Constraint based routing due to physical impairments in automatically switched transport networks

TABLE OF CONTENTS

  ACKNOWLEDGEMENTS  5

  ERKLÄRUNG  6

  EINWILLIGUNG  6

  ABSTRACT  7

  SYMBOLS AND ABBREVIATIONS  8

1  INTRODUCTION  11

1.1  NA  

  Introduction  11

1.2  NA  

  Optical communications systems 11  

1.3  NA  

  Aims and objectives 14  

1.4  NA  

  Structure of this thesis 15  

2  THEORY  16

2.1  NA  

Chapter overview  16  

2.2  NA  

  Maxwell s equations 16  

2.3  NA  

  Reflection and refraction 17  

2.4  NA  

  Dispersion and loss 18  

2.5  NA  

  Effective length and area 21  

2.6  NA  

  Nonlinear Schrödinger equation (NSE) 21  

TABLE OF CONTENTS

3  COMPONENTS  23

3.1  NA  

Chapter overview..............................................................................................  23  23

3.2  NA  

  Modulation format  23

3.3  NA  

  Filter  25

3.4  NA  

  Assessment of the signal quality  27

3.4.1  NA  

  Eye opening penalty  27

3.4.2  NA  

  Q-factor  28

4  LINEAR DEGRADATION EFFECTS  31

4.1  NA  

Chapter overview..............................................................................................  31  31

4.2  NA  

  Inter-channel crosstalk  31

4.2.1  NA  

  Continuous-wave (CW) case  31

4.2.2  NA  

  Non-return to zero (NRZ) case  33

4.2.3  NA  

  Return to zero (RZ) case  36

4.3  NA  

  Narrow-band spectral filtering  40

4.4  NA  

  Optical demux filter optimization  42

4.5  NA  

  Dispersion  45

4.5.1  NA  

  Group velocity dispersion (GVD)  45

4.5.2  NA  

  Third-order dispersion (TOD)  50

5  NONLINEAR DEGRADATION EFFECTS  53

5.1  NA  

Chapter overview..............................................................................................  53  53

5.2  NA  

  Four-wave mixing (FWM)  53

5.2.1  NA  

  Approximation of the signal-to-crosstalk ratio  53

5.2.2  NA  

  Simulations of the NRZ case  57

5.2.3  NA  

  Simulations of the RZ case  63

5.3  NA  

  Self-phase modulation (SPM)  66

5.4  NA  

  Stimulated Raman Scattering (SRS)  72

5.4.1  NA  

  Theoretical considerations  72

3  NA  

TABLE OF CONTENTS

5.4.2  NA  

  Continuous-wave (CW) case 78  

5.4.3  NA  

  Non-return to zero (NRZ) case 78  

5.4.4  NA  

  Effects of group-velocity dispersion (GVD) 81  

5.4.5  NA  

  Simulations of multi-span systems with GVD 84  

6  EXAMPLES OF NETWORK PLANNING  85

6.1  NA  

Chapter overview  85  

6.2  NA  

  Variation of the channel input power 85  

6.3  NA  

  Variation of the number of WDM channels 86  

6.4  NA  

  Variation of the channel spacing 88  

7  CONCLUSION AND OUTLOOK  90

8  REFERENCES  93

4

ACKNOWLEDGEMENTS

ACKNOWLEDGEMENTS

First and foremost, I wish to acknowledge sincerely my project supervisor Dipl.-Ing. J. Kissing for his help and guidance throughout the whole project. His excellent supervision was essential for the project work to be carried out.

In addition, I would like to thank all members of the institute of high-frequency techniques for their help and support. Without their experience many problems could not have been sorted out that quickly. In this work the simulation software PHOTOSS has been used extensively. The members of the institute of high-frequency techniques have developed this software package and I had the opportunity not only to use it, but also to read the source code and to add my own functions.

Moreover, I am grateful to the CUSANUSWERK for granting me a scholarship, which was a great help for my studies, and I gratefully appreciate the support by E-FELLOWS.NET and by NOKIA.

Finally, I would like to express my deepest gratitude to my parents for their encouragement and support throughout my studies.

