 100% reflectivity. The phase mask used had a spacing period of Λ=

1070 nm. The laser was fired yet again with a power level of 300 mW during approximately 15

minutes to fabricate this grating. The WDM source was the same used in the OADM testing.


When the tunable optical filter is stabilized at room temperature, the central wavelength is

λM=λ2, therefore switching channel 2 to Output 1. The other two channels (1 and 3) are switched

to Output 2.

In Figure 60 it can be seen that the crosstalk isolation level of channels λ1 and λ3 is -16.23 dB

(Output 2). The small residual components centered in λ1 and λ3, are due mostly to residual

reflections in the grating and circulators that give birth to Heterodyne Crosstalk.

1548 1549 1550 1551 1552 1553 1554-35

-30

-25

-20

-15

-10

-5

0

Output 1 Output 2

Opt

ical

Pow

er (d

B)

λ (nm)

Figure 60 – Optical Crossconnect 1 performance test.

As a second step in the analysis of this structure the optical filter (FBG) is detuned by

temperature variation and the central wavelength is placed at an intermediate wavelength

between λ2 and λ3 (λM=1551.1 nm). In Figure 61 we can see the power level of the signals in

both exits of the device. The three channels are switched to Output 2. In Output 1 the signal

measured has a power level 10.53 dB bellow the one at Output 2. This signal is reflected from

the grating and causes Homodyne Crosstalk. It should be noted once more that the use of

gratings with an apodized refractive index profile will reduce significantly the measured

crosstalk levels.



1548 1549 1550 1551 1552 1553 1554-35

-30

-25

-20

-15

-10

-5

0

Output 1 Output 2

Opt

ical

Pow

er (d

B)

λ(nm)

Figure 61 - Power Spectral Response of the OXC in a detuned state.

It is important to note that there are inherent losses due to the use of circulators. Each time a

signal enters a circulator and exits in the next port it suffers from a 1 dB loss to which are added

0.2 dB of losses due to fiber splices. Each signal incurs twice in these losses, which explains why

the maximum power levels at the Outputs are bellow the reference level by 2.4 dB.

One of the disadvantages of this device is the physical impossibility of placing signals with

the same wavelength in contiguous ports and effectively switching them. The grating would have

to be detuned and the conflicting signals would mix destroying all information in them.

This structure is nevertheless fully scalable (with the mentioned limitation), i.e., it is possible

to build NxN port Crossconnects using basic 2x2 OXC blocks although the described setback is

still a problem.

Even though in this structure only one FBG is used, there is a possibility of controlling more

input channels by placing other FBGs next to the one present.

7.4.3. Development of an upgraded OXC

The most significant achievement of this project was a novel Optical Crossconnect

architecture that will now be described. This structure is under proper procedures in order to be

patented. In Figure 62 a diagram of the OXC – OXC 2 is presented. Unlike the previous structure

presented (OXC 1) this novel structure has no limitations in terms of channel insertion. We can

insert any type of wavelength in any of the input ports and there is always a way to successfully

switch these wavelengths through the Crossconnect. The optical filters (FBGs) are still

controlled by the previously referred temperature controller based on a Peltier Device.



Input 1 Output 1

Input 2

Output 2

λΜ

λΜ

Signal A

Signal B

(λ2,λ3)

(λ1)

Figure 62 - OXC 2.


In the next set of Figures, we will present the signals obtained in the performance assessment

of this configuration. In order to evaluate the performance more effectively new optical sources

where used. A tunable laser with good power output was applied to Input 2 with central

wavelength with center at λ1 = 1548.8 nm. At Input 1 a Fiber Laser was used to create a signal

with two wavelengths λ2 =1549.6 nm and λ3 = 1550.4 nm. This fiber laser is built as depicted in

Figure 63.

EDFA

APC

FBGs

80:20

Circulator

PC

Losses

λ2

λ3

Figure 63 - Multiwavelength Fiber Ring Laser Source using Fiber Bragg Gratings.

The optical signals from these sources are depicted in Figure 64.



1546 1548 1550 1552-60

-50

-40

-30

-20

-10

0

Input Port 1 Input Port 2

Tran

smis

sion

(dB

m)

λ(nm)

Figure 64 - Input Signals

As a first step in this test, the optical filters (FBGs) where detuned from any of the input

wavelengths and the signals at the corresponding Outputs where measured. The results of this

test are shown in Figure 65.

1546 1548 1550 1552

-60

-50

-40

-30

-20

-10

0

Output 1 Output 2

Tran

smis

sion

(dB

m)

λ(nm)

Figure 65 - Output port signals with optical filters detuned.

As seen in this Figure, the wavelength channels are properly sent directly to their

correspondent outputs, i.e., Output 1 for channel 2 and 3; Output 2 for channel 1. The small peak

at 1548.8 nm that appears in the signal form Output port 1 is Heterodyne Crosstalk and has a

power level of -45 dBm which is fairly negligible. Insertion Losses are calculated with respect

to the Input signals and are in this case 1.31 dB. These losses are originated in the circulators and

fiber splices and grating (optical filter) imperfections. The Interchannel Crosstalk Isolation Level



is 30.72dBm. The Homodyne Crosstalk Isolation Level is 30.89dBm. These are very good

isolation levels even when comparing this device with commercial ones.

In the next step, the optical filters (FBGs) where tuned to λ1 = 1548.8 nm in order to switch

channel 1 from Input port 2 to Output port 1. The results are in Figure 66.

1546 1548 1550 1552-80

-70

-60

-50

-40

-30

-20

-10

0

Output 1 Output 2

Tran

smis

sion

(dB

m)

λ(nm)

Figure 66 - Outputs when Channel 1 is switched from Input 1 to Output 2.

As seen in this result, Channel 1 is effectively switched to Output port 1. Due to the increase

in the reflections in optical filters and also an increase in the number of times a signal has to

enter an optical circulator (thereby suffering from insertion losses) the total Insertion Losses in

this case have increased to 4 dB. In Output port 2 a -38.2 dBm Crosstalk level is seen. The

Homodyne and Heterodyne Crosstalk Isolation Levels are, respectively 20.4 dB and 21.4 dB.

These are very good results and point to the good performance of this device.

Another performance issue to be addressed is the tuning speed. The temperature control

gives stable results but only slow tunability is achieved. This can be seen in the Figure 67 where

the tuning and detuning speed of the optical filter (FBG) is seen.



0 5 10 15 20 25 30 35 40

-20

-15

-10

-5

0

FBG tuning FBG detuning

Tran

smis

sion

(dB

)

Time (sec)

Figure 67 - Tuning and detuning speed.

When tuning, i.e., reaching a specified wavelength in order to reflect it, the achieved speed is

around 10 seconds. But it should be noted that detection of the desired channels begins before

the stable wavelength is reached. As the temperature that reaches the grating increases, the

central wavelength reflected by the grating changes until it reaches the desired wavelength. To

show the stability of this temperature control two figures are presented next. In Figure 68 we can

see the power level of the Crossconnect’s Output signal as the optical filter is being tuned.

According to the wavelength that appears in the Output port, the power level increases or

decreases. The important detail to notice is the fact that, once in a specific wavelength, the

device maintains stability. This can be noticed much easily in Figure 69.



0 20 40 60 80 100 120 140 160 180 200 220 240-10

-9

-8

-7

-6

-5

-4

-3

-2

-1

0

Tran

smis

sion

(dB

)

Time (sec.)

Figure 68 - Tuning and detuning of the optical filter (FBG).

0 20 40 60 80 100 120 140 160 180 200 220 2401550,4

1550,5

1550,6

1550,7

1550,8

1550,9

1551,0

1551,1

1551,2

λ (n

m)

Time (sec.)

Figure 69 - Wavelength tuning and detuning of the optical filter (FBG).

The optical filter (FBG) seems to be a limiting component in all the configurations presented.

The filter used, although having near 100% reflectivity, does not entirely block incoming signals.

It blocks signal reducing their level by a maximum of 15 dB. Better filters could effectively be

obtained but would come to the expense of wider spectral bandwidth that would cause

interference with adjacent channels. This setback could be avoided if the apodized masks for

fabrication where available.



7.5. Summary

In this chapter, a description of the practical implementation of our project was made. An

introductory section about the grating fabrication process was presented and subsequently, the

structures implemented where described and their performance tests where shown.

In the next Chapter final conclusions about our work are given and justified.

Chapter 8– Concluding remarks


Chapter 8 – Concluding remarks

In this report we reviewed the fundamental aspects of this project and a theoretical

introduction about its area of interest – Optical Networks – is made. This enabled us to be aware

of the environment that surrounds this area of research and its importance in the

telecommunications scenario.

We reviewed the most important advances in optical networks and their evolution through

successive generations up to the present time. The paths taken to overcome the main problems of

the first two generation are described, in particular the electronics bottleneck problem.

After that, a review of the Multiplexing solutions possible is presented, namely, Optical Time

Division Multiplexing – OTDM, Optical Code Division Multiplexing – OCDM and finally,

Wavelength Division Multiplexing – WDM, which is the most pervasive optical technology

nowadays. Optical Components and network impairments are also described (Chapters 3 and 4).

Fiber Bragg Grating structures are reviewed in Chapter 5 and the state-of-the-art in Optical

Switching is presented.

Finally, attention is given to the practical implementation of the devices developed in the

project. The designed structures and performance tests made are described. As a final conclusion

we can say that the OXC has very good performance although different approaches to its control

could be made.

The Optical Crossconnect configurations, specially the last one described has very good

spectral characteristics which results in low homodyne and heterodyne crosstalk levels which

subsequently permits high cascading levels bringing this 2x2 structure to generalized NxN port

configurations.

In order to improve the spectral characteristics of the filters used, thereby reducing the

crosstalk levels and improving the performance of our solution, apodized grating filters should

be used. Unfortunately there were no facilities to fabricate apodized grating filters but reference

to this fact will be made later on in this report.

The long run goal of this project was to develop good insight on the performance

characteristics and specific impairments of this type of technology and integrate it with routing

protocols, namely, GMPLS – Generalized Multiprotocol Label Switching, that enables the

Chapter 8– Concluding remarks


upgrade of this device to a fully functional Optical Router with completely optical interfaces,

removing the O-E-O conversion which gives birth to the Electronics Bottleneck. This is an area

of intense research at this moment and several solutions to the underlying physical architecture

are in development. Nonetheless, we think that this technology has very good perspectives in

terms of market implementation because of its very good price/quality ratio. This means that it

has visible cost benefits considering the performance levels reached.

Considering other technologies this is the only one where wavelength and spatial switching is

performed at the same time. The wavelength selection, that is, splitting of the WDM channels in

a multiwavelength signal is done in a manner that is inherent to the device itself, because

gratings are, by nature, wavelength selective filters and select the channels in the composite

signal. Other technologies demand the use of prisms or other type of wavelength splitting

technologies to be used and this comes at the expense of excessive losses and switching

complexity. For example in MEMs technologies, much discussed at this time, each wavelength

has to use its specific wavelength switching plane. This increases the complexity and cost of this

solution and severely affects its flexibility.

There is a possibility of controlling the tuning of the optical filters (FBGs) through strain

using piezoelectric control and obtain higher tuning speeds but this also has its setbacks due to

the still necessary temperature control of the gratings which will complicate the control circuitry

and stability requirements.

The mentioned integration with routing protocols is one of the objectives for the other part of

this project, developed in another research unit (UTM – Telecommunications and Multimedia

Unit), which we believe was carried out also with great success.

References


References 1 P. S. André, J. L. Pinto, I. Abe, H. J. Kalinowski, O. Frazão, F. M. Araújo, “Fiber Bragg Grating for Telecommunication Applications: Tunable Thermally Stress Enhanced OADM”, Journal of Microwaves and Optoelectronics, Vol. 2, No. 3, July 2001. 2 C.A Brackett, “Dense wavelength division multiplexing networks: principles and applications”, IEEE Journal on

Selected Areas in Communications, Volume: 8 Issue: 6 , Aug. 1990, pp. 948 -964 3 P.E Green Jr., “Fiber Optic Communications”, Englewood Cliffs, Prentice Hall, 1993 4 Mestdag, D.J.G., “Fundamental of Multiaccess Optical Networks”, Norwood, MA, Artech House, 1995 5 M.S. Borella, J.P. Jue, D. Banerjee, B. Ramamurthy, B. Mukherjee, “Optical Components for WDM Lightwave

Networks", Proceedings of the IEEE, vol. 85, no. 8, pp. 1274-1307, August 1997. 6 M.J. Yadlowsky, E.M. Deliso, V.L. Silva, “Optical Fibers and Amplifiers for WDM Systems", Proceedings of the

IEEE, vol. 85, no. 11, pp. 1765-1779, November 1997. 7 P.E. Green, “Optical Networking Update", IEEE Journal on Selected Areas in Communications, vol. 14, no. 5, pp.

764-779, June 1996. 8 D. Cotter, J.K. Lucek, D.D. Marcenac, "Ultra-High-Bit-Rate Networking: From the Transcontinental Backbone to

the Desktop," IEEE Communications Magazine, pp. 90-95, April 1997. 9 M.E. Marhic, “Coherent Optical CDMA Networks," Journal of Lightwave Technology, vol. 11 no. 5/6, pp. 729-

735, May/June 1993. 10 C.A. Brackett, A.S. Acampora, J. Sweitzer, G. Tangonan, T. Smith, W. Lennon, K. Wanga, R.H. Hobbs, “ A

Scalable Multiwavelength Multihop Optical Network: A Proposal for Research on All-Optical Networks," Journal

of Lightwave Technology, vol. 11 no. 5/6, pp. 736-753, May/June 1993. 11 J.M. Senior, M.R. Handley, M.S. Leeson, “Developments in Wavelength Division Multiple Access Networking,"

IEEE Communications Magazine, pp. 28-36, December 1998. 12 R. Ramaswami, “Multiwavelength Lightwave Networks for Computer Communication," IEEE Communications

Magazine, pp. 78-88, February 1993. 13 J. M. Elmirghani, H.T. Mouftah, “Technologies and Architectures for Scalable Dynamic Dense WDM Networks”.

IEEE Communications Magazine, 38(2):5866, February 2000. 14 L. Xu, H. Perros and G. Rouskas, “Techniques for Optical Packet Switching and Optical Burst Switching”. IEEE

Communications Magazine, vol. 39, no.1, pp. 136-142, January 2001. 15 B. Mukherjee, “WDM-Based Local Lightwave Networks Part I: Single-Hop Systems", IEEE Network, pp. 12-27,

May 1992. 16 B. Mukherjee, “WDM-Based Local Lightwave Networks Part II: Multihop Systems", IEEE Network, pp. 20-32,

July 1992. 17 A. Ganz, B. Li, “Implementation Schemes of Multihop Lightwave Networks", Journal of Lightwave Technology,

vol. 11 no. 5/6, pp. 979-986, May/June 1993. 18 M. Veeraghavan, R. Karri, T. Moors, M. Karol and R. Grobler, “Architectures and Protocols that Enable New

Applications on Optical Networks”. IEEE Communications Magazine, vol.39, no. 3, pp. 118-127, March 2001.

References


19 C. Brackett, “Dense Wavelength Division Multiplexing Networks: Principles and Applications", IEEE Journal on

Selected Areas in Communications, vol. 8, no. 6, pp. 948-964, August 1990. 20 O. Gerstel, “On The Future Of Wavelength Routing Networks", IEEE Network, pp. 14-20, November/December

1996. 21 E. Karasan, E. Ayanoglu, “Performance of WDM Transport Networks", IEEE Journal on Selected Areas in

Communications, vol. 16, no. 7, pp. 1081-1096, September 1998. 22 J.M.H. Helmirghani, H.T. Mouftah “Technologies and Architectures for Scalable Dynamic Dense WDM

Networks”, IEEE Communications Magazine, vol.38, no. 2, pp. 58-66, February 2002. 23 Gerd Keiser, “Optical Fiber Communications”, Mc Graw Hill, 2000 24 D. Brooks, S. Ruschin, “Integrated electro-optic multielectrode tuneable filter”, Journal of Lightwave Technology,

Vol.13. no. 7, pp. 1508-1513, July 1995 25 D.J.Mestdagh, “Fundamentals of Multiaccess Optical Fiber Networks”, Artech House, 1995 26 A. Sneh, K.M. Johnson, “High-Speed tunable liquid crystal filter for WDM networks”, Journal of Lightwave

Technologies, Vol. 14, no. 6, June 1996 27 P.L. Chen, K.C. Lin, W.C. Chuang, Y.C. Tzeng, K.Y. Lee, W.Y. Lee, “Analysis of a liquid crystal Fabry-Perot

etalon filter: a novel model”, IEEE Photonics Technologies Letters, Vol. 9, no. 4, April 1997 28 D.A. Smith, J.E. Baran, J.J. Johnson, K.W. Cheung, “Integrated-optic acoustically tuneable filters for WDM

networks”, IEEE Journal of Selected Areas in Communication, Vol.8, no.1, pp. 1151-1159, January 1994 29 D.A. Smith, “Evolution of the acousto-optic wavelength routing switch”, Journal of Lightwave Technology,

Vol.14, no. 6, pp. 1005-1019, June 1996 30 K. Saitoh, M. Koshiba, Y. Tsuji, “Numerical analisys of integrated acousto-optic tuneable filters with weighted

coupling”, Journal of Lightwave Technology, Vol.17, no. 2, pp. 249-254, February 1999 31 M.J. Yadlowsky, E.M. Deliso, V. L. DA Silva, “Optical Fibers and Amplifiers for WDM Systems”, Proceedings of

the IEEE, Vol. 85, no. 11, November 1997 32 “Raman Amplification Design in Wavelength Division Multiplexing (WDM) Systems”,

http://www.iec.org/online/tutorials/raman/index.html, 2001 33 R. Ramaswami, K. N. Sivarajan, “Optical Networks, A Practical Perspective”, Morgan Kaufmann Series in

Networking, Morgan Kaufmann Publishers, 1998. 34 http://www.lightreading.com 35 M.S. Borella, J. P. Jue, D. Banerjee, B. Ramamurthy, B. Muckerjee, “Optical Components for WDM Lightwave

Networks”, Proceedings of the IEEE, Vol. 85, no. 8, August 1997 36 P.M.Henderson, “Introduction to Optical networks”, Mindspeed Technologies, Inc., Conexant Systems, Inc. 37 K. Hobermann, S.Kindt, D.Breuer, K. Ptermann, C. Schmidt, S. Die, H.G. Weber, “Noise characteristics of

semiconductor optical amplifiers used for wavelength conversion via cross-gain and cross-phase modulation”, IEEE

Photonics Technologies Letters, Vol. 9, no. 3, pp. 312-314, , March 1997 38 S.L. Danielsen, P.B. Hansen, K. E. Stubkjaer, M. Schilling, K. Wunstel, W. Idler, P. Doussiere, F. Pommerau,

“All optical wavelength conversion for increased input power dynamic range”, IEEE Photonics Technologies

Letters, Vol. 10, no. 1, pp. 60-62, , January 1998

References


39 A.D. Ottavi, E. Iannone, A. Mecozzi, S. Scotti, P. Spano, R. Dall’Hara, J. Eckner, G. Guekos, “Efficiency and

noise performance of wavelength converters based on FWM in semiconductor optical amplifers”, IEEE Photonics

Technologies Letters, Vol. 7, no. 4, pp. 357-359, , April 1995 40 D. F. Geraghty, R.B. Lee, K.J. Vahala, M. Verdiell, M. Ziari, A. Mathur, “Wavelength conversion up to 18 nm at

