Distributed Optical Amplification Based on Raman Scattering for DWDM Applications in C, L and U Bands

6 
Introduction 
 
The dramatic growth of Internet has invited unprecedented rapid deployment of 
Wavelength Division Multiplexed (WDM) and DWDM (Dense WDM) transmission systems 
based on Doped Fiber Amplifiers (DFAs), which are a cutting-edge technology yet.  
The insatiable demand of Internet to high-capacity data transport has dragged a huge 
amount of new inventions directly out of labs into real fields, most of which may be maturing 
in the course of field applications. As a result, WDM transmissions are now using up the entire 
avalable gain band of DFAs (S, C and L bands). Also, the spectral efficiency of transmission 
capacity has been increased up to 0.4 b/s/Hz by adopting the so-called dispersion-managed 
solitons techniques. Even though the entire band of DFAs is fully utilized and very high 
spectral efficiency has been obtained through soliton technologies, the demands from Internet 
still seem to keep on growing. It has been pointed out that the DFAs based DWDM 
transmission technology has been hitting the upper limit of transmission capacity.  
DFAs used in DWDM transmission systems are so-called lumped amplifiers in which the 
gain is lumped at a point of the transmission line. On the other hand, distributed amplifiers, 
such as fiber Raman amplifiers, retain the optical signal level over a long distance along the 
transmission line.  
This distributed amplification, in principle, shows better system performances especially in 
terms of noise.  
 
 
 
 
 
 
 
 
 
 
Figure I.1 shows how distributed amplifiers are superior to lumped amplifiers. A focus of 
system design is to optimize the signal level diagram: there is an upper limit and a lower limit 
in terms of signal level. If the signal enters into the fiber span at a too high level, it suffers 
Figure I.1: Schematic diagram to compare a distributed amplifier and a lumped amplifier. 
Introduction 
 7
from fiber nonlinearities, while if it enters at a too low level, it receives high noise that will be 
amplified at the next stage. In the systems based on lumped amplifiers, it is more difficult to 
solve this dilemma respect to those based on distributed ones, as we can see in figure I.1. Now 
that DFA-based systems are facing against a wall that bars toward higher capacity, is time to 
reconsider fiber Raman amplifiers. Raman amplification in optical fiber was first observed and 
measured by Stolen and Ippen. Their measurement showed that the Stokes shift of Silica fiber 
is approximately 13.2 THz.  
Experiments on optical data transmission using Raman amplifiers were carried out by Aoki 
et al., and Mollenauer et al. utilizing a fiber Raman amplifier to carry out optical soliton 
transmission. Although it was even before EDFAs that Raman amplifiers were demonstrated 
in experiments using large-scale solid-state lasers, they were not actually deployed in the real 
field systems. Raman amplifiers could not have been deployed until now when high-power 
diode pump sources became commercially available.  
Meanwhile, DFAs using diode pump lasers have been widely studied and actually deployed 
in the DWDM systems. 
Despite the deployment of DFAs in the real-field applications, studies on Raman 
amplifiers have been continuously carried out to suggest two critical merits of Raman 
amplifier: one is low noise and the other is arbitrary gain band, so the system based only on 
Raman amplifiers has superior noise performance to an EDFA based system. Masuda et al. 
demonstrated that a Raman amplifier placed in front of an EDFA increases the overall gain 
bandwidth to 75 nm. Authors proposed Raman amplifiers pumped with the so-called WDM 
pumping technique, in which 1.4 micron pump laser diodes are combined through WDM 
couplers to generate a very high pump power over a wide wavelength range. The wide 
wavelength range of pump light results in a wide and flat composite gain spectrum of Raman 
amplifier.  
In conjunction with the Internet’s exploding demands, the maturity of the pump sources 
has invoked resurgence of Raman amplifiers in WDM transmission systems. As was discussed 
with respect to figure I.1, a distributed amplification improves Optical Signal-to-Noise Ratio 
(OSNR). In other words, a Raman amplifier increases power budget margin in system design. 
Then, the question is what this extra budget should be used for. One answer to this question is 
to enhance spectral efficiency and realize terabit-per-second transmission. Also, Raman 
amplifiers play an important role in transoceanic multi-terabit-per-second WDM transmission.  
An interesting use of the power budget margin is to extend the fiber span distance 
between amplifiers up to 250 km, 140 km and 160 km. Hansen et al. and Takachio et al. 
Introduction 
 8
demonstrated that four-wave mixing could be suppressed even in Dispersion Shifted Fibers 
(DSF) by using Raman amplifiers. Garrett et al. have demonstrated that Raman amplifiers 
operate nearly as well as those in the lab, even though the field-deployed fiber cables contained 
many connectors causing reflections.  
In this scenario, this thesis want to demonstrate the possibility to obtain Raman 
amplification in the C (1530-1565 nm), L (1565-1625 nm) and in the far U (1625-1675 nm) 
bands with a simple and novel approach based on standards optical components. 
The first chapter shows the actual state of the optical network with its systems and 
components. Furthermore the merits, the needs and the problems of the most used optical 
amplifiers are analized introducing the Raman amplifiers. 
The second chapter explains the Raman scattering and amplification underlining the theory 
and the fundamental applications. Moreover the aspects linked to the noise and to the 
polarization are also analized. 
In the third chapter will be presenteed the results of the experimentation activity made at 
the Optical Communications laboratories of ISCOM (Istituto Superiore delle Comunicazioni e 
delle tecnologie dell’informazione) research body of the Italian Communication Ministry. With 
a Raman pump in the E band we obtain gain in C, L and U bands. 
The system has been studied in terms of gain and noise figure and its performances have 
been tested on the deployed optical fiber between Rome and Pomezia, obtaining the Q-factor 
and the eye diagrams of the modulated signals.  
Considerations on signals input polarization are also reported, showing the absence of 
polarization dependence of the presented system. 
9 
Chapter 1 
The High Speed Optical Transport Network 
 
Preface 
 
The improvement and the growth of telecommunications network efficiency have shifted 
the focus on new strategical targets [1]. The main reasons of the frenetic change are: the shift of 
voice traffic from a fix access to a mobile access; the incoming of new Operators in the telecom 
market; the growth of IP based services (Internet Protocol); the integration, in the transport 
network, of voice, data and video. 
 