5

ERKLÄRUNG

ERKLÄRUNG

Ich versichere, dass ich diese wissenschaftliche Arbeit selbständig verfasst habe und keine anderen als die angegebenen Quellen und Hilfsmittel verwendet habe. Die Stellen der Arbeit, die anderen Werken in Wortlaut und Sinn entnommen worden sind, wurden in jedem einzelnen Fall unter Angabe der Quelle als Entlehnung kenntlich gemacht. Das Gleiche gilt auch für beigelegte Skizzen oder Darstellungen. Die Arbeit hat in gleicher oder ähnlicher Form noch keiner anderen Prüfungsbehörde vorgelegen.

Dortmund, den 7. Februar 2006

EINWILLIGUNG

Hiermit erkläre ich mich damit einverstanden, dass diese wissenschaftliche Arbeit nach den Bestimmungen des § 6 Abs. 1 des Gesetzes über Urheberrecht vom 09. 09. 1965 in die Bereichsbibliothek aufgenommen und damit für Leser der Bibliothek öffentlich zugänglich gemacht wird. Ferner bin ich damit einverstanden, dass gemäß

§ 54 Abs. 1 dieses Gesetzes Leser zu persönlichen Zwecken Kopien anfertigen dürfen.

Dortmund, den 7. Februar 2006

6

ABSTRACT

ABSTRACT

Next generation optical communication systems will be characterized by increasing data rates, high signal powers and dense wavelength division multiplexing (DWDM). In future-optical networks channels will be routed through complex, meshed networks (ASTN, Automatically Switched Transport Networks, ITU-T Recommendation G.808 0/Y.1304). These networks will be able to setup transparent optical paths without converting the optical signals to electrical signals. In all-optical networks the physical impairments and degradation effects play an important role. There is a multitude of degradation effects like dispersion, noise, crosstalk, fiber nonlinearities, polarization dependent loss, etc. To enable a fast setup and the best choice of one of the available paths, the signal quality along the whole transmission distance has to be evaluated very fast.

Ideally, only a single figure of merit (FOM), e.g. the bit error rate (BER), will be computed, which incorporates all degradation effects. Therefore it is important to characterize the different physical impairments analytically. Signal distortions can be measured by an eye opening penalty (EOP) and degradation effects due to noise by the optical signal-to-noise ratio (OSNR). The goal is to find and calculate these impairments from the signal parameters (modulation format, data rate, duty cycle, channel spacing, etc.) as well as the route parameters (fiber lengths and parameters, EDFA powers, etc.). Due to the need of fast routing algorithms, time-consuming numerical methods or a complete system simulation are not practical. In addition, it is not possible to linearly accumulate the different degradation effects.

The focus of this work will be to find analytical or heuristic formulas for each degradation effect. These approximation formulas will be compared to the results obtained from a complete simulation of a reference system with the help of PHOTOSS.