10Gb/s by four-wave mixing in a semiconductor optical amplifier”, IEEE Photonics Technologies Letters, Vol. 9,

no. 4, pp. 452-454, April 1997 41 J.M. Yates, J.P.R. Lacey, M.P. Rumsewicz, M.A. Summerfield, “Performance of networks using wavelength

converters based on FWM in semiconductor optical amplifiers”, Journal of Lightwave Technology, Vol.17, no. 5,

pp. 782-791, May 1999 42 Green, P.E., Jr., “Fiber Optic Networks”, Prentice Hall, 1993. 43 D.J.Mestdagh, “Fundamentals of Multiaccess Optical Fiber Networks”, Artech House, 1995. 44 M.S. Borella, J. P. Jue, D. Banerjee, B. Ramamurthy, B. Muckherjee, “Optical Components for WDM Lightwave

Networks”, Proceedings of the IEEE, Vol. 85, no. 8, August 1997. 45 www.lightreading.com 46 A. R. Chraplyvy, “Limitations on lightwave communications imposed by optical-fiber nonlinearities”, Journal of

Lightwave Technology, Vol. 8, no. 10, pp. 1548-1557, October 1990. 47 K. Hobermann, S.Kindt, D.Breuer, K. Ptermann, C. Schmidt, S. Die, H.G. Weber, “Noise characteristics of

semiconductor optical amplifiers used for wavelength conversion via cross-gain and cross-phase modulation”, IEEE

Photonics Technologies Letters, Vol. 9, no. 3, pp. 312-314, , March 1997. 48 S.L. Danielsen, P.B. Hansen, K. E. Stubkjaer, M. Schilling, K. Wunstel, W. Idler, P. Doussiere, F. Pommerau,

“All optical wavelength conversion for increased input power dynamic range”, IEEE Photonics Technologies

Letters, Vol. 10, no. 1, pp. 60-62, January 1998. 49 A.D. Ottavi, E. Iannone, A. Mecozzi, S. Scotti, P. Spano, R. Dall’Hara, J. Eckner, G. Guekos, “Efficiency and

noise performance of wavelength converters based on FWM in semiconductor optical amplifers”, IEEE Photonics

Technologies Letters, Vol. 7, no. 4, pp. 357-359, April 1995. 50 D. F. Geraghty, R.B. Lee, K.J. Vahala, M. Verdiell, M. Ziari, A. Mathur, “Wavelength conversion up to 18 nm at

10Gb/s by four-wave mixing in a semiconductor optical amplifier”, IEEE Photonics Technologies Letters, Vol. 9,

no. 4, pp. 452-454, April 1997. 51 J.M. Yates, J.P.R. Lacey, M.P. Rumsewicz, M.A. Summerfield, “Performance of networks using wavelength

converters based on FWM in semiconductor optical amplifiers”, Journal of Lightwave Technology, Vol.17, no. 5,

pp. 782-791, May 1999. 52 K. P. Hoo, J.M. Kahn, “Methods for crosstalk measurement and reduction in dense WDM systems”, Journal of

Lightwave Technology, Vol. 14, no. 6, pp. 1127-1135, June 1996. 53 C.S. Li, F. Tong, “Crosstalk and interference penalty in all-optical networks using static wavelength routers”,

Journal of Lightwave Technology, Vol. 14, no. 6, pp.1120-1126, June 1996. 54 S. Geckeler, “Crosstalk penalties in bi-directional Fiber-Optic WDM System”, IEEE Journal in Selected Areas in

Communications, Vol. 8. no.6, pp. 1115-1119, August 1990. 55 S.D. Dods, R.S. Tucker,” A Comparison of the Homodyne Crosstalk Characteristics of Optical Add–Drop

Multiplexers”, Journal of Lightwave Technology, Vol. 19, no. 12, pp.1120-1126, pp. 1828-1838, December 2001.

References


56 P.S.André, J.L.Pinto, A.N. Pinto, T. Almeida, “Performance Degradations due to crosstalk in Multiwavelength

Optical Networks Using Optical Add Drop Multiplexers Based on Fiber Bragg Gratings”, Revista do DETUA,

Vol.3, no. 2, pp. 85-90, Setembro 2000. 57 K. O. Hill, Y. Fujii, D. C. Johnson and B. S. Kawasaki, “Photosensitivity in optical waveguides: Application to reflection filters fabrication”, Appl. Phys. Lett. 32(10), 647 (1978). 58 C. R. Giles, “Lightwave Applications of Fiber Bragg Gratings”, IEEE J. Lightwave Technol., vol. 15, NO. 8, 1391-1403, August 1997. 59 K. O. Hill and G. Meltz, “Fiber Bragg Gratings Technology and Overview”, IEEE J. Lightwave Technol. Vol. 15, No.8, 1263-1276, August 1997. 60 Andreas Othonos, “Review of Scientific Instruments: Fiber Bragg Gratings”, vol.68, NO.12, December 1997. 61 Raman Kashyap, “Fiber Bragg Gratings”, Ed. Optics and Photonics 1999. 62 T. Erdogan, “Fiber Grating Spectra”, IEEE J. Lightwave Technol. Vol. 15, No.8, 1277-1294, August 1997. 63 D. K. W. Lam and B. K. Garside, “Characterization of Single-mode Optical Fiber filters”, Applied Optics, Vol.20, No. 3, February 1981. 64 G. Meltz, W. W. Moorey and W. H. Glenn, “Formation of Bragg gratings in optical fibers by a transverse holographic method." Opt. Lett. 14, pp. 823, 1989. 65 Hill K. O., “Aperiodic distributed-parameter waveguides for integrated optics”, Applied Optics 13, 1853-1856, 1974. 66 B. Malo, K. O. Hill, F. Bilodeau, D. C. Johnson and J. Albert, “Point-by-point Fabrication of Micro-Bragg Gratings in Photosensitive Fiber Using Single Excimer Refractive Index Modification Techniques”, Electronics Letters, Vol. 29, No. 18, September 1993. 67 D. Z. Anderson, V. Mizrahi, T. Erdogan and A. E. White, “Phase-Mask Method for Volume Manufacturing of Fiber Phase Gratings”, Post-deadline paper PD16, Technical Digest of Post-Deadline Papers Proc. Conf. On Optical Fiber Communications, OFC ’93, p. 68, 1993. 68 B. F. Ventrudo, G. A. Rogers, G. S. Lick, D. Hargreaves and T. N. Demayo, “Wavelength and Intensity Stabilization of 980nm Lasers Coupled to Fiber Bragg Gratings”, Ellectro. Lett. 30(25), pp. 2147-2149, 1994. 69 F. N. Timofeev, P. Bayvel, V. Mikhailov, P. Gambini, R. Wyatt, R. Kashyap, M. Robertson, R. Campbell and J. E. Midwinter, “2.5 Gbit/s dense WDM Transmission in Standard fiber Using Fiber Grating Lasers”, Electron. Lett. 33(19), 1632-1633, 1997. 70 A. R. Chraplyvy “Limitations on lightwave communications imposed by optical-fiber nonlinearities,” J. Lightwave Technology, vol. 8, pp. 1548-1557, Oct. 1990. 71 O. Frazão, F. M. Araújo, L. A. Ferreira, J. L. Santos, “Sensores de Bragg em fibra Óptica: Aplicações em engenharia”. 72 W. W. Morey, G. Meltz and W. H. Glenn, “Fiber Optic Bragg Grating Sensors”, SPIE 1169, Fiber Optics Sensors VII, pp. 98-107, 1989. 73 Shin-Kuei liaw, Keang-Po Ho, Chinlon Lin, sien Chi, “Experimental Investigation of Wavelength Tunable WADM and OXC Devices Using Strain Tunable Fiber Bragg Gratings”, Optics Commun. 169, pp. 78-80, 1999. 74 P. A. Krug, T. Stephens, G. Yoffe, F. Oullette, P. Hill and G. Doshi, “Dispersion Compensation Over 270 Km at 10Gb/s using an Offset-Core Fibre Bragg Grating”, Electron. Lett. 19, pp. 877-879, 1994.

References


75 H. J. R. Dutton, “Understanding Optical Communications, 1st Ed.”, Raileigh, 1998. 76 R. Ramaswami and K. Sivarijan, “Optical Networks: A Practical Perspective”, Morgan Kaufmann, 1998. 77 E. L. Goldstein, F. Elrefaie, “Performance Implications Of Component Crosstalk in Transparent Lightwave Networks”, IEEE Photon. Technol. Lett., Vol. 6, pp. 657-660, 1994. 78 J. Zhou, J. O’Mahony and S. D. Walker, “Analysis of Optical Crosstalk Effects in Multi-Wavelength Switched Networks”, IEEE Photon. Technol. Lett., Vol. 6, pp. 302-305, 1994. 79 S. D. Dodds and R. S. Tucker, “Homodyne Crosstalk in WDM Ring and Bus Networks”, J. Lightwave Tecnol., Vol. 9, pp. 1285-1287, 1997. 80 W. Weierhausen and R. Zengerle, “Photonic Highway Switches based on ring resonators used as Frequency-Selective Components”, Appl. Opt., Vol.35, pp. 5967-5978, 1996. 81 M. Fukutoku, K. Oda and H. Toba, “Wavelength-Division-Multiplexing Add/Drop Multiplexer employing a Novel Polarization Independent Acousto-Optic Tunable Filter. 82 N. A. Riza and S. Sumriddetchkajorn, “Fault-tolerant dense multiwavelength add-drop filter with a two-dimensional digital micromirror device,” Appl. Opt., vol. 37, pp. 6355–6361, 1998. 83 J. E. Ford, V. A. Aksyuk, D. J. Bishop, and J. A. Walker, “Wavelength add-drop switching using tilting micromirrors,” J. Lightwave Technol., vol. 17, pp. 904–911, 1999. 84 S. Y. Kim, S. B. Lee, S. W. Kwon, S. S. Choi, and J. Jeong, “Channelswitching active add/drop multiplexer with tunable gratings,” Electron. Lett., vol. 34, pp. 104–105, 1998. 85 P. Leisching, B. H., A. Richter, D. Stoll, and G. Fischer, “Optical add/drop multiplexer for dynamic channel routing,” Electron. Lett., vol. 35, pp. 591–592, 1999. 86 H. Takahashi, O. Ishida, K. Oda, and H. Toba, “Anticrosstalk arrayedwaveguide add-drop multiplexer with foldback paths for penalty free transmission,” Electron. Lett., vol. 30, pp. 2053–2055, 1994. 87 C. Caspar, H.-M. Foisel, R. Freund, U. Krüger, and B. Strebel, “Cascadability of arrayed-waveguide grating (de)multiplexers in transparent optical networks,” in Proc. Optic. Fiber Commun. (OFC’97), Dallas, TX, 1997. 88 An Vu Tran, Wen De Zhong, R. Tucker and R. Lauder, “Optical Add-Drop Multiplexers with Low Crosstalk”, IEEE Photon. Technol. Lett., Vol. 13,No.6, pp. 582-584, June 2001. 89 Tze-Wei Yeow, K. L. Eddie Law, and Andrew Goldenberg, “MEMS Optical Switches”, IEEE Commun. Mag., November 2001. 90 Wenjie Chen, Ph.D. and Alice Liu, Ph.D. “Understanding the Principles of Thermal-optic Switching” Lightwave Microsystems Corporation. 91 Saleh e Teich, Fundamentals of Photonics, Wiley Series, 1994. 92 Ming, Max e Liu, Kang, Principles and Application of Optical Communications, McGraw-Hill, 1996. 93 http://www.labs.agilent.com/news/2000features/fea_optswitch.html - The Agilent “Champagne Switch” 94 C. Bracket, “Dense Wavelength Division Multiplexing Networks: Principles and Applications”, IEEE Journal on Selected Areas in Communications, vol. 8, no. 6, pp. 948-964, August 1990. 95 A. d’Alessandro, D. A. Smith, and J. E. Baran, “Multichannel operation of an integrated acousto-optic wavelength routing switch for WDM systems,” IEEE Photon. Technol. Lett., vol. 6, pp. 390–393, 1994.

References


96 An Vu Tran, When de Zhong, R. Tucker and Kai Song, “Reconfigurable Multichannel Optical Add-Drop Multiplexers Incorporating eight-Port Optical Circulators and Fiber Bragg Gratings”, IEEE Photon. Technol. Lett., Vol. 13,No.10, pp. 1100-1102, October 2001. 97 Ikuo Ota and Toshiaki Tsuda, “Development of Optical Fiber Gratings for WDM Systems”, Furukawa Review, No. 19, 2000. 98 P. S. André, J. L. Pinto, A. N. Pinto, T. Almeida, “Performance Degradations due to Crosstalk in Multiwavelength Optical Networks Using Optical Add Drop Multiplexers Based On Fiber Bragg Gratings”, Revista do DETUA, Vol.3, No. 2, Setembro 2000. 99 M. G. Oberg, “Optical NxN Wavelength Crossconnect – U. S. Patent No. 5.940.551”, August 1999.

Appendix A - Peltier Devices 1

Appendix A - Peltier Devices

A. Introduction

Peltier devices, also known as thermoelectric (TE) devices, are small solid-state

devices that work as heat pumps. A "typical" unit is a few millimeters thick by a few

millimeters to a few square centimeters. It is a sandwich formed by two ceramic

plates with an array of small Bismuth Telluride cubes ("couples") in between. When a

DC current is applied heat is moved from one side of the device to the other - where it

must be removed with a heat sink. The "cold" side is commonly used to cool an

electronic device such as a microprocessor or a photodetector. If the current is

reversed the device makes an excellent heater.

B. Peltier & Seebeck Effects

The cooling property of these devices is due to the Peltier Effect, while the electrical

power generating property is due to the Seebeck Effect. A thermoelectric device can

be used as either a cooler or a power generator, but not with the best efficiency.

Peltier Effect devices are almost always constructed with Bismuth Telluride (Bi2Te3)

and used around room temperature and below. Seebeck Effect power generators are

often constructed of PbTe or, SiGe as well as Bi2Te3 and are used at much higher

temperatures.

In theory, the Peltier effect is explained the following way: electrons speed up or slow

down under the influence of contact potential difference. In the first case the kinetic

energy of the electrons increases, and then, turns into heat. In the second case the

kinetic energy decreases and the joint temperature falls down. In case of usage of

semiconductors of p and n types the effect becomes more vivid. On Figure 1 you can

see how it works.

Figure 1 - Usage of semiconductors of p and n-type in thermoelectric coolers


Combination of many pairs of p and n semiconductors allows the creation of cooling

units - Peltier devices of relatively high power (Figure 2). Has we have previously

seen, thermoelectric cooling couples (Figure 1) are made from two elements of

semiconductor, primarily Bismuth Telluride, heavily doped to create either an excess

(n-type) or deficiency (p-type) of electrons. Heat absorbed at the cold junction is

pumped to the hot junction at a rate proportional to current passing through the circuit

and the number of couples.

Figure 2 - Structure of a Peltier device

In resume, a Peltier device (Figure 3a)) consists of semiconductors mounted

successively, which form p-n and n-p junctions. Each junction has a thermal contact

with radiators. When switching on the current of the definite polarity, there forms a

temperature difference between the radiators: one of them warms up and works as a

heat sink, the other works as a refrigerator.

Figure 3 – a) Peltier device; b) Peltier cascade

A typical device provides a temperature difference of several tens degrees Celsius.

With forced cooling of one of the hot radiators, the other one can reach temperatures


below 0 ºC. For more significant temperature differences the cascade connection is

used. (Figure 3 b)) The cooling devices based on Peltier devices are often called

active Peltier refrigerators or Peltier coolers. Peltier device's power depends on its

size. Low power devices might not be efficient enough. But the usage of the devices

of too high power might cause moisture condensation, what is dangerous for

electronic circuits.

Thermoelectric devices are not the solution for every cooling problem. However, you

should consider them when your system design criteria include such factors as high

reliability, small size or capacity, low cost, low weight, intrinsic safety for hazardous

electrical environments, and precise temperature control. i,ii,iii,iv,v

C. Competitive Advantages

It is possible to build thermoelectric systems in a space of less than 1 cubic inch.

More typically, thermoelectric systems occupy about 20 to 30 im3 of space. These

systems are energized by a DC power input. In addition to the space and weight

saving advantages, thermoelectrics offer the utmost in reliability due to its solid-state

construction. Another feature of importance is the ease with which a thermoelectric

can be precisely temperature controlled, which is an important advantage for

scientific, military and aerospace applications.

D. Technical aspects

Peltier device cooling & heating speed - they can change temperature extremely

quickly, but to avoid damage from thermal expansion control the rate of change to

about 1ºC per second.

These devices can heat as well as cool - they make great heaters. Just reverse the

polarity of the power supply. But, be sure not to exceed the temperature rating of the

devices, usually 80ºC for standard models to 200ºC for high temperature models.

The power supply requirement is a simple DC supply, if the AC ripple is not more

than about 10% or 15%. The devices specified Vmax should not be exceeded.

E. Peltier in Optical Communications

Peltier Devices have encountered widespread use in the temperature stabilization of

optoelectronic components (lasers, switches, detectors, etc.) in high speed and

wavelength division multiplexed (WDM) fiber optic communication systems. This is


even more so in dense WDM systems where the spacing between adjacent

wavelengths can be from 0.8nm (100GHz) to as small as 0.2nm (25GHz) vi. Since

typical InGaAsP-based DFB lasers operating around 1.55 µm have a wavelength drift

of approximately 0.1 nm/°C, the temperature must be controlled to less than a degree

of variance to prevent excessive loss in multiplexers/demultiplexers or crosstalk

interference. While Peltier devices have successfully met this requirement, they have

added greatly to the total cost of components since they are not easily integrated with

devices vii.

Another disadvantage to the use of Peltier devices is the large mismatch in thermal

mass between that of the cooler and the device. The smallest Peltier devices are a

couple of millimeters squared, whereas a typical optoelectronic device is an order of

magnitude smaller. Much work is currently underway in thin film thermoelectric

refrigeration for other applications, however the same problems of integration with

optoelectronics still exist. The InGaAsP/InP family of materials has poor

thermoelectric properties due to the inherently small Seebeck coefficient viii. However,

the use of thermionic emission in heterostructures was recently proposed and has been

demonstrated in the InGaAsP system to increase the cooling power ix,x

F. Peltier Device in IP Over WDM

The usage of the Peltier Device in this project has the unique purpose of tuning or

detuning the fiber Bragg grating, since the grating period of a FBG depends on

temperature, to reflect or not the wavelength channels according on the switching

requirements. The FBG was placed over the thermoelectric Peltier device, so that its

temperature could be easily set and controlled. The electronic circuit designed to

control the Peltier Device is briefly described in Appendix B.

i http://www.peltier-info.com/ ii http://www.peltierelement.com/english/peltierelement/index.peltierelement.html iii http://www.overclockers.com/topiclist/index21.asp#PELTIERS iv http://www.toomeycomputers.com/Peltiers.htm v http://www.melcor.com/handbook.htm


vi Y. Yamada, S.I. Nakagawa, K. Takashina, T. Kawazawa, H. Taga, K. Goto, “25GHz spacing ultra-

dense WDM transmission experiment of 1 Tbit/s (100WDM x 10Gbit/s) over 7300km using non pre-

chirped RZ format”, IEEE Elec. Lett., 35, pp. 2212, 1999. vii L. Rushing, A. Shakouri, P. Abraham, J.E. Bowers, “Micro theroelectric coolers for integrated

applications”, Proceedings of 16th International Conference on Thermoelectrics, Dresden, Germany,

August 1997. viii A. Shakouri, C. LaBounty, “Material Optimization for Heterostructure Integrated Thermionic

Coolers”, Proceedings of 18th International Conference on Thermoelectrics, Baltimore, MD, USA

August (1999). ix A. Shakouri, E.Y. Lee, D.L. Smith, V. Narayanamurti, J.E. Bowers, “Thermoelectric effects in

submicron heterostructure barriers”, Microscale Thermophysical Engineering 2, 37 (1998). x C. LaBounty, A. Shakouri, P. Abraham, J.E. Bowers, “Integrated cooling for optoelectronic devices,

Proceedings of SPIE Photonics West Conference”, San Jose, CA, USA, Jan. 2000

Appendix B – Peltier Electronic Controller 1

Appendix B – Peltier Electronic Controller

A. Introduction

To control Bragg wavelengths by temperature alteration it was necessary make use of

the Peltier device described in the previous appendix. To control this device a precise

electronic control circuit is required. Basically the Electronics’ Peltier controller is

composed of three different stages: Digital Controller, Analogical Controller and a

Power Circuit, to which the Peltier Device is connected. The block diagram of the

Peltier Controller is shown in Figure 1.