1.1 Growth and targets of OTN (Optical Transport Network) 
 
The creation of an open network able to integrate several network platforms based on the 
IP protocol is fundamental. In forecast of this new background we can see, already today in a 
few cases, the decrease of the hierarchical levels, the concentration in few network knots of 
complicated elaboration functions and the reduction of  management costs.  
Furthermore, the request for high bit-rate services has assumed remarkable 
proportions, simultaneously to a sudden decrease of  bit/km cost. All this has brought, and is 
bringing, to a modernization of the network based on the capillary diffusion, at the Access 
Level, of X-DSL technologies (ADSL-Asymmetric Digital Subscriber Line, VDSL-Very high 
bit rate DSL and SDSL-Symmetric DSL). At the Transport Level, there is a deepest interest 
focused on an intelligent exploitation of the optical backbone through the use of WDM 
(Wave Division Multiplexing) and DWDM (Dense WDM) systems. 
The strong interaction of heterogeneous technologies has, in this sense, excellently 
answered to the evolution of the global communication and so to its new necessities. 
 
 
 
28 
Chapter 2 
Raman Scattering and Amplification 
 
Preface 
 
As already analysed, internet traffic is increased at a rate faster than that would been 
expected on the basis of Moore’s law, generating an urgent need for development of 
technologies that can serve as an alternative to conventional WDM and DWDM transmission 
based on EDFAs [12]. The research is working to enhance DWDM technologies so that 
higher bit-rates per channel can be transmitted. It was inevitable that pursuit of such 
improvements in DWDM technologies would narrowed the margins of system design. For this 
reason, researchers have begun seeking alternative enabling technologies capable of avoiding 
system impairments such as noise and non-linear effects, and handling bandwidths wider than 
those that can be managed by use of EDFAs. One of these new technologies is the distributed 
fiber Raman amplification which is intrinsically characterized by low noise and by the ability to 
provide gain over a wide wavelength range. 
 
2.1 Research on Raman amplification 
 
Research on Raman amplification in optical fibers started early in the 1970s with the 
work “Raman gain in glass optical waveguides” written by Stolen and Ippen [13]. The benefits 
from Raman amplification in the transmission fiber were already being investigated in the 
middle 1980s principally by Mollenauer, Gordon and Islam with the work “Soliton 
propagation in long fibers with periodically compensated loss”. 
However, Raman gain requires more pump power, roughly tens of milliwatts per dB of 
gain, as compared to the tenths of a milliwatt per dB required by EDFAs for small signal 
powers. This disadvantage, combined with the scarcity of high power pumps at appropriate 
wavelengths, meant that Raman amplifier research subsided during the commercialization of 
EDFAs in the early 1990s. 
 51 
Chapter 3 
Cascaded Raman Amplification 
 
Preface 
 
The introducton of lumped optical amplifiers, as the EDFAs, to overcome the losses 
encountered by the WDM streams, opened the window to the implementation of long-haul 
large-capacity transmission systems [33]. The problems induced by this kind of amplification 
technique are typically related to noise accumulation and nonlinearity which result in signal 
distortion. 
On the other hand, the so-called distributed optical amplification, which can use 
ordinary transmission fibers as gain medium, is advantageously less noisy and also more 
suitable for reduced nonlinear effects. For these reasons, distributed Raman amplification is 
nowadays recognized as an enabling technology for high capacity systems. 
Furthermore, the increase in the bit-rate per channel and in the transmission distance, 
lead to the introduction of chromatic dispersion management based on the adoption of 
dispersion compensation units at the network nodes. 
This chapter is the result of the research activity in the FUB (Fondazione Ugo Bordoni) 
Optical Communications laboratories at ISCOM (Istituto Superiore delle Comunicazioni e 
delle tecnologie dell’informazione) research body of the Italian Communication Ministry. The 
research aim is the analisys and characterization of a novel distributed Raman amplification 
system. 
The designed novel approach is based on something modelled and characterized [34] but 
considered often only an impairment in literature [35]: the Raileigh backscattering. Controlling 
this process we can obtain a triple C, L and U bands amplification. 
 
80 
Appendix 
Optical Components and Fibers 
 
Preface 
 
In this section two important elements of the set-up are described: the multi wavelength 
Keopsys
®
 Raman fiber laser and the fiber Bragg gratings. Fiber connectors and some 
important operations on fibers are also showed.  
 
A.1 Keopsys
®
 Raman fiber laser 
 
This new generation of device offers many new features in a turnkey benchtop package 
(fig. A.1). 
The principle of operation involves utilization of a high power Ytterbium fiber laser 
source to pump a cascaded Raman resonator. 
This type of laser acts as a triple pump producing signals at 1428, 1445 and 1466 nm; 
their power can be regulated until a total value of 30 dBm. The power distribution changes 
from 63% to 100% for 1428 nm pump and from 0% to 100% for 1445 and 1466 nm pumps. 
 
 
 
Figure A.1: High power Keopsys
®
 Raman pump.