7

SYMBOLS AND ABBREVIATIONS

SYMBOLS AND ABBREVIATIONS

ASE Amplified spontaneous emission

ASTN Automatically switched transport network

BER Bit error rate

BERT Bit error rate testing device

BR Bit rate

CS-RZ Carrier suppressed return to zero

CW Continuous wave

DCF Dispersion compensating fiber

DEMUX Demultiplexer

DGD Differential group delay

DMD Dispersion management device

DSF Dispersion shifted fiber

DUCS Distributed undercompensation scheme

DWDM Dense wavelength division multiplexing

EDFA Erbium-doped fiber amplifier

EO Eye opening

EOP Eye opening penalty

FEC Forward error correction

FOCS Full inline- and optimized post-compensation scheme

FOM Figure of merit

FSK Frequency shift keying

8

SYMBOLS AND ABBREVIATIONS

FWHM Full-width at half maximum

FWM Four-wave mixing

GaAs Gallium arsenide

GI Graded index fiber

GVD Group velocity dispersion

IM Intensity modulation

ISI Intersymbol interference

Laser Light amplification by stimulated emission of radiation

LEAF Large effective area fiber

NEB Noise equivalent bandwidth

NMS Network management system

NRZ Non-return to zero

NSE Nonlinear Schrödinger equation

NZDSF Non-zero dispersion shifted fiber

OADM Optical add/drop multiplexer

OBS Optical burst switching

OSA Optical spectrum analyzer

OSNR Optical signal-to-noise ratio

OTN Optical transport network

OXC Optical crossconnect

PDF Probability density function

PDL Polarization dependent loss

PMD Polarization-mode dispersion

9

SYMBOLS AND ABBREVIATIONS

PRBS Pseudo random bit sequence

PSK Phase shift keying

RX Receiver

RZ Return to zero

Sech Hyperbolic secant

SI Step index fiber

SPM Self-phase modulation

SRS Stimulated Raman Scattering

SSF Split-step Fourier method

SSMF Standard single mode fiber

TOD Third order dispersion

TX Transmitter

UNI User network interface

VOA Variable optical attenuator

WDM Wavelength division multiplexing

XPM Cross-phase modulation

10

CHAPTER 1 I NTRODUCTION

1 INTRODUCTION

1.1 Introduction

The significance of fiber optics has grown rapidly in the last decades. In 1980, the first fiber networks were installed in the US and not earlier than 1988, the first transatlantic optical fiber cable (TAT-8) was installed. It operated at mere 140 Mbit/s and used electrical regenerators [22].

Optical fibers are used in different fields of science and technology. First of all, fiber optical devices are used in modern optical telecommunications. The main arguments for the use of fiber optics in this field are the extremely low loss and the very high bandwidth. However, there are also other fields, which make use of fibers extensively. In medicine – for example – the requirements are quite different, and the small size and high flexibility of the fiber are the main arguments in favour.

Therefore it is understandable that this area of science is a field that needs a lot of research to satisfy not only the demands of future telecommunications but also of many other fields.

1.2 Optical communications systems

The invention of the telephone by Graham Bell allowed global communication via telephone networks. At first twisted pair wires were used and were later replaced by coaxial cables with higher data rates and lower loss. As the amount of the transmitted data increased continuously, this technology also reached its limit very fast.

The invention of optical fibers was a revolution for long distance communications, because it allowed transmitting signals of very high bandwidths. At first the losses in the fiber were very high, but already in 1966, Kao predicted an attenuation of 3 dB/km, and in 1968, a fiber with 20 dB/km was realized by Maurer and Corning.

Today fibers with an attenuation as low as 0.2 dB/km are used. Another important step was the invention of the laser by Maiman in 1960. With the laser it was possible to use a coherent light source, which is needed to couple a light beam into the fiber. Further inventions like the GaAs laser, EDFAs, low-loss fibers and sophisticated multiplexing techniques made it possible to transmit several data streams simultaneously over a single optical fiber (WDM). Due to the exponential growth of the world-wide-web, the demand for very fast broadband transmission media to exchange large amounts of data

11

1.2 OPTICAL COMMUNICATIONS SYSTEMS

all over the world has grown rapidly. Optical fibers, which are now commonly used, are the best choice for this application. Modern optical fiber amplifiers offer the possibility to compensate the loss occurring during the transmission.

Two of the main problems in the transmission of light signals are the occurring dispersion and nonlinear effects. These effects limit the maximal bandwidth or the length of the fiber. Being able to compensate these degradation effects is becoming even more important, since erbium doped fiber amplifiers (EDFA) are available and there may be very long distances of up to several thousand kilometres without (electrical) regeneration of the signal. The idea of the erbium doped fiber was developed at Southampton University. EDFAs make it possible to amplify a weak optical signal (in the C-band around 1545-1560 nm) without converting it into an electrical signal. Recent developments employ differently doped materials, like thulium doped fibers, to amplify different band regions (e.g. S-band at 1470-1510 nm) [15]. Instead of opening up new bandwidth regions to increase the amount of data to be transmitted, also the spectral efficiency of a signal could be enhanced [17]. This could be done, on the one hand, by utilizing polarization multiplexing, which means that two signals with orthogonal polarization states are launched simultaneously. On the other hand, the channel spacing could be reduced, which requires precise wavelength stabilization of the laser sources. Currently wavelength tolerances can be stabilized within about r150 MHz.