Figure 1 - Peltier Controller Stages

B. Digital Controller

As described bellow the first of the three stages of the Peltier Controller is the Digital

Controller. In reality it wasn’t necessary to design this part of the circuit, since a

circuit with similar functionalities had already been designed for a previous project in

Lasers Modulators at UOSEi. That digital controller has the capability of generating

digital voltage steps (Vdac), which will be used to control the Peltier device current

(IP). The voltage Vdac is controlled with a microcontroller of the 80C51 family and the

selection of the multiple voltage steps could be done with a personal computer (PC)

(since that stage is connected to the PC through the serial port) or with a simple

keyboard connected directly to the digital controller.

C. Analogical controller ii

The purpose of the second stage, the Analogical Controller, is to convert the digital

signal VDAC into analogical voltage signals (VA1(1), VB1(1), VA2(1), VB2(1)) that will be

sent to the Power Circuit. The electronics behind the Analogical Controller is quite

simple and based in several operational amplifiers µA741, and some passive

PeltierDevice

V dac V A1

V B1 VA2

V B2

1

2

Analogical Controller Power Circuit Digital Controller


components such as resistors, potentiometers, diodes, and capacitors. The schematic

representation of this circuit is shown in Figure 2.

Figure 2 - Peltier Device Analogical Controller

The controller’s configuration is described next through a series of schematics.

a. Input stage

In Figure 3 the configuration of the controller Input stage is shown.

Figure 3 - Input stage

As seen in the previous figure the input stage is a non-inverting configuration with a

voltage reference at the non-inverting input terminal. Equation 1 describes this circuit

configuration.

12

2

2

22 1 REF

up

downdac

up

downoutU V

RVRVV

RVRVV •

−•

+=

Equation 1


The Vdac voltage variation is from 0V to 5V. VREF1 has a constant value as will be

calculated next. The VU2out signal varies between +15V and –15V and his value is 0V

when Vdac=2.5V.

b. VREF1 Generator

In the Figure 4 is shown the configuration of the VREF1 Generator.

Figure 4 - VREF1 Generator

The operational amplifier U1 works as buffer. The Equation 2 that describes a simple

voltage divider gives his output voltage.

downup

upREF RVRV

RVV

11

11 30

+•=

Equation 2

In this project the voltage VREF1 used is equal to 3V. It’s interesting to detach that this

value of VREF1 imposes VU2out=0V when Vdac is equal to 2.5V.

c. VREF2(1) and VREF2(2) Generators

In Figure 5 the configurations of the VREF2(1) and VREF2(2) Generators are shown.

Figure 5 - VREF2(1) and VREF2(2) Generators


The op. amps. U3 e U4 work as comparators with reference voltages in the inverting

terminal. The Equation 3 gives these reference voltages.

updown

downREF

updown

downREF RVRV

RVVRVRV

RVV44

4)2(2

33

3)1(2 15;15

+•=

+•=

Equation 3

Both reference voltages VREF2(1) and VREF2(2) are equal to 0.1V. The purpose of these

low voltage levels is to provide a faster response to this configuration when the VU2out

starts increasing (case of U3) or decreasing (case of U5).

d. Inverting configuration with unitary gain

In Figure 6 the configuration of the inverting configuration for a amplifier with a

unity gain is shown.

Figure 6 - Inverting configuration with unitary gain

In the Equation 4 R5 must equal R4 allowing the inverting amplifier to have unity

gain.

outUB VRRV 2

4

52 •−=

Equation 4

e. The Diodes functions

D1 allows positive voltage excursions to reach operational amplifier U3 and the

output B1, while D2 allows negative voltage signal excursions to the operational

amplifiers U4.

D3 and D4 are respectively connected to the op. amps. U3 e U5 with the purpose of

providing only 2 steps, high or null to the output signals of A1 and A2.


f. The resistors R3 and R6 functions

The insertion of these resistors in the digital controller circuit has the unique purpose

of avoiding voltage floating points which could impair the circuit’s performance.

g. The Analogical controller functional description

The functional description of this circuit is done in a very simple way appealing to the

following Table 1.

Vdac (V) VU2out (V) VA1(1) (V) VB1(1) (V) VA2(1) (V) VB2(1) (V) 0 -15 0 0 0 13.6 ⇓ -15 ⇒ 0 0 0 0 ⇒14.3 13.6

2.5 0 0 0 0 0 ⇑ 0 ⇒ 15 0 ⇒14.3 13.6 0 0 5 15 0 13.6 0 0

Table 1

D. Power Circuit ii

Since the analogical control doesn’t supply large currents and the Peltier device drives

lots of power, i.e., it needs large currents to operate, it was necessary to design a

Power circuit that works as an interface between the control and the Peltier device,

producing enough current (IP) to drive the Peltier device. This Power circuit is shown

in Figure 7.

Figure 7- Power circuit


The previous circuit is based in two sets of transistors, one composed by four bipolar

transistors (Q2, Q4, Q6 and Q8) and another composed by four bipolar power

transistors (Q1, Q3, Q5, and Q7). The pairs Q1 and Q2, Q3 and Q4, Q5 and Q6, Q7

and Q8, are Darlington pairs. The use of the Darlington configuration is very

important in this case because the output current of the Analogical controller is very

small. Using Darlington configuration reduces the base current requirements for a

power stage and provides a very high current gain. This happens because the current

gains of both transistors are multiplied by each other.

The entry’s 1 and 2 are respectively connected to the outputs 1 and 2 of the analogical

controller. When Vdac spans between 2.5V and 5V the Peltier device acts as a heat

source. The Darlington pair composed by Q1 and Q2 acts like an interrupter while the

other one composed by Q7 and Q8 provides the increase or decrease of the Peltier

current by changing the Darlington base voltage. When the Vdac spans between 0V

and 2.5V the Peltier device act as a cooling source because the other entries (3 and 4)

provide voltage levels that make this possible.

The Peltier device is connected to the output ports (1 and 2 – interface P6) of the

Power circuit stage.

E. Peltier device controller pictures

Figure 8 – Analogical controller Stage


Figure 9 - Board with four analogical controller stages

Figure 10 - Power Circuit

i V. Barbosa, “Transmissor Óptico Analógico”, INESC Porto - UOSE, 2001 ii A. S. Sedra, K. C. Smith, “Microelectronic Circuits”, Oxford University Press, 1998

Appendix C – OADM paper submitted in Física 2002 1

Appendix C – OADM paper submitted in Física 2002

AVALIAÇÃO DE DUAS ARQUITECTURAS DE OADM BASEADAS EM CIRCULADORES ÓPTICOS E REDES DE BRAGG EM FIBRA ÓPTICA

I. Terrosoa#, J. P. Carvalhoa#, O. Frazãoa, M. Ricardoa,b , H. M. Salgadoa,c a INESC Porto - UOSE , Rua do Campo Alegre 687, 4169-007 Porto – Portugal

b INESC Porto - UTM , Praça da República 93, 4050-497 Porto, Portugal c FEUP - DEEC, Rua Dr. Roberto Frias, 4200-465 Porto - Portugal

A tecnologia de multiplexagem em comprimento de onda DWDM – Dense Wavelength Division Multiplexing tem evoluído rapidamente pois parece ser a única capaz de satisfazer a crescente necessidade de largura de banda em redes de comunicações ópticas. O aumento do tráfego nas redes de comunicação por fibra óptica, resultante da procura de serviços de Internet e múltimedia, impõe o desenvolvimento de redes baseadas na tecnologia DWDM, para aumento da capacidade, que sejam simultaneamente capazes de efectuar o encaminhamento e comutação de comprimentos de onda, por oposição a sistemas de transmissão ponto-a-ponto. A implementação dessas técnicas de remoção de canais requerem o desenvolvimento de nós ópticos de remoção e adição de canais – OADM Optical Add-Drop Multiplexers. Estes dispositivos são utilizados para remover e adicionar selectivamente um ou vários canais numa rede óptica DWDM, aumentando desta forma a sua flexibilidade. Neste artigo apresenta-se um OADM baseado em circuladores ópticos e em redes de Bragg em fibra óptica (FBG – Fiber Bragg Gratings) [1]. A diafonia (Crosstalk) é um dos problemas de OADM’s deste tipo. Neste trabalho apresenta-se uma avaliação do desempenho de duas arquitecturas OADM (ver figura 1) em termos de diafonia. O OADM apresentado na figura 1, utiliza uma FBG que está estabilizada através de um elemento de Peltier a uma temperatura constante. As FBGs reflectem um determinado comprimento de onda de acordo com a condição em que se encontra a rede de difracção de Bragg segundo a seguinte equação, λB = 2 neff Λ em que neff é o índice de refracção efectivo do modo de propagação guiado, e Λ é o período de modulação do índice de refracção na fibra.

(λ1,λ2,λ3)

Entrada Saída

λ2

Remoção Adição

Sinal A

Sinal C

Sinal B (λ1,λ3)

(λ2)

a)

(λ1,λ2,λ3)

Entrada Saída

λ1

(λ1,λ3)

λ3

Remoção Adição

Sinal A Sinal B λ2 λ2

Isolador Óptico

Sinal C (λ2) b)

Figura 1) Representação das duas arquitecturas em estudo

# Telf: 22.608.2601 Fax: 22.608.2299 e-mail: [email protected] , [email protected]

Appendix C – OADM paper submitted in Física 2002 2

Na Figura 1 temos representadas as duas arquitecturas em estudo. A Figura 1 a) representa o OADM convencional [2] onde existe diafonia a prejudicar o desempenho, e na Figura 1 b) já temos uma arquitectura capaz de reduzir significativamente a diafonia.

Na Figura 2 estão representados os espectros (reflexão/transmissão) do FBG usado no OADM. O filtro óptico é uma rede de Bragg em fibra óptica centrada em 1550,9 nm, com uma reflectividade próxima de 100% e FWHM igual a 0,2 nm. Foram introduzidos 3 canais WDM espaçados a 100 GHz, nas duas arquitecturas. Na Figura 3 estão representados os três canais introduzidos nos OADM’s na porta de entrada (Sinal a tracejado – A). O sinal B é o sinal à saída do OADM após a remoção do canal 2 e está livre de diafonia pois o filtro óptico apresenta caracte-rísticas excelentes de filtragem. O sinal C, que é obtido na porta de remoção do OADM (referente à primeira arquitectura) já apresenta picos de diafonia referentes aos canais 1 e 3. Estes picos são eliminados graças à arquitectura do OADM (segunda arquitectura) que apresenta FBGs na saída de remoção para eliminar os picos de diafonia presentes na primeira arquitectura, como se pode observar na figura 4, em que se compara o sinal C, obtido com a primeira arqui-tectura e o sinal D obtido com a segunda arquitectura. Foram apresentadas duas arquitecturas de OADM’s baseadas em dois circuladores ópticos e FBG’s. A segunda arquitectura mostrou um melhor desempenho em relação à primeira em termos de diafonia. Referências [1] C.R. Giles, Journal of Lightwave Technology, Vol. 15, Nº. 8, August 1997 [2] P.S. André e J.L. Pinto, I. Abe, H.J. Kalinowski, O. Frazão, F.M. Araújo, Journal of Microwaves and

Optoelectronics, Vol. 2, N.º 3, July 2001.

1548 1549 1550 1551 1552 1553 15540,0

0,2

0,4

0,6

0,8

1,0

Tran

smis

são

(dB)

Ref

lect

ivid

ade

λ (nm)

-16

-14

-12

-10

-8

-6

-4

-2

0

Figura 2 – Sinal do filtro óptico baseado em FBG em transmissão e em reflexão.

1548 1549 1550 1551 1552 1553 1554-25

-20

-15

-10

-5

0

A B C

Tran

smis

são

(dB

)

λ (nm)

Figura 3 - A – Sinal de entrada com 3 canais WDM, B – Sinal após a remoção do canal 2 na 1ª arquitectura de OADM; C – Sinal após remoção dos canais 1 e 3 na 1ª arquitectura.

1548 1549 1550 1551 1552 1553 1554-25

-20

-15

-10

-5

0

C D

Tran

smis

são

(dB)

λ (nm)

Figura 4 - D – Sinal do canal 2 removido passando por 2 filtros ópticos na 2ª arquitectura.

Appendix D – OXC architecture paper submitted in Física 2002 1

λ1, λ2 , λ3

Entrada 1

λM

Entrada 2

Saída 1

Saída 2

Fig.1 – Arquitectura do comutador

óptico (OXC)

Appendix D – OXC architecture paper submitted in Física 2002

COMUTADOR ÓPTICO (OXC) BASEADO EM CIRCULADORES ÓPTICOS E NUMA REDE DE BRAGG EM FIBRA ÓPTICA

J. P. Carvalhoa#, I. Terrosoa#, O. Frazãoa, V. Barbosaa, M. Ricardob,c, H. M. Salgadoa,c a INESC Porto - UOSE , Rua do Campo Alegre 687, 4169-007 Porto – Portugal

b INESC Porto - UTM , Praça da República 93, 4050-497 Porto, Portugal c FEUP - DEEC, Rua Dr. Roberto Frias, 4200-465 Porto - Portugal

O incremento nas taxas de transmissão de dados devido ao uso cada vez mais acentuado da Internet e de outras aplicações multimédia tornaram as redes de telecomunicações totalmente ópticas, baseadas na tecnologia de multiplexagem densa de comprimento de onda (DWDM-Dense Wavelenght Division Multiplexing), no candidato maioritário à constituição do backbone que suportará o tráfego global de dados num futuro próximo. O DWDM e o consequente recurso às técnicas de encaminhamento por comprimento de onda, tornam os comutadores ópticos (OXC-Optical Cross Connect) com selecção de comprimento de onda, dispositivos chave neste tipo de redes, dado que permitem aos pontos terminais de rede a possibilidade de comunicarem de forma transparente, flexível e reconfigurável. As propostas para arquitecturas de OXC’s têm sido várias, nomeadamente: as tecnologias baseadas em micro-espelhos reguláveis sobre bases de silício (MEMS-Micro Electro Mechanical System), guias de onda baseados em bolhas, e ainda soluções baseadas em redes de Bragg em fibra óptica (FBG-Fiber Bragg Gratings) [1]. Um grave problema destes sistemas de encaminhamento de comprimento de onda é a diafonia (crosstalk), que causa uma degradação acentuada na performance do sistema. Esta pode ser de dois tipos: heterodina (heterodyne crosstalk) – resultante da interferência de pequenas fracções de potencia fora da banda do canal, ou homodina (homodyne crosstalk) – resultante de interferências dentro da banda do canal. Neste artigo apresenta-se uma arquitectura de um OXC de 2 × 2 portas baseada numa FBG e circuladores ópticos [3]. Apresenta-se uma avaliação do comportamento deste dispositivo (OXC) face a variações controladas da FBG para o reencaminhamento dos comprimentos de onda de entrada. A configuração utilizada é representada na Figura 1, e consiste em dois circuladores (JDS Uniphase) e uma FBG sintonizável controlada por um elemento de Peltier. O comprimento de onda de Bragg λB ocorre quando a constante de propagação do modo guiado no núcleo se encontra em ressonância com a modulação espacial do índice estabelecendo a condição de Bragg: λB = 2 neff Λ, em que neff é o índice de refracção efectivo, Λ é o período de modulação do índice de refracção na fibra [2]. Foi fabricado uma FBG do tipo uniforme com

comprimento de onda 1550.7 nm, FWHM de 0.2 nm e uma reflectividade de próxima de 100%. O desempenho do OXC apresentado foi testado com três canais WDM na entrada 1, de comprimentos de onda λ1=1549.9 nm, λ2=1550.7 nm e λ3=1551.5 nm, espaçados de 0.8 nm, o que corresponde a um intervalo em frequência de 100 GHz. Quando o filtro óptico FBG se encontra estabilizado à temperatura ambiente, a FBG está sintonizada em λM=λ2, encaminhando dessa forma o canal 2 para a saída 1 do OXC, enquanto que os outros

# Telf: 22.608.2601 Fax: 22.608.2299 e-mail: [email protected] , [email protected]

Appendix D – OXC architecture paper submitted in Física 2002 2

1548 1549 1550 1551 1552 1553 1554-35

-30

-25

-20

-15

-10

-5

0

Saída 1 Saída 2

Pote

ncia

Ópt

ica

(dB)

Comprimento de Onda (nm)1548 1549 1550 1551 1552 1553 1554

-35

-30

-25

-20

-15

-10

-5

0

Saída 1 Saída 2

Pote

ncia

Ópt

ica

(dB)

Comprimento de Onda (nm) a) b)

Fig.2 – Potência Espectral no comutador óptico: a) Canal 2 sintonizado e encaminhado para a porta 1; b) Dessintonia da rede de Bragg e encaminhamento dos três canais para a saída 2.

dois canais são direccionados para a saída 2. Na Figura 2 a) verificamos que a rejeição feita pelo FBG dos canais λ1 e λ3 é de -16.23 dB (saída 2). As pequenas componentes espectrais centradas em λ1 e λ3, devem-se sobretudo a reflexões residuais no FBG e nos circuladores e originam diafonia heterodina. Numa segunda fase a rede de Bragg é dessintonizada, por variação de temperatura e o comprimento de onda de Bragg alterado para um intermédio entre λ2 e λ3 (λM=1551.1 nm). Na Figura 2 b) podemos ver o traçado da potência dos sinais em ambas as saídas deste dispositivo, sendo que os três canais de entrada foram encaminhados para a saída 2. Na saída 1 é verificado um nível de sinal com menos 10.53 dB que na saída 2, resultante da reflexão do FBG que provocará diafonia homodina. Note-se que a redução acentuada dos níveis de diafonia referidos anteriormente pode ser

conseguida recorrendo ao uso de FBG’s apodizados. É importante notar que existem perdas associadas ao trânsito de potência duma porta dum circulador para a porta adjacente.

Neste caso essas perdas são de 1.2 dB e dado que cada canal tem de sofrer obrigatoriamente duas perdas deste tipo, assim se explicam os valores máximos de potência registados nas saídas que rondam os -2.4 dB. Este dispositivo apresenta como limitação a impossibilidade de introdução em portas adjacentes do mesmo canal dado que esta situação provocaria níveis de diafonia elevadíssimos no caso da FBG estar sintonizada e colisões do sinal óptico no caso da dessintonia. A configuração estudada é ainda totalmente escalável sendo possível através de blocos básicos do OXC de 2 × 2 portas elaborar comutadores de N × N portas em que cada canal pode ser encaminhado para qualquer uma das portas de saída desde que seja respeitada a limitação acima referida. Embora nesta configuração apenas seja utilizado um filtro óptico FBG, existe ainda a possibilidade de se usarem M filtros ópticos que no limite seriam tantos quanto o número de diferentes comprimentos de onda introduzido nas portas deste OXC. O comutador óptico aqui descrito e estudado apresentou um desempenho razoável. Actualmente estão a ser estudados e implementadas novas configurações de OXC’s com reduzida diafonia baseadas nesta mesma arquitectura. Referências [1] Y. W. Song, Z. Pan, D. Starodubov, V. Grubsky, E.Salik, S. A. Havstad, Y. Xie, A. E.

Willner, J. Feinberg, “All-Fiber WDM Optical Crossconnect Using Ultrastrong Widly Tunable FBGs", IEEE Photonics Technologhy Letters, VOL. 13, NO. 10, Outober 2001.

[2] Andreas Othonos, Kiriacos Kalli, "Fibre Bragg Gratings – Fundamentals and Applications in Telecommunications and Sensing”,Artech House, London, 1999.

[3] Xiangnong Wu, Xau Lu, Z. Ghassemlooy, Yixin Wang, "Evaluation of Intraband Crosstalk in an FBG-OC-Based Optical Cross Connect”, IEEE Photonics Technologhy Letters, VOL. 14, NO. 2, February 2002.

PART II

INESC - UTM Porto, July 2002



IP over WDM


Part II

Supervisors Students Henrique Salgado, PhD Manuel Ricardo, PhD

Bruno Leite Fernando Pinto

Table of contents

i

Table of contents

Table of figures............................................................................................................................... ii Chapter 1 – Introduction..................................................................................................................1

Abstract ........................................................................................................................................................................1 Document Structure .....................................................................................................................................................1 Background ..................................................................................................................................................................2

Telecommunications Networks – the optical era ...................................................................................................2 Data traffic emergence – the Internet revolution....................................................................................................2 Moving to a distributed standardized control plane ...............................................................................................3

The Objectives..............................................................................................................................................................4 Our work.......................................................................................................................................................................5

Chapter 2 - Optical technology networks ........................................................................................6 Background ..................................................................................................................................................................6 Network architectural considerations ..........................................................................................................................6

The usage of WDM technology..............................................................................................................................6 Mesh topology vs ring topology .............................................................................................................................6 A new control plane is required. .............................................................................................................................6 Network integration models – IP over WDM protocol stack. ...............................................................................7

Layer 2 technologies in use..........................................................................................................................................8 Synchronous Digital Hierarchy...............................................................................................................................9 Optical Transport Network G.709 ........................................................................................................................10 Optical Internetworking Forum - UNI 1.0............................................................................................................11 GigaBit Ethernet....................................................................................................................................................12

Migrating from management based solutions to automatic routing and provisioning.............................................15 Moving to IP protocol arena ......................................................................................................................................16 Using MPLS framework ............................................................................................................................................16 Evolving to GMPLS...................................................................................................................................................17

The need for different control plane models ........................................................................................................18 Conclusions ................................................................................................................................................................18

Chapter 3 - GMPLS.......................................................................................................................19 Architecture................................................................................................................................................................19

Switching domains ................................................................................................................................................19 1. Fiber-Switch Capable (FSC)...................................................................................................................................................................19 2. Wavelenght Switch Capable (λSC) ........................................................................................................................................................20 3. Waveband Switch Capable (WSC).........................................................................................................................................................20 4. Time Division Multiplexing Capable (TDMC)......................................................................................................................................20 5. Packet Switch Capable (PSC).................................................................................................................................................................20

OXC control plane functions ................................................................................................................................20 Control plane..............................................................................................................................................................21

1. Link management....................................................................................................................................................................................21 2. Intra domain routing protocols ..............................................................................................................................................................22 3. Inter domain routing protocols ..............................................................................................................................................................22 4. Signaling protocols .................................................................................................................................................................................22

Inband/Outband control channel................................................................................................................................23 Traffic Engineering ....................................................................................................................................................24 Conclusions ................................................................................................................................................................24

Chapter 4 - Our approach ..............................................................................................................25 Core architecture. .......................................................................................................................................................25 How different components interact ...........................................................................................................................26

Forward Information Base (FIB) ..........................................................................................................................26 Traffic Engineering Topology DB........................................................................................................................26 LMP.......................................................................................................................................................................26 OXC Controller .....................................................................................................................................................26 RSVP-TE and OSPF-TE.......................................................................................................................................27 Traffic Engineering control...................................................................................................................................27

Signaling (RSVP and CR-LDP) ...............................................................................................................................27 How to request an LSP..........................................................................................................................................28 Generalized Label Request ...................................................................................................................................29 Label Suggestion by the Upstream .......................................................................................................................30 Label Restriction by the Upstream .......................................................................................................................30

Table of figures

Routing (OSPF and BGP)..........................................................................................................................................31 The roots of GMPLS routing ................................................................................................................................31

Some challenges to consider …..................................................................................................................................................................32 …and some possible solutions....................................................................................................................................................................32

Link management (LMP)...........................................................................................................................................34 Managing links ......................................................................................................................................................35 LMP for DWDM Multiplexer...............................................................................................................................35

Traffic Engineering ....................................................................................................................................................36 Conclusions ................................................................................................................................................................37

Chapter 5 – The proposed architecture..........................................................................................38 OXC Controler ...........................................................................................................................................................38 RSVP ..........................................................................................................................................................................43

OXC controller communication functions. ..........................................................................................................43 RSVP signaling messages.....................................................................................................................................44

1. Generalized label request object .............................................................................................................................................................44 2. Generalized label object..........................................................................................................................................................................45 3. Suggested label object.............................................................................................................................................................................45 4. Label Set object.......................................................................................................................................................................................46

Communication interface...........................................................................................................................................47 Using a PCI board. ................................................................................................................................................47

Conclusions ................................................................................................................................................................48 Chapter 6 – Conclusions................................................................................................................49 References .....................................................................................................................................51

Papers .........................................................................................................................................................................51

Table of figures Figure 1 – Optical Network Model..................................................................................................5 Figure 3 - Protocol Stack evolution for IP-over-WDM solutions. ..................................................7 Figure 4 – 10 Gigabit Ethernet Protocol Stack..............................................................................13 Figure 5 - Optical Media for 10Gigabit Ethernet . ........................................................................14 Figure 6 - 10 Gigabit Ethernet in metropolitans network.............................................................14 Figure 7 – Switching domains. ......................................................................................................19 Figure 8 - Control plane architecture.............................................................................................25 Figure 9 - LSP hierarchical structure traditional in GMPLS networks. ........................................34 Figure 10 - Control structures........................................................................................................39 Figure 11 - OXC Controller Data Structures, full picture. ............................................................40 Figure 12 - FIB data structures. .....................................................................................................40 Figure 13 - Forwarding Information Base Hash Table..................................................................41 Figure 14 - OXC controller and FIB joint operation. ....................................................................41 Figure 15 - Overall program diagram............................................................................................42 Figure 16 - Generalized Label Request object format...................................................................44 Figure 17 - Generalized Label Object format................................................................................45 Figure 18 - Label Set object format ..............................................................................................46

Chapter 1 – Introduction

IP over WDM II - 1


Abstract

The main objective of this work consists in defining the architecture of an OXC control

plane. This architecture must cope with two main issues.

The first is the diversity of deployed equipments and technologies. That requires a multi-

protocol solution support and some backward compatibility. The second is the movement of IP

related technologies from the edge to the core of the network.

The proposed approach combines the recent GMPLS advances with OXC technology to

provide a framework for real-time routing and provisioning of optical channels and allow the use

of uniform semantics for hybrid network management and operations control.

The development of GMPLS requires modifications to signalling and routing protocols.

The protocols being considered are originated from the IP arena, thus the movement of IP

technologies to the network core is a fact.

Document Structure

This document is divided in chapters, each one covering a given subject.

Chapter 1 - Introduction covers the project presentation and its key elements.

Chapter 2 - Optical Technology Networks and Chapter 3 - GMPLS present actual

technology state and define the main guidelines of IP over WDM control plane within the

definitions of GMPLS.

Chapter 4 - Our approach presents a detailed description of the OXC control plane

components and which are the key elements in this technology. This chapter starts the control

plane definition and references the protocols to be used.

Chapter 5 – The proposed architecture reinforces previous chapter subjects and covers the

work done and the guidelines to follow to complete the implementation.

Chapter 6 – Conclusions, the final chapter, presents the main conclusions and a small

description of the complete OXC usage and final implementation.


2 - IP over WDM II

Background

Telecommunications Networks – the optical era

Nowadays optical transport networks assume the central role in the telecommunications

networks stage. Synchronous Digital Hierarchy established itself as the absolute must have for

any telecommunications corporation, worthy of bearing that name. However, despite its

widespread use and proven capabilities to explore the benefits of the optical medium, SDH

networks can no longer take their future for granted as data traffic emerges to become the

dominant. The design of SDH networks in the 1980s did not foresee the emergence of data

networks in the late 1990s, thus SDH was developed as an high performance voice traffic carrier,

rigidly structured to carry voice signals and unable to take advantage of statistical multiplexing -

the key word in data networks.

Though, as many said, ‘rumours of the impending death of SDH might have been

exaggerated’, and SDH will keep on being the dominant technology for years to come. The

investments made in building an SDH infrastructure, the tremendous experience and known-how

accumulated in long years of operation and management are a too precious asset to throw away

overnight. Wavelength Division Multiplexing was a significant breakthrough in the physical

layer, providing SDH with an n-fold increase in available bandwidth in a single fibre, where n is

expected to grow from two to four channels today to hundreds in the near future. Though

providing SDH with an absolute surplus value, WDM being a new multiplexing technology

brought an extra burden in an already over burdened network management department. As

carriers strive to keep up with Internet providers ever changing demands and pressure to drop

circuit prices, while having to deal with incompatible network management systems, manually

hop-by-hop configuration of WDM equipment. Whilst keeping up with strong competition from

new comers, in the context of the telecommunication market liberalization.

Data traffic emergence – the Internet revolution

The exponential growth of the Internet in the late 1990s brought great challenges to

backbone data networking technologies. Data networking technologies were aimed always at

small geographic areas, because data networks were usually small and most of the traffic was

internal to the network. Longer links were easily provided in early days through modems using

the public voice network, as demands grew larger using protocol overlay solutions and rented

links (E1 lines, ATM VCs or even STM links). Overlay networks where no more than a short-


IP over WDM II - 3

lived approach to scale IP networking technology. Stability problems (difficult convergence of

routing protocols in large meshed networks with many adjacencies, specially in failure events),

ineffective use of layer two capabilities, excessive overhead due to large protocol stacks,

expensive framing conversion and the impossibility to do traffic engineering, soon forced the

development of new solutions.

Moving to a distributed standardized control plane

Multi-Protocol Label Switching came as a radical revolution in networking concepts.

Applying the maxim - route once, switch many - and adopting the virtual circuit concept of ATM

(now Label Switched Path), MPLS is taking over Internet Service Provider backbones by

allowing easy and effective integration with layer two technologies (specially ATM but also

emerging Gigabit Ethernet) and proving to be an effective tool for traffic engineering. MPLS has

leveraged IP networking from a collection of small islands interconnected by the public

telephone network to a metropolitan and even national integrated network.

Notwithstanding, this a far from optimal solution, transporting packet traffic, multiplexed

statistically, even using MPLS in an underlying circuit switching technology has been a cause of

major headaches for network management departments all over the world. Today, more than

extra bandwidth, provided by WDM for years to come, fast and flexible provisioning, low cost

and rapid network deployment are the commons goals to the industry, from equipment

manufacturers, standards committees to telecommunications corporations.

International Telecommunications Union - Telecom is devoting considerable effort in the

development of a large number of standards considering distributed network management. Its

main initiatives are the G.ASTN (Automatic Switched Telecommunications Networks) and the

G.ASON (Automatic Switched Optical Networks), both of them are due to produce standards by

early 2002 to ratified by mid that year. The Optical Internetworking Forum has also developed

valuable work in this field, by standardizing the User to Network Interface for signalling, already

approved and tested – UNI-1.0. The Internet Engineering Task Force is producing considerable

work in developing a full control network specification starting from the standardizations

introduced by ITU-T and specially OIF but evolving to an IP like architecture - GMPLS.

As data traffic becomes even more dominant, data networks concepts and protocols will

certainly shape transport networks. This can be already seen in a progressive change in voice

applications towards the usage of IP and the new generation of mobile phone networks - UMTS.

However, IP type networks have more to offer than a flexible network protocol. They were


4 - IP over WDM II

conceived to allow different multi-vendor equipment to interoperate, resilient to withstand

failure and unreliable media, flexible and auto-managed to simplify its deployment and

survivability. IETF best impersonates the Internet attitude, where an industry community

develops open advisories, which are indeed the strongest standards in the networking world.

Common Control and Management Plane and IP over Optical workgroups have produced a large

number of drafts, proposing the adoption of routing and signalling protocols from the IP world to

provide transport networks with a distributed control plane architecture. This new specifications

were produced in conjunction with ITU and mainly OIF standardization efforts. This will

provide optical networks with an end to end signalling and automatic routing infrastructure.

The Objectives

Some intermediate steps are needed in order to test and define the final architecture.

The goals of this project are:

- Build an Optical Add/Drop multiplexer.

- Build an Optical Cross Connect.

- Define Optical IP Router architecture.

- Specify and develop the Optical IP Router control plane.

The project is divided between two groups each one with a different part of the project.

One part of the project deals with physical optical components, fabrics and the electronic

control. The objective is to design an OXC to be controlled by the control plane defined in the

other part of the project.

Our part of the project discusses and defines, based in the latest developments and

standards in this area, the structure of the protocol stack, needed to establish the control plane

that will control the OXC.

This Optical IP Router is only fiber and wavelength switching capable. Due to this

limitation it is equivalent to a common OXC in the data plane, but in control plane it is capable

of routing and signalling in a GMPLS context.

At physical level the breakthroughs of our OCX are the usage of fully passive optical

technology and the capability of wavelength and spatial (fiber) switching simultaneously.


IP over WDM II - 5

Our work

Our work focuses on providing the state of the art WDM OXC, designed by our fellow

colleagues and presented in the first part of this document, with a control plane as specified by

GMPLS drafts for wavelength capable equipment. We strongly believe our OXC can have an

important role to play in large data backbones, which we consider will be governed by GMPLS

routing and signalling protocols. Switching transparently high bandwidth WDM channels, of

more and more IP traffic, each time transported closely to the optical medium.

AON

Client Netw ork

λSR

λSR

λSR

λSR

λSR

Client Netw ork

Client Netw ork OXC

OXC

OXC OXC

OXC

AON

Client Netw orkClient Netw ork

λSRλSR

λSRλSR

λSRλSR

λSRλSR

λSRλSR

Client Netw orkClient Netw ork

Client Netw orkClient Netw ork OXCOXC

OXCOXC

OXCOXC OXCOXC

OXCOXC

Figure 1 – Optical Network Model.

The network model represented in Error! Reference source not found. states our view

of the future for optical networking. Our WDM OXC, now a GMPLS lambda-router fits in the

core of this network, future developments (providing it with packet our TDM switching

capabilities) can take it to the border of the WDM cloud acting as an edge lambda router.

Chapter 2 - Optical technology networks

6 - IP over WDM II


Background

Today, in the Internet era, traffic patterns are unpredictable, what stills remains

predictable is the traffic exponential growth. Operators need a solution that scales well and

allows them to reconfigure the network to cope with rapidly changing traffic patterns. Add to

this the fact that many different types of traffic are entering the network now at optical speeds

(including ATM, gigabit Ethernet, high-speed TDM circuits, and IP). So being, a carrier needs a

way to manage all these protocols and wavelengths through simplified control structures. Today

a carrier may have to enlist three or more different management systems to provision a single IP

connection across the country, and many occasions with manual interventions required along the

way, thus the goal is to have a simplified control structure that selects conduit, fiber, wavelength,

optical switch ports, router ports, and a variety of restoration paths in a single step, according to

traffic engineering constraints.

Network architectural considerations

The usage of WDM technology

It is important to identify the latest changes in optical transport network architecture.

One is the use of Wavelength multiplexing (WDM) in order to explore the available

bandwidth in the fiber, and now at an higher rate with the use of Dense WDM (DWDM).

Mesh topology vs ring topology

Another change is that the early implementations of optical networks were based on ring

topologies. The goal at that time was to have point-to-point connection in the core network. Now

the trend is migrating this technology closer to the end user, so the mesh topology becomes

necessary. Mesh architectures are easy to deploy, allowing a more powerful implementation.

A new control plane is required.

A new optical control plane and a new unifying networking architecture are required that

are equally adapted to manage connections all the way from the heavy-duty fiber in the core to

individual packet flows within a single link.


IP over WDM II - 7

Carriers can continue to survive, for now, without optical signalling or any other

integrated control schema, but it is a sure thing that they could benefit from such a solution.

Early runners in implementing integrated signalling systems will have the advantage in the near

future.

Network integration models – IP over WDM protocol stack.

Is important at this point to look at the protocol stack. Current implementations of optical

/ WDM technology are usually based on SONET/SDH and ATM, which acts as an interface to

higher layers.

The emergence of Internet and its related applications based on Internet Protocol (IP) are

making IP the dominant protocol, to which all communication network technology converge.

The solution for communication networks will become IP over WDM. IP will become the

convergence layer; WDM is and shall be the high-bandwidth carrier.

OpticalLayer

SDHATM

IPIP

Traditional SDH approaches. - IP/ATM/SDH - POS ( IP/PPP/HDLC/SDH)

Traditional SDH based approaches, with WDM layer.

IP over WDM overlay approach

Direct MPλS approach

WDM Adaptation Layer

HDLCIP

GbE

IP

SDHATM

IP

WDM / OTNOptical LayerGigabit Ethernetover SDH Aproach

PPPIP

HDLCPPP IP

GbEGFP

WDM

PPP EthIP

GFPPPP Eth

ATMIP

OpticalLayer

SDHATM

IPIP

Traditional SDH approaches. - IP/ATM/SDH - POS ( IP/PPP/HDLC/SDH)

Traditional SDH based approaches, with WDM layer.

IP over WDM overlay approach

Direct MPλS approach


HDLCIP

GbE

IP

SDHATM

IP

WDM / OTNOptical LayerGigabit Ethernetover SDH Aproach

PPPIP

HDLCPPP IP

GbEGFP

WDM

PPP EthIP

GFPPPP Eth

ATMIP

Figure 2 - Protocol Stack evolution for IP-over-WDM solutions.

As seen in Figure 2, the first step towards faster networks was the introduction of a WDM

adaptation layer to the traditional SDH approach. This new layer would manage WDM channel

setup/takedown and provide some level of protection recovery. This IP/ATM/SDH based

solution configures an overlay model network (CLIP, MPOA), and carries a significant overhead

due to the four framing layers and extra management burden. Using Packet over SONET (POS),

based in IP/PPP/HDLC into SDH framing some overhead is eliminated. The use of Generic

Framing Procedure (GFP) reduces even more this overhead since GFP uses a more efficient

framing technique than HDLC, and less error prone. In this solution IP runs over Ethernet or


8 - IP over WDM II

PPP. Traffic adaptation mechanisms such as ATM, HDLC or GFP shape the traffic to the SDH

layer.

GFP and Ethernet (i.e. Gigabit Ethernet) can also run over fiber directly. So it is possible

to eliminate SDH framing, and reduce even more the overhead. This solution is not yet widely

implemented in existing communications infrastructures based in Sonet/SDH, but with the

standardization of the both protocols it has the potential to become dominant, for its simplicity

and low cost.

Once eliminating SDH framing it is possible to maintain GFP and Ethernet to assure the

framing. Functions such as protection/recovery, if needed can be implemented at WDM optical

layer. But, as Ethernet supports fiber directly it is possible to use direct GMPLS. This is the

lighter and more flexible approach since a direct framing without monitoring is used, expensive

synchronizations. Provisioning and survivability actions can be taken by GMPLS. For carrier-

class reliability an optional framing/monitoring sub-layer can be used.

Layer 2 technologies in use

This project focused mainly in building a physical layer device and designing a control

plane according to GMPLS forthcoming standards. Being a fully transparent switch designed for

the core transport networks it was of little concern in our design which would be the protocols

used to carry the data itself, alias its transparency is one of its key advantages.