For transmitting optical signals over a long distance, normally NRZ-modulation is employed (non-return to zero). NRZ allows minimizing the signal distortions due to fiber nonlinearities and chromatic dispersion. The energy of a single bit is distributed uniformly over the whole bit period, thus both the peak power as well as the optical bandwidth are relatively small. Unfortunately, the use of NRZ modulation is limited to low signal powers and bit rates. If a higher optical signal-to-noise value (OSNR) is required and as a consequence the signal power is increased, signal distortions will occur due to the nonlinear optical effect of self-phase modulation (SPM), which cannot be compensated.

A new approach is the use of solitons in optical fibers. This technique allows compensating the occurring dispersion by the interaction of the nonlinear self-phase modulation of the pulse and the nonlinearities of the optical fiber. The soliton is RZ- modulated with a hyperbolic secant pulse shape. The utilization of soliton pulses (or “sech”-pulses) makes it unnecessary to employ dispersion compensating fibers (DCF).

12

CHAPTER 1 I NTRODUCTION

Today most of the existing optical networks are circuit switched. An optical path (wavelength) is setup permanently for a specific user. Changes can only be made by the network operator, therefore the net is called “static”. In the near future all-optical networks will be in use. In these networks an optical path can be setup automatically over several nodes without the need of converting the optical signal into an electrical one. Such a system is also referred to as an “automatically switched transport network” (ASTN). One important advantage of these systems is that the data – that is passed through a node – remains in the optical domain, and only the data intended for that node has to be processed electronically. Hence, the burden on the underlying electronics at the node for very high data rates will be reduced significantly [3].

The key network elements needed for all-optical networks are optical add/drop multiplexers (OADMs) and optical crossconnects (OXCs). An OADM takes in a WDM signal at multiple wavelengths and selectively drops some of these wavelengths. It also selectively adds wavelengths to the composite outbound signal. OXCs have a large number of input and output ports. They are able to switch wavelengths from one input port to another. Both of these devices may incorporate wavelength conversion capabilities [7]. In today’s optical transport networks (OTN) an optical path consists of a cascade of optical transmission lines (fibers) and optical switching centres (OXCs). The switching fabric of an OXC can be realized with mirrors. The whole system is controlled by a centralized network management system (NMS). A great disadvantage of these systems is that the path setup has to be done manually. Thus, such networks are also referred to as “semi-static”. In the future the path setup will be performed automatically by the network management. A subscriber may request a certain transmission bandwidth over the user network interface (UNI). This end-to-end signalling functionality also enables new services like bandwidth-on-demand or optical virtual private networks (OVPN) [5]. Another point important to mention is the one of diversity [6]. Two light paths are said to be diverse, if they have no single point of failure. In the event of system failure (e.g. through cut fiber cables), diversity allows to route the data streams simply through other routes to the desired destination. It should also be stated that transparent networks not only offer advantages. A transparent network must standardize technology choices at the outset, and thus cannot exploit new developments such as broader amplifier bandwidths or narrower channel spacing as they become available [21].

Also optical packet switching could become reality in the near future. Optical packet switching will not only route a certain wavelength from one point to the other, but allows routing single data packets. These packets consist of a specific header, which

13

1.3 AIMS AND OBJECTIVES

contains all-important routing information, and a payload, which can be flexible and may change with the amount of data to be transmitted. In contrast to conventional packet switching, the payload is remaining completely in the optical domain, and no electrical detection of the signal is needed. Ideally, also the routing of these packets would be performed in the optical layer, but in practice, certain functions, such as processing the header and controlling the switch, will be relegated to the electronic domain [23]. Optical packet switching can also be subdivided in conventional packet switching and optical burst switching (OBS). In OBS, a certain number of packets with the same destination are combined to a burst. The header of a burst is transmitted in a separated channel from the payload, and the path is setup before the payload will be sent. This may be advantageous because no optical delay line is needed during the setup of the path, and the time needed for the reconfiguration of the path may be much larger than in conventional optical packet switching. In today’s first optical packet switched networks OBS is already utilized.

A few words should also be said about the routing algorithms. In general, routing algorithms can be ranked according to their blocking probability, connection setup time and bandwidth requirements for control messages. Examples for such algorithms are the alternate shortest-path routing (ASPR) or multi-path routing (MPR) [14].