However we find from the utmost significance to briefly discuss here layer 2 technologies

current available and how they fit in a GMPLS world. Sometimes, however, some confusion

might arise over which technology refers to which layer, we will consider everything that

concerns physical interfacing as layer 1 and all the rest that fits beneath IP as layer 2, even

though sometimes there will be more than one layer 2 protocol, as in Packet Over SDH where

beneath IP lies PPP, HDLC and SDH, in this case SDH is a layer one protocol in what concerns

to physical media specifications and framing.

We shall not discuss ATM networks in this chapter, not that we find them outdated but

we do think they were extensively covered in MPLS standards and the road towards bringing

closer IP and the optical layer will certainly take ATM as its first victim. By this, with do not

mean ATM is unworthy of our attention, just that it didn't succeed and was clearly defeated by

the combination of IP and Ethernet. ATM's only battle victories were in large networks’

backbones, though its deployment was always tampered by too complex protocol stacks and


IP over WDM II - 9

large overheads, or in an initial stage as a WAN interconnection technology. But now the war

seems definitively lost, as we shall see bellow and only MPLS will keep ATM going in a near

future.

However ATM's effort wasn't in vain and despite defeat it enjoys a sweet vengeance as IP

world sees itself obliged to accept the superiority of most of ATM's concepts, embraced by

MPLS and latterly GMPLS protocols. Route once, switch many left a bitter taste in the mouth of

IP winners.

Synchronous Digital Hierarchy

Long from its introduction in the 80s, SDH has become the widely adopted technology in

carrier telecommunication networks. Its capability to take advantage of the high bandwidths

provided by the optical medium, recently upgraded using DWDM physical layer techniques,

hierarchic structure, tools for medium and equipment monitoring, fast protection and restoration

took it to the status of standard in telecommunications networks all over the world.

SDH is best known for the wideband and ultra fast switching capabilities, however SDH

was conceived not to carry data but voice signals. This has a dramatic impact on its structure,

optimised for multiplexing large amounts of voice traffic coming from lower hierarchies and

switching them in a static fashion - circuit mode. This inability to effectively carry datagram

traffic from the IP world forced the development of several adaptation protocols that ran on top

of SDH structure in an overlay model, most of them inducing large overheads and proving to be

ineffective in using SDH resources.

In an early stage ATM was used as an adaptation layer, due to its cell switching

technology, midway from packet switching and circuit switching. Although ATM was designed

to run over SDH networks and simplified adapting IP nodes to use SDH links, having an ATM

network underneath the IP layer meant a significant traffic overhead, overwhelming effort to

manage three distinct networks (IP, ATM and SDH) and the well known shortcomings of

overlay model networks, namely ineffective usage of network capabilities and instability in

failure events. MPLS solved the overlay problem and raised the billion-dollar question - why use

ATM if SDH can handle the job?

Eager bandwidth network operators were soon taken over SDH high capacity and the

trend to evolve to smaller protocol stacks led to widespread use of SDH in backbone router

interconnection as bandwidth requirements increased dramatically – though still in an overlay

manner. IETF developed a recommendation to meet these requirements, Packet over


10 - IP over WDM II

SDH/SONET (PoS). IP packets are encapsulated in PPP, which in turn is mapped into an HDLC

like frame. This frame is then mapped together with other frames into the payload of a STM-1

frame, if no data is to be transported in one frame the entire payload is filled with HDLC idle

frames. SDH allows multiplexing the STM-1 frame upon climbing higher in the hierarchy,

nowadays reaching 10Gb/s at framing layer with STM-64.

The Point-to-Point Protocol (PPP) provides a standard method for transporting multi-

protocol datagrams (e.g. IP packets) over point-to-point links. Initially, PPP was used over Plain

Old Telephone Services (POTS). However, since the SDH technology is by definition a point-to-

point circuit, PPP/HDLC is well suited for use over these lines. PPP is designed to transport

packets between two peers across a simple link.

PoS and the use of DWDM in the physical layer have allowed carriers to cope with the

ever growing demand for bandwidth, DWDM makes it possible to aggregate in one link up to

1Tb/s and possibly even more. The solid growth in the number of channels in DWDM systems

and higher switching speeds is changing the network bottleneck from link bandwidth to

provisioning times, as will be discussed in the next section.

Optical Transport Network G.709

SONET/SDH matured to become the standard in transport technology, established in

almost every country in the world. It was conceived at a time when traffic voice was utterly

predominant, thus switching circuit concepts was the sensible choice. With the ever-growing

demand for large bandwidths, especially in data communications driven by the large growth

Internet has experienced, it became clear that even using WDM technology and faster TDM

circuits (STM-64) the limits of the transport medium (the optical fibre) and more critically the

switching capability of electronic circuits were reaching its limit.

The latest recommendation in this field is G.709 "Interface for the Optical Transport

Network" builds on the experience gained from SDH to provide a route to the next generation

optical network. The OTN takes the concepts and structure from SDH and broadens them

essentially to achieve larger bandwidths. OTN shares with traditional SDH a layered structure, in

band service performance monitoring, protection and several other management functions.

However, some key elements were added to improve performance and reduce operational costs,

namely more flexible management of optical channels in optical domain and the introduction of

forward error correction techniques to improve error performance in longer and faster optical

spans.


IP over WDM II - 11

OTN also standardises the method for managing optical wavelengths (channels), avoiding

the need for expensive opto-electric-opto (O-E-O) conversion at each node in the link, which has

been demonstrated to be the main limiting factor to larger bandwidths in current networks. This

clearly opens the door to integrate new-coming all optical switching technology in current

networks, with extensive cost saving, simplicity and scalability enhancements. In band

monitoring, protection and fast restoration techniques derived from SDH will account for highly

resilient networks and the several overheads allow transporting GMPLS control plane

information in band.

In a nutshell, OTN paves the way for GMPLS deployment in carrier networks by

standardising the management of optical nodes, allowing more transparent switching (fibre,

waveband or wavelength level) and providing error correction to cope with larger speeds and

spans of optical links without compromising data integrity.

Optical Internetworking Forum - UNI 1.0

Dense Wavelength Division Multiplexing has become quite a standard in wide band

telecommunications systems as the most cost-effective technology for increasing capacity of

optical fibre networks. This new optical network layer promises intelligent transport services to

clients such as backbone IP routers interconnection, initially using SDH, but soon evolving to

other optical interfaces.

The bandwidth requirement growth has increased by several others of magnitude

recently, overwhelming carriers management systems, most of them manually operated, to

control large and intricate optical networks. Currently, optical networks are provisioned through

supplier-specific element management systems. Provisioning end-to-end connections across

multi-vendor equipment has involved the use of incompatible EMSs, manual operations and

even hop by hop reconfiguration, leading to long provisioning times in an era when costumers

demand faster responses from their carriers. The long awaited solution for this time-consuming

matter is to bring MPLS like control intelligence to switching networks. To accomplish this OIF

defined a new standard signalling interface - UNI - between client and optical networks to enable

dynamic provisioning requests.

OIF addressed this critical issue by defining an interface that is compatible with the latest

GMPLS signalling specifications. In addition to signalling, the UNI specifications also addresses

two other subjects fundamental in simplifying management of large networks. The first one is a


12 - IP over WDM II

neighbour discovery mechanism, which will allow adjacent nodes to identify each other.

Neighbour discovery will allow management systems to build inter-connection maps

automatically. The second important aspect of UNI is a service discovery mechanism enabling

clients to determine the services provided by the optical network. Service discovery will allow

clients to automatically discover and take advantage of new services provided by the network as

they are introduced over time. Most of UNI services have been integrated in GMPLS

architecture, specially the mechanism for neighbour and service discovery, which are the basis of

the Link Management Protocol. Though GMPLS will use the services provided by the UNI, a

substantial difference exists between OIF’s point of view and the GMPLS peer to peer network

concept. Whilst GMPLS makes no distinction between client and provider equipment (only the

use of BGP might introduce a concept of border in this otherwise flat architecture), OIF defines

different interfaces, one for inner networks nodes (UNI-N) and one other to external nodes (UNI-

C). The UNI-C is more limited, preventing outer networks equipment to be aware of the inner

network topology in an overlay model, so to speak.

GigaBit Ethernet

Ethernet has come a long way since its introduction as a small and simple network for office

buildings by Xerox in early 1970s. From 10Mb/s over coaxial copper cabling to 1Gb/s over

single mode fibre the protocol evolved to become the dominant technology in local networking

and even menacing to take over WAN networking with its 10Gb/s, STM-64 framing on WDM

support standard. Its low cost technology, easy deployment and ubiquity from desktop

connectivity to campus backbones seized the Internet with over 250 million Ethernet ports

installed.

Moving towards gigabit speeds meant quite a revolution in lower stack layers. In order to

accelerate from speeds 100 Mbps Fast Ethernet up to 1 Gb/s, several changes needed to be made

to the physical interface. It has been decided that Gigabit Ethernet will look identical to Ethernet

from the data link layer upward. The challenges involved in accelerating to 1 Gb/s have been

resolved by merging two technologies together: IEEE 802.3 Ethernet and ANSI X3T11

FiberChannel. Leveraging these two technologies means that the standard can take advantage of

the existing high-speed physical interface technology of FibreChannel while maintaining the

IEEE 802.3 Ethernet frame format (MAC sublayer remains untouched), backward compatibility

for installed media, and use of full or half duplex carrier sense multiple access collision detection

(CSMA/CD). This scenario helps minimize the technology complexity, resulting in a stable


IP over WDM II - 13

technology that can be quickly developed. The Gigabit interface converter (GBIC) allows

network managers to configure each gigabit port on a port-by-port basis for short wave (SX),

long wave (LX), long haul (LH), and copper physical interfaces (CX). LH GBICs extends the

operations using SM fibre to span up to 40km, making it perfect for metro networks

interconnection and perhaps even WAN.

PMD

LAN PCS WAN PCS

MAC

PhysicalLayer

LLCData LinkLayer

64b/66b encodingSTM64 line rate

SDH framing

8b/10b encoding10.3GBaud

Same as in 802.3

10 Gigabit EthernetProtocol StackIEEE 802.3ae

Figure 3 – 10 Gigabit Ethernet Protocol Stack.

10Gigabit Ethernet is much more then a ten fold increase in link bandwidth, being able to

aggregate several Gigabit Ethernet in one link it presents itself as the optimal solution to scale

enterprise and service providers Ethernet backbones. Being compatible with traditional Ethernet

frame format, preserving minimum and maximum frame size (except Jumbo payload from

Gigabit Ethernet), supporting only full duplex operation and compatible Gigabit Media

Independent Interface makes it the best solution to leverage network equipment with Ethernet

interfaces.

10Gigabit Ethernet standardizes two new and different physical interfaces, one targeted

to the traditional LAN market and the new one to the metropolitan links using SDH networks. In

one hand, LAN PHY operating at 10Gb/s up to 300m spans on MM fibre and 2 to 40km spans

on SM fibre for office backbones and interconnection in small distances. WAN PHY, one the

other hand, made compatible with STM-64 framing - ethernet frames are carried in the payload

of a VC4 without the need to do any protocol conversion. Using DWDM physical layer

techniques it is possible to span almost to 1000km. WAN PHY aims high by threatening to take

on ATM and PoS, since Internet end points are ethernet powered, it makes all sense to avoid

expensive framing conversions in WAN links. WAN PHY accomplishes just that by being

compatible with the fastest up to day SDH hierarchy. The two physical interfaces are broken in

two sub-layers, the upper one – Physical Coding Sublayer – allows different coding,


14 - IP over WDM II

multiplexing and serialization. The bottom one is the Physical Media Dependent, which specifies

the physical characteristics of transmitters, optical medium and receivers as described in Figure 4

- Optical Media for 10Gigabit Ethernet . The two physical layers only differ in the PCS and fully

support all the interfaces defined in the PMD sub-layer.

Figure 4 - Optical Media for 10Gigabit Ethernet .

10 Gigabit Ethernet is perhaps the most promising technology at the moment and the one

that can most benefit from GMPLS. The possibility to provision a link dynamically, on client

request granted by a transparent network is the sensible choice when even the framing is

preserved from end to end. Using the WAN PHY it will be easy to integrate existing SDH

networks with new Ethernet links for data networks.

10Gb LAN PHY MM fibre <40km

10Gb LAN PHY MM fibre <300m

10Gb WAN PHY MM fibre <300m

TDMTDM

TDM TDM TDMTDM

WDM MetroNetwork

10Gb WAN PHY SM fibre WDM

ISP PoP

Carrier PoP

Figure 5 - 10 Gigabit Ethernet in metropolitans network.


IP over WDM II - 15

Migrating from management based solutions to automatic routing and

provisioning.

The transition to optical signaling allows the use of automatic routing and provisioning. It

doesn’t have to take place in one profound leap, but can progress in stages, according to a

carrier’s needs.

The first step is already taking place in many carrier networks today. It consists of

improving operational support systems (OSSs) to allow for real point-and-click provisioning of

optical bandwidth.

The second step a carrier may take moves them beyond the management of optical

bandwidth and begins to collapse the boundary between the data services layer of the network

and the optical layer. This step involves implementing a user-to-network-interface, or UNI, that

allows network equipment to “ask” for connectivity across the optical network by signaling for

it. This does require specifications for signaling and provisioning, and it's on the way from the

Optical Internetworking Forum (OIF). In this step, inside the optical "cloud" signaling remains

proprietary, whereas outside the cloud a standardized interface allows client devices attached to

optical systems to talk to each other as though they were neighbours, speeding service creation

and improving resource management.

The third step is the complete standardization of signaling and control planes. With this in

place, carriers can begin to collapse layers of the network, reaching two (a packet layer IP and a

transport layer WDM) or even one, adopting a unified control plane that touches all the

equipment in a carrier network, allowing them all to communicate in real time, dynamically,

asking for bandwidth, connections, grades of restoration, and just about anything they could

want from any layer of the network.

One important question to make is how far to push this technology. Should routers,

dynamically provisioning themselves wavelengths, be making million-dollar decisions on their

own? Can data equipment be expected to manage restoration in the transport layer? Is optical

signaling really the last step to allow a migration to mesh networking in the optical core?


16 - IP over WDM II

The answer to those questions is not the goal of this document, but the idea is to keep in

mind is that optical signalling and routing is only part of a larger movement toward unifying

control planes for multiple network layers

Moving to IP protocol arena

It is important to note here that the emerging Optical Internet will require both signalling

and routing. The signals provide the communication between formerly disparate network

domains; and the routing, in the form of extensions to existing IP routing protocols, is there to

provide enhanced capabilities in managing optical network resources, such as WDM

wavelengths.

A method was required that would ease the burden put on core routers, where traffic

quantity was highest, by relieving these routers of the task of examining each packet header in its

entirety and instead just reading a “label” and passing the packet along according to label

switching rules. Optical signalling and routing are starting their first steps but the fact that it is

drawing upon recent advances in MPLS control plane technology has accelerated its momentum

considerably.

Using MPLS framework

The foundation of MPLS was constraint-based routing, which provides IP devices the

ability to establish and maintain paths through the network that are optimal with respect to a pre-

determined set of metrics and constraints. These constraints can be either resource-related, such

as bandwidth, or administrative related, such as restricting paths to particular links.

In short, MPLS was created as a combination of a forwarding mechanism (label

switching), connection establishment protocols, and defined mappings onto Layer 2

technologies. Thus, MPLS could behave as a next-gen ATM, improving routing in the core of

the network by putting traffic where the bandwidth is, and enabling a range of new services,

including network-based VPNs, circuits over MPLS, and differentiated data services.

The Internet has demonstrated that the non-centralized control plane of MPLS is required

to reduce provisioning and planning costs, while being extremely robust. The essential feature of

MPLS is to apply virtual circuit notions (such as those used in frame relay or ATM) to IP

networks to support quality of service (QOS) and traffic engineering.


IP over WDM II - 17

MPLS bases in Link state routing (LSR) protocols, which are used to obtain network

topology information, and signaling/label distribution protocols, which are used to set up virtual

circuits (or LSPs) across the MPLS network. These labels only have local significance to the

switch, called a label switching router (LSR), and do not require the router to perform any time-

consuming route table lookup.

Fundamentally, this means that in MPLS forwarding information is separate from the

content of the IP header; and through this separation of the forwarding plane and data plane, any

kind of data can be mapped into LSPs, which is already making MPLS as attractive to

edge/aggregation systems as it is to core routers.

Evolving to GMPLS

MPLS evolved the fundamental architecture of routed networks by separating the

forwarding plane (looking up packets) from the control plane (deciding where they go). With this

separation in place, a “best-effort” IP network can now support a variety of protection and

restoration functions, as well as providing some measurable level of QOS, reducing or

eliminating the need for an ATM layer in the network, and improving the IP network’s long-term

scaleability.

With all the work underway developing Multiprotocol Label Switching (MPLS) and its

control plane, it became obvious that the same control plane could be abstracted to the lower

layers of the network, namely Sonet/SDH and the DWDM layer.

If core routers are being simplified with a new switching scheme that relies on a separate

control plane, and optical transmission networks are being simplified with the addition of a

switch, then this new control plane for routers could be applied equally well to both routers and

optical switches.

So the Internet Engineering Task Force (IETF) has been off and running, developing its

Generalized MPLS (GMPLS) for just this purpose, applying MPLS control plane techniques to

optical switches and IP routing algorithms to manage lightpaths in an optical network.

Some IETF groups are working in this area. Groups such CCAMP (Common Control

and Measurement Plane) group, IPO (IP over Optical) group and MPLS (Multiprotocol Label

Switching) group are making the most important contributions to GMPLS standardization.


18 - IP over WDM II

The need for different control plane models

The key distinction between MPLS and GMPLS is that, whereas the control plane for

MPLS was separate from the data plane, in GMPLS it can also be physically separate from the

signal. The GMPLS control plane allows for a wide variety of control plane operation models

and architectures that meet different network operation scenarios. This allows a GMPLS control

plane to manage connectivity and resource management among multiple layers of the network,

from fibers to wavelengths from IP networks do SDH circuits.

Conclusions

The evolution of optical networks towards a simplification and standardization of the

control structures and the trend to carry IP with the least overhead possible in the optical

medium, especially in 10Gigabit Ethernet, opens the door to the introduction of IP protocols in

the management of the AON. As switched networks become more complex, centralized control

systems strive to keep up with the growing pressure for fast network deployment and

reconfiguration. GMPLS with its peer model decentralized control using traditional IP routing,

with traffic engineering extensions, promises to ease network management in a more efficient

and cost-effective way.

Classical MPLS (meaning, derived from IP network elements) and optical MPLS (those

extensions to classical MPLS that control optical network elements) are only subsets of GMPLS.

From now all efforts in MPLS and optical layer management will now be developed under the

rubric of GMPLS.

Chapter 3 - GMPLS

IP over WDM II - 19

Chapter 3 - GMPLS

Architecture

GMPLS assumes a unique control plane, derived from MPLS, that is extended to include

a group of network elements that do not make forwarding decisions based on the information

carried in packet or cell headers, but rather based on time slots, wavelengths, or physical ports.

It is important to understand how these kinds of information channels are related and how

they must be treated.

Switching domains

Those different information channels in the network and the different manners how they

could be switched could be considered as different switching domains. In current drafts of

GMPLS signalling, five types of interface or switching domains are described.

TDMC

FSC/λSC

OXCOXCOXCλSR λSR

PSC

SONETSONET

TDMC

FSC/λSC

OXCOXCOXCOXCOXCOXCλSR λSRλSR

PSC

SONETSONETSONETSONET

Figure 6 – Switching domains.

1. Fiber-Switch Capable (FSC)

These interfaces do not need to recognize bits or frames and do not necessarily have

visibility of individual wavelengths or wavebands. This data forwarding based on the position of

the data in physical space, such as the interfaces on an optical cross-connect that can operate at

the level of single or multiple fibers. These would be found on automated fiber patch panels,

fiber protection switches, or photonic crossconnects that operate at the level of a fiber.

Chapter 3 - GMPLS

20 - IP over WDM II

2. Wavelenght Switch Capable (λSC)

These interfaces do not need to recognize bits or frames. They forward data based on the

wavelength and port on which the data is received, such as an optical cross-connect or

wavelength switch. These are not assumed to be capable of receiving and processing control

plane information on an in-band channel. Examples are interfaces on an all-optical add/drop mux

(OADM) or all-optical crossconnect (OXC).

3. Waveband Switch Capable (WSC)

If adjacent wavelengths are grouped together and the switch has the capability to switch

all of them as a group, this capability is defined as waveband switching. This functionality could

be implemented at the fabrics level, but is easier to implement at control level, as a group of λSC

interfaces.

4. Time Division Multiplexing Capable (TDMC)

These interfaces also recognize bits, though focus on the repeating, synchronous frame

structure of Sonet/SDH. These interfaces forward data on the basis of a time slot within this

structure, and are capable of receiving and processing control plane information sent in-band

with the synchronous frames. Examples are interfaces on Sonet/SDH add/drop multiplexers,

digital cross-connects, and OEO switching systems.

5. Packet Switch Capable (PSC)

Interfaces that make forwarding decisions based on information in the packet or cell

header, such as routers and ATM switches. These interfaces recognize bit, packet, or cell

boundaries and can make forwarding decisions based on the content of the appropriate MPLS

header. Notice, these are also capable of receiving and processing routing and signalling

messages on in-band channels. Examples include interfaces on routers, ATM switches, and

Frame Relay switches that have been enabled with an MPLS control plane.

OXC control plane functions

In order to discuss and develop OXC technologies, it’s important to look at the OXC

place in LSP’s. As presented in Figure 6, OXCs are placed inside FSC cloud. It’s capabilities are

to switch fibers and individual wavelengths from fiber to fiber (FSC and λSC). The possibility to

make wavelength conversion or switch wavebands is an option inside switch architecture. The

option at the OXC is only path’s decision (Fiber and Wavelength switching). Packets decision is

Chapter 3 - GMPLS

IP over WDM II - 21

not possible. So, the goal here is the Wavelength Switch and Fiber Switch Capability and the

related control plane functions.

Control plane

Creating a standardized control plane for optical networks gives carriers a comfort level

that they won’t be tied into a single vendor solution (as they were in the ATM days) and

provides them with a toolkit to support a variety of protection and restoration schemes, traffic

engineering in the optical layer, and provision of optical channels across their networks.

How does a control plane accomplish all this? Which protocols should be used and which

functions should they perform?

1. Link management

The use of technologies like Dense Wavelength Division Multiplexing (DWDM) implies

that we can now have a very large number of parallel links between two directly adjacent nodes

(hundreds of wavelengths, or even thousands of wavelengths if multiple fibers are used). Such a

large number of links was not originally considered for an IP or MPLS control plane, although it

could be done. Some slight adaptations of that control plane are thus required if we want to

better reuse it in the GMPLS context.

For instance, the traditional IP routing model assumes the establishment of a routing

adjacency over each link connecting two adjacent nodes. Having such a large number of

adjacencies does not scale well. Each node needs to maintain each of its adjacencies one by one,

and link state routing information must be flooded throughout the network.

To solve this issue the concept of link bundling was introduced. Moreover, the manual

configuration and control of these links, even if they are unnumbered, becomes impractical. The

Link Management Protocol (LMP) executes automated provisioning of an optical network by

discovering neighbour nodes and their capabilities. LMP allows neighbouring nodes to exchange

identities, link information, and negotiate the functions to be supported between the nodes.

LMP is also used to allow adjacent OXCs to determine IP addresses of each other and

port-level local connectivity information, such as, which port on one optical switch is connected

to which port on a neighbour. In GMPLS, the optical control plane will include capabilities of

establishing, maintaining, and tearing down optical channels in much the same way MPLS-

enabled routers establish label switched paths (LSPs).

Chapter 3 - GMPLS

22 - IP over WDM II

2. Intra domain routing protocols

Resource discovery is a good start, but it does not provide enough information to route

connections across a network.

Basically, there are two different ways for performing distributed routing:

• Link-state protocols involve reliably flooding all changes in network topology to each

network node, after which the node uses this to calculate its routing tables;

• Distance-vector protocols involve the network nodes participating in a joint

calculation of what the least-distance path is to a destination.

When allowing optical switches and network elements to disseminate information about

the network topology and resource availability, topology information can be exchanged only

using a link-state protocol. Reachability information (what end stations or nodes can be reached)

can be distributed via either link state or vector distance.

Link-state protocols have superior speed of convergence and freedom from routing loops,

while distance-vector protocols are somewhat less complex.

Routing link-state protocols such as OSPF step in at this point to distribute current

information about the topology of the network to each node. In GMPLS, extensions are being

defined to allow OSPF to be used for disseminating routing information for optical networks.

OSPF must be considered as an “intra domain” routing protocol.

3. Inter domain routing protocols

Considering OSPF as a “intra domain” routing protocol, an “inter domain” routing

protocol is necessary. This function is to be performed by BGP. New extensions are being

developed in order to adapt BGP to GMPLS arena.

4. Signaling protocols

Path setup and control protocol allows the connection to be created through switch-to-

switch signaling, without the need for network management intervention at intermediate nodes.

There are many protocols in use today that provide an analogous function in packet and

circuit networks. In circuit networks, this function is provided by SS7 or QSIG protocols, in

ATM networks by the PNNI or INNI protocols. In IP networks, a similar function is provided by

RSVP, which uses signaling to reserve resources across the IP network to support a new

Chapter 3 - GMPLS

IP over WDM II - 23

information flow. RSVP has not been extensively deployed, due to scalability concerns, but

extensions have been defined to improve its scalability.

Participants in the IETF group dealing with this issue were unable to reach a unified

approach towards setup signalling, and wound up with multiple signaling standards, leaving it to

the market to decide. This situation has unfortunately extended to GMPLS, where equivalent

modifications have been defined for both RSVP and LDP.

The OIF has similarly been unable to resolve the issue and has incorporated both RSVP

and LDP into its UNI specification. The extensions required for both RSVP and LDP in order to

support optical network signaling are significant. New parameters or formats have been defined

to take into account the need to specify time slots, wavelengths, and wavebands, instead of

packet header labels. New parameters have also been defined to allow connection requirements

such as protection and diversity to be specified.

Inband/Outband control channel

In Generalized MPLS, a control channel can be separated physically from the data

channel. Such a separation brings issues to use of RSVP and RSVP-TE for signaling.

In original RSVP, signaling is assumed to be in-band. Each RSVP message is sent hop-

by-hop between RSVP-capable routers as an IP datagram. The IP addresses of RSVP

downstream messages (Path, PathTear and ResvConf) must be set to DestAddress for the

session. Also these messages must be sent with Router Alert IP option in their IP headers. Thus,

all routers along a route examine received IP packets carrying RSVP downstream messages, but

only RSVP-aware routers recognize and process these messages. The delivery of RSVP

downstream messages to the session destination is based on IP routing scheme and unaffected by

non-RSVP routers on the path.

The assumption of in-band signaling is unchanged in RSVP-TE. Therefore the

convention referred above is inherited. However, the use of some functions is limited to cases

when all routers along the explicit route support RSVP those functions.

On the contrary, Generalized MPLS handles non-packet-switch-capable interfaces. Thus

an in-band signaling cannot be assured any longer. An RSVP messages may travel out-of-band

with respect to an LSP data channel. A Path or PathTear message should be addressed directly to

an address associated with the control plane of the node, which is known to be adjacent at the

data plane, without Router Alert option.

Chapter 3 - GMPLS

24 - IP over WDM II

Traffic Engineering

Traffic engineering (TE) must be considered as an optimization to the network

technology. Its goal is to reduce the overall cost of operations by more efficient use of bandwidth

resources, preventing a situation where some parts of a service provider network are overloaded

(congested), while other parts are idle.

Due to its applicability to transport networks where TE issues are unsurpassable, GMPLS

extends the existing signalling protocols defined for MPLS-TE (i.e. RSVP-TE or CR-LDP) in

order to support TDMC, λSC and FSC traffic engineering.

Conclusions

Communication networks must support different switching domains. Devices that

perform packet switching are needed, but also those that perform switching in time, wavelength

and space domains.

Inside the transport network core, the needs are fast switching and traffic aggregation. On

the contrary, in the edge the needs are packet switching in order to route traffic to its destination.

In communication networks lots of different devices perform those different functionalities. Such

different devices must use a unified control plane that supports and manage all those differences

and deal with available functionalities at each network equipment.

GMPLS assumes itself as a solution to implement a complete and integrated distributed

control plane supporting many different control plane operation models.

Chapter 4 - Our approach

IP over WDM II - 25


Core architecture.

Based in the study being done in this project, our approach to the GMPLS control plane

to implement in OXC is the one presented in Figure 7. This represents one of many possible

solutions though it is the most common approach done by the market.

It is important to note that the positioning of the OXC inside the network core define

some constraints to the architectural model of the control plane. I.e., inter domain routing

protocols (BGP) are not needed when considering a single carrier scenario.

Traffic Enginnering Control

FIB

OSPF - TETE Topology DB

OXC controller

LMP

RSVP - TE

Path and Wavelength selectionTraffic Enginnering Control

FIB

OSPF - TEOSPF - TETE Topology DB

OXC controllerOXC controller

LMPLMP

RSVP - TERSVP - TE

Path and Wavelength selection

Figure 7 - Control plane architecture.

It is also important considering the need of traffic engineering support in all the protocols.

The use of GMPLS and its related protocols implements distributed and automatic routing and

provisioning. But the functions used inside network core imply the usage of management-based

solutions. Centralized management solutions regarding traffic engineering are still important for

carriers.

GMPLS control plane architecture implements automatic management based in OSPF,

LMP and RSVP, but traffic engineering control is always possible.


26 - IP over WDM II

Adapted versions of OSPF-TE, RSVP-TE and LMP are needed in order to support the

GMPLS specifications. Out of band signaling, different interfaces and functionalities in network

nodes, forwarding adjacency, path bundling and unnumbered links are some of the changes to

implement.

The OXC controller includes all the functions to configure and control the OXC in order

to identify the network its capabilities and to receive and execute network configuration

instructions.

FIB will include the new data structures needed to exchange information between OSPF,

LMP, RSVP and OCX controller and to represent switch configuration and capabilities.

How different components interact

Forward Information Base (FIB)

The FIB stores the current status of the data flows traversing the switch. Specifying for

each the origin IP address, input port and wavelength set, and the respective destination IP

address, output port and wavelength set.

In OXC controller software description, in Chapter 5 – The , a more detailed description

of FIB information could be found.

Traffic Engineering Topology DB

TE Topology DB is the core component of the control plane, here is stored topology, link

forwarding adjacencies and respective control channel attributes, link TE characterization (i.e.,

bandwidth, protection, priority, node switching types).

LMP

This protocol populates the TE topology DB, with information gathered from its

neighbour and service discovery mechanisms and control channel negotiations.

LMP is the only protocol that directly accesses OXC controller to be able to perform

monitoring of the physical layer according to the switch capabilities.

OXC Controller

The OXC controller uses FIB information to setup and configure the OXC fabrics.


IP over WDM II - 27

RSVP-TE and OSPF-TE

These protocols act as common signaling and routing protocols but using the GMPLS

and traffic engineering extensions to interact with FIB and TE topology DB.

OSPF-TE will fill and update the information stored in the TE topology DB and use this

same information to distribute to its peers. OSPF not only depends on the information stored in

the TE topology DB but also from the TE configurations from TE Control. Wavelength routing

subject to other TE constraints (protection, traffic balancing, bandwidth limitations) – constraint

shortest path algorithms must be used instead of standard shortest path calculation.

RSVP-TE uses TE topology DB to be able to propagate Path and Resv messages

according to routing info. It will use the information contained in the FIB together with the

information in the TE topology DB to select appropriate switch configuration (port and

wavelength) to fulfill each signaling request (Path and Resv) messages. After a successful

signaling request (Path message) it commands the OXC controller via setting a new FIB entry. A

similar procedure is executed in when a Tear message is received, removing an entry in the FIB.

Every FIB change triggers the OXC to update the fabrics state.

Traffic Engineering control

Traffic Engineering control is not mandatory for this implementation. Its functions

concern providing network operators manual control over certain TE parameters such as load

balancing, protection requirements, recovery rules, etc. TE control has proved to be a major

advantage by allowing operators to interfere with automatic routing calculation, mapping traffic

flows to the network physical configuration.

Path and wavelength selections are the most complex subject in new generation optical

networks. The mathematical aspects of CSPF algorithms with wavelength constraints have been

a matter of extensive research, though we consider it out of our scope. Diverse solutions, most of

them combining offline and online calculations are available and could easily be integrated in

our control architecture.

Signaling (RSVP and CR-LDP)

GMPLS does not specify any profile for RSVP-TE and CR-LDP implementations in

order to support GMPLS - except for what is directly related to GMPLS procedures. It is to the

manufacturer to decide which are the optional elements and procedures of RSVP-TE and CR-


28 - IP over WDM II

LDP that need to be implemented. Some optional MPLS-TE elements can be useful for TDM,

λSC and FSC layers, for instance the setup and holding priorities that are inherited from MPLS-

TE.

Nevertheless Generalized MPLS formalizes possible separation between control and data

channels, it is unlikely that these control channels are realized via completely different service

provider networks. It is rather reasonable to consider that: there should be direct connectivity for

communication in the control plane, between immediate neighbors, which are connected

physically in the date plane. Even if there is no actual wire, there can be a logical connection by

means of IP tunnels.

Assuming such an existence of one-to-one relationship between a communication

channel in the control plane and a physical connection in the data plane, the conventional manner

of RSVP messages can be considered still effective. Particularly, it is useful within the condition

that a network topology is subject to change. Specifically, setups and teardowns of FA-LSPs in

GMPLS make a network topology transitional itself. For the support of Explicit Route (ERO), it

is even presumable that all the associated nodes in a network support RSVP. Every node on a

path examines received IP packets according to the Router Alert option. If a packet has protocol

number 46, the router recognizes and processes it as an RSVP packet. The router looks the ERO

in the packet, and then determines an appropriate next hop based on its up-to-date understanding

of the network topology.

In signaling of hierarchical LSPs, an ingress node is supposed to build an ERO that

consist of subobjects including LSP region nodes for a Path message. Using conventional RSVP

manner, an LSR can receive this message, even if ERO subsequence, which includes the LSR is

extracted by an upstream node. It gives the LSR a chance to examine the message. Thus, if the

LSR can provide a more optimal route, it may pick up and modify the message. Otherwise, if the

LSR determines to be a transit for the FA-LSP, it acts like a non-RSVP node.

In addition, when a RSVP message is delivered to terminator node directly jumping

intended transit nodes, it is even desirable to provide mechanism that can save the signaling

session from an error.

How to request an LSP

A TDMC, λSC or FSC LSP is established by sending a PATH/Label Request message

downstream to the destination. This message contains a Generalized Label Request with the type


IP over WDM II - 29

of LSP (i.e. the layer concerned), and its payload type. An ERO is also normally added to the

message, but this can be added and/or completed by the first/default LSR.

The requested bandwidth is encoded in the RSVP-TE SENDER_TSPEC object.. Specific

parameters for a given technology are given in these traffic parameters, such as the type of

signal, concatenation and/or transparency for a SDH/SONET LSP. For some other technology

there be could just one bandwidth parameter indicating the bandwidth as a floating-point value.

The requested local protection per link may be requested using the Protection Information

Object. The end-to-end LSP protection is for further study and is introduced LSP

protection/restoration section.

Additionally, a Suggested Label, a Label Set and a Waveband Label can also be included

in the message. Other operations are defined in TE. The downstream node will send back a

Resv/Label Mapping message including one Generalized Label object that can contain several

Generalized Labels. For instance, if a concatenated SDH/SONET signal is requested, several

labels can be returned.

Generalized Label Request

The Generalized Label Request is a new object to be added in an RSVP-TE Path message

instead of the regular Label Request. Only one label request can be used per message, so a single

LSP can be requested at a time per signaling message.

The Generalized Label Request gives three major characteristics (parameters) required to

support the LSP being requested: the LSP Encoding Type, the Switching Type that must be used

and the LSP payload type called Generalized PID (G-PID).

The LSP Encoding Type indicates the encoding type that will be used with the data

associated with the LSP. For instance, it can be SDH, SONET, Ethernet, ANSI PDH, etc. It

represents the nature of the LSP, and not the nature of the links that the LSP traverses. This is

used hop-by-hop by each node. A link may support a set of encoding formats, where support

means that a link is able to carry and switch a signal of one or more of these encoding formats.

The Switching Type indicates then the type of switching that should be performed on a particular

link for that LSP. This information is needed for links that advertise more than one type of

switching capability. Nodes must verify that the type indicated in the Switching Type is

supported on the corresponding incoming interface; otherwise the node must generate a

notification message with a "Routing problem/Switching Type" indication. The LSP payload


30 - IP over WDM II

type (G-PID) identifies the payload carried by the LSP, i.e. an identifier of the client layer of that

LSP. For some technologies it also indicates the mapping used by the client layer, e.g. byte

synchronous mapping of E1. This must be interpreted according to the LSP encoding type of the

LSP and is used by the nodes at the endpoints of the LSP to know to which client layer a request

is destined, and in some cases by the penultimate hop.

Other technology specific parameters are not transported in the Generalized Label

Request but in technology specific traffic. Currently, two set of traffic parameters are defined,

one for SONET/SDH and one for G.709.

Label Suggestion by the Upstream

GMPLS allows for a label to be optionally suggested by an upstream node. This

suggestion may be overridden by a downstream node but in some cases, at the cost of higher

LSP setup time. The suggested label is valuable when establishing LSPs through certain kinds of

optical equipment where there may be a lengthy (in electrical terms) delay in configuring the

switching fabric. For example physical motion inside the OXC is needed and subsequent it takes

time. If the labels and hence switching fabric are configured in the reverse direction as usual, the

MAPPING/Resv message may need to be delayed by hundreds of milliseconds per hop in order

to establish a usable forwarding path. It can be important for restoration purposes where alternate

LSPs may need to be rapidly established as a result of network failures.

Note that the use of of Suggested Label is only an optimisation.

Label Restriction by the Upstream

An upstream node can optionally restrict the choice of label of a downstream node to a

set of acceptable labels. Giving lists and/or ranges of acceptable or unacceptable labels in a

Label Set provides this restriction. If not applied, all labels from the valid label range may be

used. There are at least four cases where a label restriction is useful in the "optical" domain.

1. The first case is where the end equipment is only capable of transmitting and receiving

on a small specific set of wavelengths/bands.

2. The second case is where there is a sequence of interfaces, which cannot support

wavelength conversion and require the same wavelength be used end-to-end over a sequence of

hops, or even an entire path.


IP over WDM II - 31

3. The third case is where it is desirable to limit the amount of wavelength conversion

being performed to reduce the distortion on the optical signals.

4. The last case is where two ends of a link support different sets of wavelengths.

The receiver of a Label Set must restrict its choice of labels to one that is in the Label Set.

A Label Set may be present across multiple hops. In this case each node generates it's own

outgoing Label Set, possibly based on the incoming Label Set and the node's hardware

capabilities. This case is expected to be the norm for nodes with conversion incapable interfaces.

Routing (OSPF and BGP)

GMPLS standards committee (namely the IETF standards) has defined several extensions

to routing protocols towards their integration in GMPLS architecture. Up to now the Interior

Gateway Protocols were the main focuses with several drafts being released concerning

extensions to both ISIS-TE and OSPF-TE (please refer to reference section for draft references).