1.3 Aims and objectives

The aim of this project is to characterize the different physical impairments on an optical signal, which is transmitted in an optical network, analytically. In contrast to small optical networks, in transparent optical networks it is mandatory to consider the physical impairments. It cannot be assumed that for all optical paths the needed BER can be reached. These physical impairments, as well as economical factors, are considered in “constraint based routing”. It has been shown that the total network cost can be decreased drastically with the help of constraint based routing [12]. In future- systems not only a single dominant degradation effect (e.g. PMD) needs to be considered. Due to the high signal power a multitude of nonlinear effects, which accumulate differently, will be deteriorating the signal. Beyond their importance for constraint based routing, the knowledge of the physical impairments is also important for network planning to find easy design rules.

In this work the different degradation effects will be investigated individually, and for each effect an analytical approximation formula will be given. The results obtained from these approximation formulas will be compared to the results obtained from a full- scale simulation with PHOTOSS. Furthermore, it will be important to show how these

14

CHAPTER 1 I NTRODUCTION

effects change, if multiple fiber spans are considered. Finally, a single figure of merit (e.g. EOP or BER) will be calculated to evaluate the signal degradation.

1.4 Structure of this thesis

After a brief introduction into optical communications has been given in this chapter, essential parts of the theory needed for understanding the propagation of light will be explained in the next chapter. The theory is based on Maxwell’s equations as well as on the nonlinear Schrödinger equation (NSE).

Chapter 3 will introduce the different parameters needed for the characterization of a signal (modulation format, data rate, duty cycle, etc.) and the different values used for the assessment of the signal quality (BER, OSNR, Q-factor, etc.).

In Chapter 4 and Chapter 5 the different degradation effects, first the linear ones and afterwards the nonlinear ones, will be examined. For each effect an analytical approximation will be given, and the results will be assessed and compared to the results of the full-scale simulation.

Chapter 6 will be built upon the results of the previous chapters and will present some easy network planning considerations.

In the last chapter a conclusion will be drawn, and the whole project will be critically reviewed. Some suggestions for further work in this area will be proposed.

15

2.1 CHAPTER OVERVIEW

2 THEORY

2.1 Chapter overview

This chapter will introduce the basic theory, which is required to understand the

principles of light propagation. First of all Maxwell’s equations will be explained. The

principles of reflection and refraction will be mentioned briefly. Afterwards the major

problems in optical communications namely dispersion, loss, as well as the nonlinear

effects – described by the nonlinear Schrödinger equation (NSE) – will be discussed.

2.2 Maxwell’s equations

The theory explaining the propagation of light is split into two different parts. It is

said that light has a dual nature. There is on the one hand the wave theory by Hooke and

Huygens and on the other hand the corpuscular theory by Newton [13]. In 1864,

Maxwell combined the equations of electromagnetism and showed that they suggest the

existence of transverse electromagnetic waves. These equations can be written either in

integral or in differential form. For the analysis of light propagation it turns out that the

differential form is more convenient. In eqs. (2.1) - (2.4) these equations are given in

differential form [2]:

wD

u ’

H

E u ’ ˜ ’ D

˜ ’ B

where H is the magnetic field intensity [A/m], J the current density [A/m²], D the

electric flux density [C/m²], E the electric field intensity [V/m], B the magnetic flux

density [Wb/m²] and U the charge density [C/m³]. The current density J and the charge

density U can be neglected in this project because only dielectric materials will be

utilized.

16

CHAPTER 2 THEORY

The electric flux density D and the electric field intensity E are related by the

following equation:

D

where P stands for the induced electrical polarization.

Similarly, the relation describing the magnetic properties is given by:

B

Because all materials are considered isotropic, the permittivity H and the permeability

µ are scalars and not tensors. The wave equation that describes light propagation in

optical fibers can be easily obtained from the above equations:

u ’ u ’ E

0

where c 0 is the speed of light in vacuum.