The work bases were the traffic engineering extensions introduced to OSPF and ISIS in MPLS

standardization. GMPLS requires IGP protocols to deal with a much more heterogeneous

network (optical and circuit switching networks like SDH and OTN) and the possibility of using

an out of band control channel. Following on we will briefly discuss some of this new demands.

Border Gateway Protocol was not yet addressed, but it is expected to receive the

CCAMP group attention in the near future due to its great importance in the inter-connection

between different carriers’ networks (Autonomous Systems). While IGPs will be extensively

used inside each network, BGP’s strong policies and flexibility will be essential to operators who

which to protect their routing information but keeping connection and routing updates from

neighbour networks.

The roots of GMPLS routing

GMPLS is indeed based on the Traffic Engineering (TE) extensions to MPLS, MPLS-TE.

As most of the technologies that can be used below the Packet Switch Control (PSC) level

require some traffic engineering. The placement of LSPs at these levels needs in general to take

several constraints into consideration (such as framing, bandwidth, protection capability, etc)

and to bypass the legacy Shortest-Path First (SPF) algorithm when required.

In order to facilitate constraint-based SPF routing of LSPs, the nodes performing LSP

establishment need more information about the links in the network than standard intra-domain


32 - IP over WDM II

routing protocols provide. The TE attributes are distributed using the transport mechanisms

already available in IGPs (IP encapsulation and flooding) and taken into consideration by the

LSP routing algorithm. Optimization of the routes may also require and benefit from some

external offline simulation (e.g. some wavelength routing algorithms are quite heavy to perform

online) using suitable heuristics that act as input for the actual path calculation and LSP

establishment process. Extensions to traditional routing protocols and algorithms are required to

uniformly encode and carry TE link information.

Some challenges to consider …

Some of the issues addressed by the CCAMP group were:

The MPLS label space is comparatively large, whereas there are a limited number of

wavelengths and TDM channels.

MPLS LSPs can be allocated any bandwidth value within bounds, whereas in optical

networks bandwidth can only be allocated statically and in a small number of values with little

granularity.

In MPLS there is usually only one link between two adjacent nodes while in TDM and

DWDM networks there might be thousands of labels if multiple fibres are used to interconnect

them.

Assigning IP addresses to each link in an MPLS network is common practice, assigning

IP address to every port, wavelength our TDM slot is a serious concern – even with IPv6.

Identifying which port in a network element is connected to is a complex matter in all

optical networks and one subject to high management burden and error prone.

Fast fault detection, isolation, reroute and protection typical of circuit switched networks

must be preserved.

In transparent domains, such as all optical networks, LSAs cannot be distributed in band

with current technology, this small change wreaks havoc in current LSA flooding techniques

because from now on adjacencies in control channel no longer match data channel adjacencies.

…and some possible solutions

To tackle the first two problems LSP hierarchy can be used. By aggregating traffic that

enters and exits part of the network in common nodes (tunnelling, so to speak) it is possible to

define multiple hierarchies throughout the network. Fine-grained LSPs can be used in outskirts,

while for scalability concerns bundling traffic in coarse-grained LSPs can be performed in the


IP over WDM II - 33

core. This also helps to deal with the discrete nature of bandwidth in circuit switched networks,

several small bandwidth LSPs can be tunnelled in one gigantic WDM wavelength (describe by

only one label) optimizing bandwidth usage and loosening the discreteness of the bandwidth in

circuit-switched domain.

A natural hierarchy exists in GMPLS networks due to an architectural restriction

mandating an LSP to begin and terminate in similar switching equipment (e.g. PSC LSP cannot

terminate in a WDM transparent switch). This architectural aspect will shape the architecture of

GMPLS networks as can be seen in the Figure 8. In this figure, similar to Figure 6, each cloud

represents a different multiplexing level with LSP grain getting coarser towards the inner clouds.

Notice, the forwarding adjacencies changing according switching technologies. The hierarchical

structure, bundling and forwarding adjacencies will account to the high scalability of GMPLS

networks though introducing complex questions in routing and signaling protocols

An LSP that starts in a packet switching node (a router) might be nested with several

others to form a new TDM LSP. This same LSP can be nested once again in a wavelength LSP,

which in its turn can be nested in a fibre LSP that will travel with some other fibres in the same

cable, thus forming a fibre and bundle LSPs. In what concerns routing protocols, this nesting will

help shorten routing databases and reducing the traffic in LSA flooding updates. Some extra

demands are introduced though, because no longer an adjacency in control channel (e.g. between

a PSC node and a TDM node) will represent a routing adjacency. Despite this tunnelling the

network continuous to act in a peer-to-peer configuration, each node is fully aware of the entire

path though meeting scalability concerns.


34 - IP over WDM II

PSC TDM λSC FSC

Bundle

Combining packetsinto one LSP

Combining loworder LSPs to

fill a f ibre

FA-PSC

FA-TDMFA-LSC

PSC TDM λSC FSC

Bundle

Combining packetsinto one LSP

Combining loworder LSPs to

fill a f ibre

FA-PSC

FA-TDMFA-LSC

Figure 8 - LSP hierarchical structure traditional in GMPLS networks.

Link bundling is one other technique, which presents useful in reducing even further

routing databases and traffic updates. By gathering several LSP our tunnels, which share

common characteristics, it is possible to achieve a large reduction in LSA number and size. One

suitable case is several optical fibres in the same cable. Though some degree of detail will be lost

in the bundling process it will be clearly outweighed by the scalability enhancements brought to

the link state protocols.

As early discussed it might be unpractical to attribute an IP address to every interface

connecting to nodes. Unnumbered links are used in such situations. The link is referred using a

set containing the node ID (usually its IP address) and the link ID, an ID attributed by each node,

unique in the node’s scope. Link Management Protocol will be used to synchronize the IDs used

in each end of the link and discovery of who connects to whom. Link State Protocols have been

extended to cope with the new identification set.

Link management (LMP)

LMP runs between data plane adjacent nodes and is used to manage TE links.

Specifically, LMP provides mechanisms to maintain control channel connectivity, verify the

physical connectivity of the data-bearing links, correlate the link property, and manage link

failures. A unique feature of LMP is that it is able to localize faults in both opaque and

transparent networks.


IP over WDM II - 35

LMP is defined in the context of GMPLS, but is specified independently of the GMPLS

signaling specification since it is a local protocol running between data-plane adjacent nodes. As

a result, LMP can be used in other contexts with non-GMPLS signaling protocols.

LMP control channel management is used to establish and maintain control channels

between nodes. Control channels exist independently of TE links, and can be used to exchange

MPLS control-plane information such as signaling, routing, and link management information.

Managing links

An "LMP adjacency" is formed between two nodes that support the same LMP

capabilities. Multiple control channels may be active simultaneously for each adjacency. A

control channel can be either explicitly configured or automatically selected, however, LMP

currently assume that control channels are explicitly configured while the configuration of the

control channel capabilities can be dynamically negotiated.

For the purposes of LMP, the exact implementation of the control channel is left

unspecified. The control channel(s) between two adjacent nodes is no longer required to use the

same physical medium as the data-bearing links between those nodes. For example, a control

channel could use a separate wavelength or fiber, an Ethernet link, or an IP tunnel through a

separate management network.

A consequence of allowing the control channel(s) between two nodes to be physically

diverse from the associated data-bearing links is that the health of a control channel does not

necessarily correlate to the health of the data-bearing links, and vice-versa. Therefore, new

mechanisms have been developed in LMP to manage links, both in terms of link provisioning

and fault isolation.

LMP for DWDM Multiplexer

In an all-optical environment, LMP focuses on peer communications (e.g. OXC-to-

OXC). Is possible to obtain important link information between two OXCs in the DWDM

Terminal multiplexers at the edge of the network. Exposing this information to the control plane

can improve network usability by further reducing required manual configuration and also by

greatly enhancing fault detection and recovery.

DWDM extensions to LMP are defined for use between and OXC and the DWDM muxs.

Fault detection is particularly an issue when the network is using all-optical switches (OXC).

Once a connection is established, OXCs have only limited visibility into the health of the


36 - IP over WDM II

connection. Even though the OXC is all-optical, long-haul DWDM muxs typically terminate

channels electrically and regenerate them optically, which presents an opportunity to monitor the

health of a channel between OXCs. DWDM extensions to LMP can then be used by the DWDM

mux to provide this information to the PXC.

In addition to the link information is also possible to OXC and DWDM mux to exchange

other types of information, such as information regarding alarm management and link

monitoring.

Traffic Engineering

Traditionally, a TE link is advertised as an adjunct to a "regular" OSPF or IS-IS link, i.e.,

an adjacency is brought up on the link, and when the link is up, both the regular IGP properties

of the link (basically, the SPF metric) and the TE properties of the link are then advertised.

However, GMPLS challenges this notion in three ways:

- First, links that are non-PSC may yet have TE properties; however, an OSPF adjacency

could not be brought up directly on such links.

- Second, an LSP can be advertised as a point-to-point TE link in the routing protocol, i.e.

as a Forwarding Adjacency (FA); thus, an advertised TE link need no longer be between two

OSPF direct neighbours. Forwarding Adjacencies (FA) are further described in a separate

section.

- Third, a number of links may be advertised as a single TE link (e.g. for improved

scalability), so again, there is no longer a one-to-one association of a regular adjacency and a TE

link.

Thus we have a more general notion of a TE link. A TE link is a logical link that has TE

properties, some of which may be configured on the advertising LSR, others which may be

obtained from other LSRs by means of some protocol, and yet others which may be deduced

from the component(s) of the TE link.

An important TE property of a TE link is related to the bandwidth accounting for that

link. GMPLS will define different accounting rules for different non-PSC layers. Generic

bandwidth attributes are however defined by the TE routing extensions and by GMPLS, such as

the unreserved bandwidth, the maximum reservable bandwidth, the maximum LSP bandwidth.


IP over WDM II - 37

It is expected in a dynamic environment to have frequent changes of bandwidth

accounting information. A flexible policy for triggering link state updates based on bandwidth

thresholds and link-dampening mechanism can be implemented.

TE properties associated with a link should also capture protection and restoration related

characteristics. For instance, shared protection can be elegantly combined with bundling.

Protection and restoration are mainly generic mechanisms also applicable to MPLS.

It is expected that they will first be developed for MPLS and later on generalized to

GMPLS.

A TE link between a pair of LSRs doesn't imply the existence of an IGP adjacency

between these LSRs. A TE link must also have some means by which the advertising LSR can

know of its liveness (e.g. by using LMP hellos). When an LSR knows that a TE link is up, and

can determine the TE link's TE properties, the LSR may then advertise that link to its GMPLS

enhanced OSPF or IS-IS neighbors using the TE objects/TLVs. We call the interfaces over

which GMPLS enhanced OSPF or ISIS adjacencies are established "control channels".

Conclusions

Signaling will probably the first GMPLS technology to be deployed in

telecommunications networks due to the significant overhead reduction in provisioning. It is

expected that in this early stages there will be no routing protocols running and much of the

topological configurations will be set manually at each node.

However, routing protocols are the core of distributed control architecture, and GMPLS

aims just to achieve that. Link state protocols will be fundamental to ease network management,

fasten network deployment and provide high degrees of resilience in future telecommunications

networks.

Chapter 5 – The proposed architecture

38 - IP over WDM II


The study of GMPLS protocols and its extensions is the main stream of the work done

since we started this project. It has been done considering two key aspects. One is that the

control plane to develop assumes an OXC fiber switch and wavelength switch capable (FSC and

λSC). The other was that the primary objective is to switch and route IP traffic.

Once defined the major guidelines we realized there was not enough time to develop the

complete control plane. It is necessary to create new OSPF, RSVP, LMP and OXC controller

software.

Considering different functionalities and some test possibilities, it is possible to get some

results by developing the OXC controller and the new RSVP. So we focus all our efforts in those

two components while keeping in mind the OSPF, LMP and TE integration.

OXC controller and RSVP are covered in following topics, but its important to note that

the complete control plane implementation lies on a extensive protocol integration. The OSPF

and LMP development must follow considerations presented in previous chapters.

OXC Controler

Our initial approach towards designing a full-blown control plane, according to what was

stated in chapter 4, focused primarily in developing the essential components to demonstrate our

router – namely the control program, the interface to the electronic switch controller and the

interface to the RSVP signaling module.

The first and most essential component to be designed is, of course, the OXC controller.

It works as a digital high-level controller. Its data structures keep the current configuration

settings, which are set downwards via the serial communication interface (serial port) and

upwards via the messages received from RSVP.

The switch is described using three different structures. The simplest one is the lambda,

which describes the status of an individual wavelength. Apart from simple status and failure

counters variables it includes two pointers, one for the destination wavelength (within switch

scope) – lambda structure – and one other to the destination port.


IP over WDM II - 39

Failure Counter

Delta

Failure Counter

Lambda ID

Active

Operative

Status

In tunning

Dest Lambda

Dest Port

Lambda

Failure Counter

Delta

Failue Counter

Port ID

Num. Lambdas

Operative Status

Num. Lambdas

Active

Lambdas Array

Port

Input/ Output

Next Hop Address

Wave length step

Max Wave length

Operative Status

Num. Input Ports

Num Lambdas

Num. Output Ports

Ports Array

Optical Switch

Lambda Converter

Min Wave length

Figure 9 - Control structures.

The port structure describes each switch port in a similar way to the lambda structure. It

differs from it as it specifies the port type (input or output), the remote node address (in a

sockaddr_in or similar structure), number of wavelength (total and active). The last element is an

array of lambda structures, describing the status of each port’s wavelength, as above-mentioned.

Finally, the optical switch structure stores information respective to the optical switch

technological characteristics, such as wavelength conversion capabilities, wavelength range and

step, input and output port number. It gathers all the ports in a single array, input ports firstly

placed, followed by the output ports.

The full picture can be seen above where is described a simple four by four switch, with

two wavelengths per port. Only one connection is represented: from port one, wavelength one,

bound to port three, wavelength one.


40 - IP over WDM II

Lamb

PortPort

Port

Lamb

Optical Switch

Port

Port Array

Input Ports

Output

LambLamb Lamb

Lambda

Lambda

Lambda

Port 1 Lambda

Port 2 Lambda Array

Port 3 Lambda

Port 4 Lambda

OXC controller

Figure 10 - OXC Controller Data Structures, full picture.

Output Port

Input Lambda

Input Port

Previous Entry

Output Lambda

Next Entry

FIB Entry

FIB

Hash seed

FIB Hash Table

Hash Table Size

0

FIB Hash Table

[...]

Hash Size-1

Figure 11 - FIB data structures.

The Forward Information Base is quite simple in this kind of switching architecture. It

must only contain origination and destination port and wavelength. Our view led as to a very

simple yet effective implementation. Instead of building a full set of data structures, most of

them replicating the ones allocated by the controller, we decided to use only pointers to the

original controller structures. The FIB itself is laid out as double linked bucket hash table. Some

discussion still remains about what hash seed to use, our present view is that the combination


IP over WDM II - 41

input port, input lambda and output port should prove effective in most circumstances. Figure 11

shows a detailed view of the data structures. Figure 12 gives a broader view of the data structure

usage. Figure 13 represents the full control plane, FIB data structures and their interaction in the

configuration stated above.

FIB Entry

FIB

Hash

[...]

Hash size-

0

FIB Entry

FIB Entry

Forwarding Information Base

Figure 12 - Forwarding Information Base Hash Table.

Lambd

Lambd

Lambd

OXC Controller

Data

Forward Informat

ion

Figure 13 - OXC controller and FIB joint operation.


42 - IP over WDM II

The data structures above mentioned are allocated, populated and used by the main

program. Two interfaces were designed at this initial stage. One regarding the communications

interface with the RSVP module and the other one interfacing with the fabrics.

Unix sockets were our choice to interface the RSVP, due to its simplicity, bidirectional

operation and similarity with Netlink sockets. Netlink sockets are used in Linux to allow a user

space program to send and receive data from a procedure in the kernel. As our program might be

ported in the near future to kernel space as a driver module, should a PCI card be used to control

the fabrics, it is rather important to account this future development at this time.

mb

mb

Main Program

Unix Socket

Interface

Serial

Port File Descipto

r

Communications Thread

Serial Port Communications

Thread

RSVP module

Serial Port Device

Fabrics

RS-232

Figure 14 - Overall program diagram.

The serial port was a sensible choice for a fabrics control channel for its simplicity and

readiness of implementation. An existing Stop and Wait protocol was adapted to meet our

requirements; simplicity was the key word as it had to be implemented in a microcontroller at the

fabrics side.

Both sides of the interfaces use threads to allow full asynchronous behaviour in read and

write operations. Multi-threading was preferred to Unix signals, as the formers are easy to


IP over WDM II - 43

implement and offer more predictable behaviour. Synchronization using mutexes is left for

further study.

The global operation mode is depicted in Figure 14.

Our software is still in a rather early development state. We could even say it is pre-alpha

release, so to speak. Only the main components were implemented and even in this small part of

the full architecture proposed there is much to be defined and tested. Though, we do believe it

settles a good start towards future developments. Porting to PCI control card and subsequent

control plane implementation in kernel space are reasonably easy to achieve. The most

problematic aspect of this task would be to integrate the FIB with the routing and neighbour

structures inside the kernel, which was also evaluated. All the protocols specified could run in

user space, exception made perhaps to LMP.

RSVP

The work under this topic was defining the changes to implement in RSVP in order to

support GMPLS.

Some issues must be considered in advance. The objective is the RSVP adaptation to

support OXC functions and integrate GMPLS specifications. Thus a wise approach is to use an

RSVP package already developed and tested and change to meet our demands. As justified

before, traffic engineering is a key component, so the use an RSVP-TE packet must be

considered.

Two kinds of changes must be made: OXC communication functions and RSVP’s

signaling messages.

OXC controller communication functions.

OXC controller and RSVP communication is performed using the Forward Information

Base (FIB). FIB retains adjacency and link information from LMP and switch characteristics

and current configuration from OXC controller. RSVP forwards signaling messages using FIB

information in the same way that OXC controller uses it to configure physical optics

components.

The work to do under this topic is developing RSVP to FIB read/write functions. Using

these functions RSVP will be able to setup the OXC and to process signaling messages

according to its capabilities.


44 - IP over WDM II

Note that the FIB data structure is explained under OXC controller topic.

RSVP signaling messages

RSVP messages must prepared to support new objects and new fields inside current

objects. Based on current drafts the new objects to be considered are a generalized label request,

a generalized label, suggested label and label sets.

1. Generalized label request object

A Generalized label request object is set by the ingress node, transparently passed by

transit nodes, and is used by the egress node. The message should contain as specific an LSP

Encoding Type and the Switching type may be updated hop-by-hop

0 1 2 3

0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1

Length Class-Num (19) C-Type (4 | TBA)

LSP Enc. Type Switching type G-PID

Figure 15 - Generalized Label Request object format.

A node processing a PATH message containing a Generalized Label Request must verify

that the requested parameters can be satisfied by the interface on which the incoming label is to

be allocated, the node itself, and by the interface on which the traffic will be transmitted.

The node may either directly support the LSP or it may use a tunnel. In either case, each

parameter must be checked. Nodes must verify that the type indicated in the Switching Type

parameter is supported on the corresponding incoming interface. If the type cannot be

supported, the node generates a PathErr message with a "Routing problem/Switching Type"

indication.

The G-PID parameter is normally only examined at the egress. If the indicated G-PID

cannot be supported then the egress generates a PathErr message, with a "Routing

problem/Unsupported L3PID" indication.

Bandwidth encodings are carried in the SENDER_TSPEC and FLOWSPEC objects.

Other bandwith/service related parameters are transported in the object are ignored and

transported transparently.


IP over WDM II - 45

2. Generalized label object

The Generalized Label travels in the upstream direction in Resv messages carrying

generalized label information.

As seen in Figure 16 this object is very simple including only one field, the label. The

interpretation of this field depends on the type of link over wich label is used. When using fiber

and wavelengths, as the present case, label indicates fiber or lambda to be used, from sender’s

perspective, and is 32 bit long.