If we include only the third-order nonlinear effects, the induced polarization will

consist of two parts such that

r P

) , (

t

where the linear part P L and the nonlinear part P NL can be calculated by the

following equations:

f ³ c c c F H ) 1 ( ˜

r P

) , (

t L f

f f f ³³³ F H ) 3 ( ˜

r P

) , (

t NL f f f

2.3 Reflection and refraction

Reflection is an important physical effect for the propagation of light in an optical

fiber. The light, which is fed into a fiber, will follow the fiber with few losses as long as

the bend is not too sharp. The first condition for total internal reflection is that the

refractive index of the cladding is lower than that of the core. The amount of light that is

reflected depends on the nature of the surface of the interface. To minimize the loss, the

17

2.4 DISPERSION AND LOSS

surface has to be very regular and smooth. According to the law of reflection, the angle

of incidence equals the angle of reflection, and the incident ray, the reflected ray as well

as the normal ray always lie in the same plane.

But not all rays will be reflected. There may also be a ray passing from the higher

refractive index side to the lower refractive index side. In this case the refracted ray

increases its speed and changes its direction such that the angle between the ray and the

normal gets larger. When the ray comes from the other direction, passing from the

lower refractive index side to the higher one, the speed of the ray, as well as its angle to

the normal, will decrease.

This behaviour can be calculated using Snell’s law:

˜

sin

n

1

where n 1 and n 2 are the refractive indices of the two materials. The refractive index is

defined as the ratio of the speed of light in vacuum c 0 to the speed of light in the

medium v.

c

0 n (2.12)

v

At some point T 2 will reach 90°. Then T 1 will be called the critical angle, T crit .

If T 1 is increased any further, total internal reflection will occur as described above. In

this case T 2 equals 90°, and the critical angle can be calculated as follows:

n T

1

2 sin (2.13)

crit n

1

2.4 Dispersion and loss

Dispersion describes the spreading of a signal in time. At the input a series of pulses

– representing binary information – is launched into an optical waveguide. Dispersion

causes each of the pulses to spread in time. When they arrive at the output, the pulses

have broadened to the point where they begin to overlap adjacent pulses. This is called

intersymbol interference (ISI). The spreading limits the maximum data rate of a

communication link. There are three types of dispersion, which can occur in optical

waveguides.

18

CHAPTER 2 THEORY

Material dispersion is caused by the material itself, which has different refractive indices for different wavelengths. In other words, the refractive index n is a function of the frequency. Therefore each wavelength will travel at a different speed, and the signal pulses will smear out. This effect is even stronger in semiconductor materials.

Waveguide dispersion occurs due to different time delays or velocities at different wavelengths due to different power-distributions in the core and the cladding. It can be altered by careful design, and thereby different fiber types like DCFs and DSFs can be produced.

Modal dispersion only occurs in multimode fibers. Each allowed mode in the waveguide travels with a different group velocity. In step-index (SI) fibers pulse spreading occurs due to different geometrical paths. The paths taken by particular modes may be longer than the paths taken by other modes, and the pulse tends to spread out. This problem can be overcome by using graded-index (GI) fibers. In these fibers the refractive index is maximal in the centre of the core and gets smaller to the cladding. Thus the light will travel faster in the outer regions of the core and compensates the longer distance it has to travel compared to an axial ray directly through the centre of the core.

In reality fibers are not perfectly circular symmetric. Thus, practical fibers are slightly birefrigent and the two orthogonal polarized modes have slightly different propagation constants. Because the energy of a pulse will normally split between these two modes, the birefringence will lead to pulse spreading. This phenomenon is called polarization-mode dispersion (PMD). In principle, this effect is similar to pulse spreading in multimode fibers. A typical value for the time spread due to PMD, which is also called differential group delay (DGD), is 'W = 0.5 ps/km [3]. However, the propagation constants for each polarization mode vary over the length of the fiber, and the PMD effects are not as bad as indicated.

In single-mode fibers only material dispersion, waveguide dispersion and PMD exist. To avoid the effects of dispersion it is possible to use dispersion-shifted fibers (DSF). These fibers are designed in such a way that the waveguide dispersion is enhanced to compensate material dispersion. However, these fibers have a very high nonlinear constant and are thus very sensitive to high signal powers. To compensate the accumulated dispersion, also dispersion-compensation fibers (DCF) can be utilized. These fibers have a high positive dispersion because the waveguide disersion is much higher than the material dispersion, and thereby a short DCF can compensate the accumulated dispersion of an SSMF, which is approx. 20 times longer.

19