0 1 2 3

0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1

Length Class-Num (16) C-Type (2 | TBA)

Label

Figure 16 - Generalized Label Object format.

Values used in this field have local significance and the receiver may need to convert the

received value to a value with that has local significance. Values may be configured or

determined using FIB information from LMP.

The recipient of a Resv message containing a Generalized Label verifies that the values

passed are acceptable. If values are unacceptable then the recipient generates a ResvErr message

with a "Routing problem/MPLS label allocation failure" indication.

Note that the presence of both a generalized and normal label object in a Resv message is

a protocol error and should treated as a malformed message by the recipient.

3. Suggested label object

The Suggested Label object is used to provide a downstream node with the upstream

node’s label preference. This permits the upstream node to start configuring it’s hardware based

on the proposed label before the label is communicated by the downstream node.

The format of a Suggested_Label object is identical to a generalized label. It is used in

Path messages. A Suggested_Label object uses Class-Number TBA and the C-Type of the label

being suggested.

As an optimisation object is possible in a previous approach to ignore it. It is important

however to support the data structures and option values from the beginning.


46 - IP over WDM II

4. Label Set object

The Label Set object is used to limit the label choices of a downstream node to a set of

acceptable labels. This can be used on a per hop basis.

There are four cases where Label Set is useful in the optical domain as referred in

“Chapter 4 - Our approach”. Label Set is used to restrict label ranges that may be used for a

particular LSP between two peers. The receiver of a Label Set must restrict its choice of labels

to one which is in the Label Set. Much like a label, a Label Set may be present across multiple

hops.

The use of Label Set is optional, but its importance in optical domain implies that it must

be implemented from the start.

0 1 2 3

0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1

Length Class-Num (TBA) C-Type (1)

Action Reserved Label Type

Subchannel 1

…

Subchannel N

Figure 17 - Label Set object format .

In Figure 17 is presented Label Set object format. Action field defines the type of the

object defined in subchannel fields.

Depending on the “Action” value those fields could represent a label list or a label range.

“Action” value also defines if the list or range is to include or exclude. A Label Set is defined via

one or more Label_Set objects. Specific labels can be added to or excluded from a Label Set via

Action zero (0) and one (1) objects respectively. Ranges of labels can be added to or excluded

from a Label Set via Action two (2) and three (3) objects respectively. When the Label_Set

objects only list labels/subchannels to exclude, this implies that all other labels are acceptable.

The absence of any Label_Set objects implies that all labels are acceptable. A Label Set

is included when a node wishes to restrict the label(s) that may be used downstream.

On reception of a Path message, the receiving node will restrict its choice of labels to one

which is in the Label Set. If the node is unable to pick a label from the Label Set or if there is a


IP over WDM II - 47

problem parsing the Label_Set objects, then the request is terminated and a PathErr message

with a "Routing problem/Label Set" indication is generated.

On reception of a Path message, the Label Set represented in the message is compared

against the set of available labels at the downstream interface and the resulting intersecting Label

Set is forwarded in a Path message. When the resulting Label Set is empty, the Path must be

terminated, and a PathErr message, and a "Routing problem/Label Set" indication is generated.

When processing a Resv message at an intermediate node, the label propagated upstream

MUST fall within the Label Set.

Note, on reception of a Resv message a node that is incapable of performing label

conversion has no other choice than to use the same physical label (i.e. wavelength) as received

in the Resv message. In this case, the use and propagation of a Label Set will significantly

reduce the chances that this allocation will fail.

Note also that the node is capable of performing label conversion may also remove the

Label Set prior to forwarding the Path message.

Communication interface

One point to be defined is how to implement the communication interface between the

host holding the control plane implementation and the OXC fabrics.

Physical optics, thermal control and a digital electronic controller are the major blocks

inside the OXC fabrics. Thermal control is used to tuning physical optics block in order to switch

wavelengths from port to port. The digital electronic controller is the “intelligent” component

inside OXC fabrics. Its function is to setup and monitoring physical optics according to

information from control plane.

In order to implement the information exchange between control plane and digital

electronics controller, a communication interface definition is required.

Our first approach, considering a test lab implementation, was to use the host serial port.

A more professional approach has been evaluated. The use of a PCI board interface seems to be

a versatile and efficient solution, specially considering evolving to an edge router configuration.

Using a PCI board.

In order to implement this solution some changes are required.


48 - IP over WDM II

In the first place it is necessary to design a PCI board. Which functions could be

implemented there? Could the software be simplified by transferring some functions to

hardware? Could some of the functions actually defined for the digital electronic controller

inside OXC fabrics be moved to PCI board? A complete study and evaluation must be performed

in order to reach a functional effective solution.

Some changes might also be essential to introduce in the OXC Controller.

In the OXC controller software there are also some changes needed. It is necessary to

implement different functions and new PCI driver modules in kernel space. This is quite more

complex than the use of serial port functions implemented in user space.

Conclusions

The OXC controller software must be defined according to the needs and capabilities of

the OXC fabrics.

OSPF, LMP and RSVP are of more complex to implementation. As we should not start

by reinventing the wheel, we found reasonable to use open software packages already available

an implement only the modifications needed. This kind of approach was tested with RSVP and

proved appropriate.

Our OXC design intends to define an IP Optical Router. A complete IP Optical router

should be fully capable of supporting packet, wavelength and fiber switching/routing.

Considering fully passive optical equipment, packet switching is not yet possible, a first

step, due to those technological issues, is to implement wavelength and fiber switching/routing.

As GMPLS implements a unique integrated control plane, a fully functional OXC should

be able to exchange information with all types of GMPLS nodes. This includes the reception and

forwarding of GMPLS Path and Resv messages, which could be originated in an SONET/SDH

multiplexer, an ATM switch or even an IP MPLS Edge Router upgraded to a GMPL control

plane.

Our OXC goes beyond protocol stacks and framing as it appears completely transparent

to upper layer protocols. In this context it makes no sense to define it as an IP Router. What

makes it comparable to IP traditional router is its GMPLS control plane, which uses protocols

bred in the IP world but extended to support non-IP traffic.

Chapter 6 – Conclusion

IP over WDM II - 49

Chapter 6 – Conclusions

The evolution of optical networks takes us towards simplification and standardization of

the control structures. The trend is to carry IP, with the least overhead possible, in the optical

medium. Gigabit Ethernet will play a dominant role in this approach and latest developments in

this technology just confirm that.

As communication switched networks become more complex, centralized control

systems strive to keep up with the growing pressure for fast network deployment and

reconfiguration. Different equipments have different switching capabilities, and control protocol

stacks must be able to deal with all those differences in the same network or even in the same

link. GMPLS with its peer decentralized control model and traffic engineering extensions

promises to ease network management in a more efficient and cost-effective way.

GMPLS assumes itself as a solution to implement a complete and integrated control plane

with many different operation models.

The need of overhead reduction in provisioning makes signaling the first GMPLS

technology to be deployed in telecommunications networks. In early stages much of the

topological configurations will be set manually at each node, however, routing protocols are the

core of distributed control architecture. Link state protocols will be fundamental to ease network

management, fasten network deployment and provide high degrees of resilience.

The switching domains concept must be always present. According to GMPLS

specifications all nodes must be able to interact with its neighbours at control plane level no

matter which are their switching capabilities. Thus two approaches are possible:

- If the node is able to intercept and understand data flows traversing it, more complex

functions could be performed based on higher-level information. More valuable data

could also be shared with the entire network.

- If the node does not understand those data flows, it acts transparently based on

configuration messages receive by its neighbours. When acting transparently the path

configuration messages either are manually defined or start in “non-transparent” nodes

following the path based in a hop-by-hop basis.

The use of OSPF and LMP protocols and the independent control plane concept present

in GMPLS, allows the automatic routing and provisioning to be used in transparent nodes.

Chapter 6 – Conclusions

50 - IP over WDM II

An OXC fiber and wavelength switch capable (FSC and λSC) acts transparently

regarding data flow type crossing it. It switches WDM channels independently of what is

transported in them.

Considering this, once the control plane is fully implemented, our OXC will act as a

network core WDM Optical OXC integrated in the GMPLS distributed architecture. Performing

automatic, on client demand, provisioning, wavelength routing and routing information

distribution in peer-to-peer model.

It will be able to switch GMPLS WDM channels no matter what type of data being

carried, such as IP over Gigabit Ethernet frames or ATM cells over SDH frames, integrated in

distributed control architecture.

References

IP over WDM II - 51

References

Papers

[1] Ghani, N., Dixit, S., Wang, T., On IP-over-WDM Integration, IEEE communications Magazine, March 2000.

[2] Banerjee, A., Drake, J., Lang, J.P., Turner, B., Kompella, K., Rekhter, Y, GMPLS: An Overview of Routing and Management Enhancements, IEEE communications Magazine, January 2001.

[3] Jerran, N., Farrel, A., MPLS in Optical Networks, http://www.dataconnection.com, October 2001

[4] Banerjee, A., Drake, J., Lang, J.P., Turner, B., Kompella, K., Rekhter, Y, GMPLS: An Overview of Signaling Enhancements and Recovery Techniques, IEEE communications Magazine, July 2001.

[5] Hernandez-Valencia, E., Scholten, M., Zhu, Z., The Generic Framing Procedure (GFP): An overview, IEEE communications Magazine, May 2002.

[6] Awduche, D., Rekhter, Y, Multiprotocol Lambda Switching: Combining MPLS Traffic Engineering Control with Optical Crossconnects, IEEE communications Magazine, March 2001.

[7] O. Aboul-Magd, et. al., Automatic Switched Optical Network (ASON) Architecture and Its Related Protocols", Internet Draft, draft-ietf-ipo-ason-01.txt, June 2002.

[8] Matsuura, N., Katayama, M., Shiomoto, K., Requirements for using RSVP-TE in GMPLS signaling, Internet Draft, draft-matsuura-gmpls-rsvp-requirements-01.txt, June 2002.

[9] Ashwood-Smith, P. et al, Generalized MPLS - Signaling Functional Description, Internet Draft, draft-ietf-mpls-generalized-signaling-08.txt, April 2002.

[10] Ashwood-Smith, P. et al, Generalized MPLS Signaling - RSVP-TE Extensions, Internet Draft, draft-ietf-mpls-generalized-rsvp-te-07.txt, April 2002.

[11] Lang, et al. Link Management Protocol, Internet Draft, draft-ietf-mpls-lmp-02.txt, March, 2001.

[12] Kompella, K., et al, Routing Extensions in Support of Generalized MPLS, Internet Draft, draft-ietf-ccamp-gmpls-routing-00.txt, September, 2001.

[13] Kompella, K., Rekhter, Y., Signaling Unnumbered Links in RSVP-TE, Internet Draft, draft-ietf-mpls-rsvp-unnum-01.txt, February 2001.

[14] 10 Gigabit Ethernet Alliance, 10 gigabit Ethernet – an introduction, September 2000

References

52 - IP over WDM II

[15] Acterna Communications, Packet Over Sonet/SDH Pocket Guide, 2000

[16] Agilent Technologies, An overview of ITU-T G709, Optical Transport Network

[17] Optical Internetworking Forum, OIF UNI 1.0 – Controlling Optical Networks, 2001

[18] Papadimitriou, Dimitri, Enabling Generalised MPLS Control for G.709 Optical Transport Networks, Alcatel, October 2001

[19] Kompella, et al., OSPF Extensions in support of Generalized MPLS, IETF, draft-ietf-ccamp-ospf-extensions.txt, 2000

[20] Fredette, Andre, et al., Link Management Protocol (LMP) for DWDM Optical Line Systems, IETF, draft-ietf-ccamp-lmp-wdm-00.txt, February 2002

[21] Mannie, Eric, et al., Generalized Multi-Protocol Label Switching (GMPLS) architecture, IETF, draft-ietf-ccamp-gmpls-architecture-01.txt

[22] Rosen, E., et al., Multi-Protocol Label Switching Architecture, IETF,rfc-3051,2001

[23] Xu, Y., et al, A Framework for Generalized Multi-Protocol Label Switching (GMPLS), IETF, draft-many-ccamp-gmpls-framework-00.txt, January 2002

[24] Papadimitriou, Dimitri, et al., GMPLS signalling extensions for G.709 Optical Networks Control, IETF, draft-ietf-ccamp-gmpls-g709-02.txt

[25] Linux How-To’s at www.linuxdoc.org

Kernel How-To

Linux Networking Overview How-To

Linux IPMasquerade How-to

Linux IPChains How-To

[26] Herrin, Glenn, Linux IP Networking – A guide to the Implementation and Modifications of the Linux Protocol Stack, May 2000

[27] Rusling, David A., The Linux Kernel, January 1999

Poster

Metropolitan GMPLS Network

ATM MPLS Network

Metropolitan SDH Network

All Optical Network / OTN 10 Gigabit EthernetMPLS Network

OXC - Optical CrossConector

SDH Ring

OXC

OXC OXC

IP Network

IP Network

E-LSR

Router IP

IP Network

IP Network

Fabrics

IP over WDMDesigning an IP Optical Router

Supervisors• Henrique Salgado (UOSE)• Manuel Ricardo (UTM)

Students• Igor Terroso (UOSE)• Joel Carvalho (UOSE)

The road to the OXC

Router IP

• Bruno Leite (UTM)• Fernando Pinto (UTM)

http://www.fe.up.pt/~ee97159/pstfc

Control Plane

Voice Network

TDMTDM

Router IP w/ TDM

λSR

λSR

λSR

λSR

E-LSR

E-LSR

LSR

ATM

ATM

ATM

ATM

Cellular Network

Cellular Network

IP Network

Router IP

LSR

LSR

λSR

Voice Network

OXC

OpticalLayer

SDHATM

IP

IPGbE

OpticalLayer

Traffic Enginnering Control

FIB

OSPF - TETETopologyDB

OXC controller

LMP

RSVP - TE

Path and Wavelength selection

Thermal controller

Digital Electronic controller

Optical components

IPGbE

WDM Layer

CommunicationsInterface

OpticalLayer

SDH

ATM

IP

IP

Traditional SDH approaches.- IP/ATM/SDH- POS (IP/PPP/HDLC/SDH)

Traditional SDH based approaches, withWDM layer.

IP over WDM overlayapproach

Direct MPλSapproach


HDLC

IP

GbE

IP

SDH

ATM

IP

WDM / OTNOptical Layer

Gigabit Ethernetover SDH Aproach

PPP

IP

HDLC

PPP IP

GbEGFP

WDM

PPP Eth

IP

GFP

PPP Eth

ATM

IP

Protocol Stack Evolution for IP over WDM SolutionsFabrics pictures

HDLC

OpticalLayer

SDH

PPPIP

SDHWDM Adapt.

Optical Layer

HDLCPPPIP

WDM Layer

IP

GFPPPPEth

IP

GbESDH

WDM Adapt.

Optical Layer

SDH

IPATM

WDM Adapt.

Optical Layer

OXC

λSR

Objectives• Developing OADM Architectures• Enhancement of the OADM Architecture

• Define GMPLS router architecture• Specify and develop router control plane

Optical Add – Drop Multiplexer 2

Drop 2

λ2

Add

Input

λ1, λ2, λ3

Output

λ1, λ3λ1

λ2λ2

λ3

Optical Add – Drop Multiplexer 1

Drop 1

λ2

Add

Input

λ1, λ2, λ3

λ2

Output

λ1, λ3

1548 1549 1550 1551 1552 1553 1554-25

-20

-15

-10

-5

0

A

Tra

nsm

issi

on(d

B)

λ (nm)

1548 1549 1550 1551 1552 1553 1554-25

-20

-15

-10

-5

0

C

D

Tra

nsm

issi

on(d

B)

λ (nm)

1548 1549 1550 1551 1552 1553 1554-35

-30

-25

-20

-15

-10

-5

0

Output 1Output 2

Opt

ical

Pow

er(d

B)

λ (nm)

1548 1549 1550 1551 1552 1553 1554-35

-30

-25

-20

-15

-10

-5

0

Output 1Output 2

Opt

ical

Pow

er(d

B)

λ(nm)

1546 1548 1550 1552

-60

-50

-40

-30

-20

-10

0

Input 1Input 2

Tra

nsm

issi

on(d

Bm

)

λ(nm)

1548 1549 1550 1551 1552 1553 15540,0

0,2

0,4

0,6

0,8

1,0

Tra

nsm

issi

on(d

B)

Ref

lect

ivity

λ (nm)

-16

-14

-12

-10

-8

-6

-4

-2

0

OXC optical components Peltier controller – Power circuit Peltier controller - Analogical

Reducing the protocol stack:Eliminating multiple framing levels

Reducing overhead

Use of IP as the convergence layer

Using WDM as Transport Layer:Simple thin optical layer

Use of additional sub-layer for carrier-class reliability

Input signals

Bar-state Output

Input 2 crossingGrating transmission / reflection

1546 1548 1550 1552-80

-70

-60

-50

-40

-30

-20

-10

0

Output 1Output 2

Tra

nsm

issi

on(d

Bm

)

λ(nm)

1548 1549 1550 1551 1552 1553 1554-25

-20

-15

-10

-5

0

A

B

Tra

nsm

issi

on(d

B)

λ (nm)

1546 1548 1550 1552-80

-70

-60

-50

-40

-30

-20

-10

0

Output 1Output 2

Tra

nsm

issi

on(d

Bm

)

λ(nm)

Input signal – WDM source A – Output signal B – Dropped signal 1 C – Dropped signal 1 D – Dropped signal 2 λM = λ2 λM between two channels

Final OXC Architecture• Reduced crosstalk

• High performance levels

• Good channel selectivity

• High scalability

Implementation focused on building OXC controller and the Forwarding Information Base

OXC controller

Main program and two threads

Main program manages OXC according to what is written in the FIB by the RSVP module, via UNIX socket read by communications thread

Main program responds to RSVP module via UNIX socket

Serial port chosen to control the OXC fabrics due to simple implementation – simple Stop and Wait data link protocol adapted

Threads act as listeners, providing fully asynchronous communication (preferred for simplicity to UNIX signals)

FIB implemented as a double linked bucket hash table

Each entry contains pointers to OXC controller data structures (conceptually: input port, input wavelength, output port, output wavelength)

Hash function key – input port, output port, input wavelength

RSVP

New objects and messages required:

. Generalized label request object . Generalized label object

. Sugested label object . Label Set Object

Output 1

Output 2

Optical Cross – Connect 2

Input 1

λ2, λ3

Input 2

λ1

λM

λM

Patent pending

Optical Cross – Connect 1

Input 2

Output 1

Output 2

Input 1

λ1, λ2, λ3

λM

US Patent 5940551

Published article: I. Terroso, J.P. Carvalho, O. Frazão, M. Ricardo, H.M. Salgado,

“Avaliação de duas arquitecturas de OADM Baseadas em circuladores ópticos e redes de Bragg em fibra óptica.”, Física 2002Published article: J.P. Carvalho, I. Terroso, O. Frazão, M. Ricardo, H.M. Salgado, “Comutador óptico (OXC)Baseado em circuladores ópticos e numa rede de Bragg em fibra óptica.”, Física 2002

Submitted Patent: “Comutador óptico (OXC) 2x2 portas para sistemas demultiplexagem em comprimento de onda e escalável a NxN portas”

RS

-232

Lambda

Lambda

Lambda

Main Program

UnixSocket

Interface

SerialPort FileDesciptor

CommunicationsThread

Serial PortCommunications

Thread

RSVP module

Serial PortDevice

Project objective reached

Documents

IP over WDMpaginas.fe.up.pt/~hsalgado/pstfc/Projecto_IP_WDM_Final.pdfFigure 15 - Mach-Zehnder Interferometer; b) Three Mach-Zehnder Chain .....22 Figure 16 - Basic acousto-optic tunable