Dense wavelength division multiplexing....

Performance Analysis of DWDM Optical Transmission System
CVSR College of Engineering 1
1. INTRODUCTION
In this topic we will assess the performance of a dense wavelength division multiplexing
(DWDM) optical network. The DWDM technique multiplexes up to 132 wavelengths, or colors,
over a single mono modal fiber or single mode fiber allowing the coexistence of many logical
channels over the same physical medium; the DWDM frequency grid at 100-GHz channel
spacing is defined by the ITU-T G.694.1, which fixes the reference frequency at 193.1 THz.
Different colors, i.e., communication links, can be added (multiplexed) and dropped (de
multiplexed) from the fiber by means of passive optical components. The passive mux/demux
process, compared to active sectioning of data, guarantees independence from specific protocols
and data rates allowing all the allocated channels to have the same latency. These features have
great impact on the reliability and the simplicity of the network; moreover, it is possible to
exploit the huge fiber bandwidth transmitting at the same time many high-speed streams
This technology cheaply solves the problem of increasing communication channels
without deploying new cables or when the number of fibers in the cable is limited; this solution
makes it very attractive not only for telecommunication products, but also for the design of
real-time data acquisition systems when the required bandwidth is on the order of many
gigabits per second.
Since the introduction on the market of small form factor optical transceivers, which are
also interoperable between different manufacturers, the DWDM technique is receiving great
attention: The last generation of DWDM lasers exhibits lower power consumption than
previous devices, can be easily inter- faced with most of the high-speed electrical transceivers
available, reaches data rates up to 2.7 Gb/s, and, not least, is not extremely expensive anymore.
Thanks to the mentioned characteristics, submarine experiments started to rely on DWDM
solutions to fit as many channels as possible into the standard telecommunication electro optical
cables in order to decrease the overall costs of connecting the submerse apparatus. A smaller
number of fibers inside the cable requires connectors with less optical pins: This choice reduces

the costs of the cable and the connectors itself because standard devices can be bought and
increase overall system reliability. This technology has already been successfully adopted in
NEMO [1] and ANTARES [2], two high-energy physics submarine experiments, and is planned
to be used by the KM3NeT Consortium.
To test the behavior of the optical network, we used a board designed for the NEMO
experiment that hosts a DWDM link for data transmission, as explained in Section II; such a
link was designed as a pluggable module to permit its reuse in other projects. In Section III, the
experimental test-bench setup to assess the performance of the network is described. In Section
IV, the results of this characterization will be shown, and, finally, the conclusions are drawn in
Section V.

2. LITERATURE SURVEY
Before starting DWDM technique and its application to find BER and OSNR lets go with
the optical fiber and its transmission concept and various multiplexing techniques used in it.
2.1 Optical Fiber:
Optical fiber consists of thin glass fibers that can carry information at frequencies in the
visible light spectrum and beyond. The typical optical fiber consists of a very narrow strand of
glass called the core. Around the core is a concentric layer of glass called the cladding.
A typical core diameter is 62.5 microns .Typically cladding has a diameter of 125
microns. 100 microwatts power (roughly) a light emitting diode can couple into an optical fiber.
Coating the cladding is a protective coating consisting of plastic, it is called the Jacket.
The loss in signal power as light travels down the fiber is called attenuation an important
characteristic of fiber optics is refraction. Refraction is the characteristic of a material to either
pass or reflect light. When light passes through a medium, it ―bends‖ as it passes from one
medium to the other. An example of this is when we look into a pond of water If the angle of
incidence is small, the light rays are reflected and do not pass into the water.
If the angle of incident is great, light passes through the media but is bent or refracted.
Optical fibers work on the principle that the core refracts the light and the cladding reflects the
light. The core refracts the light and guides the light along its path. The cladding reflects any
light back into the core and stops light from escaping through it - it bounds the medium! fast data
transmission rate is an advantage to using fiber optics data transmission.

2.2 Introduction to Previous Multiplexing Techniques
2.2.1 Optical time division multiplexing (OTDM):
Electronic circuits meet practical limitations on their speed of operation at frequencies
around 10 GHz. Therefore, although more recently the feasibility of 10 G bits-1 direct intensity
modulation and transmission over some distances (100 km) has been demonstrated electronic
multiplexing at such speeds remains difficult and presents a restriction on the bandwidth
utilization of a single-mode fiber link. An alternative strategy for increasing the bit, rate of
digital optical fiber systems beyond the bandwidth capabilities of the drive electronics is known
as optical time division multiplexing (OTDM).
Advantages of TDM
1. The user gets full bandwidth of the channel in a particular time slot.
2. For bursty signals such as voice or speech TDM gives maximum utilization of the channel
3. most suitable technique for digital transmission.
Disadvantages of TDM
1. it is not much suitable for continues signals
2. extra guard time are necessary
3. synchronization is necessary.
2.2.2 Optical Frequency Division Multiplexing:
In telecommunications, frequency division multiplexing (FDM) is a technique by which
the total bandwidth available in a communication medium is divided into a series of non-
overlapping frequency sub-bands, each of which is used to carry a separate signal. This allows a
single transmission medium such as a cable or optical fiber to be shared by many signals. An

example of a system using FDM is cable television, in which many television channels are
carried simultaneously on a single cable. FDM is also used by telephone systems to transmit
multiple telephone calls through high capacity trunk lines, communications satellites to transmit
multiple channels of data on uplink and downlink radio beams, and broadband DSL modems to
transmit large amounts of computer data through twisted pair telephone lines, among many other
uses.
FDM has both advantages and disadvantages relative to TDM. The main advantage is
that unlike TDM, FDM is not sensitive to propagation delays. Channel equalization techniques
needed for FDM systems are therefore not as complex as those for TDM[8] systems.
Disadvantages of FDM include the need for band pass filters, which are relatively expensive and
complicated to construct and design (remember that these filters are usually used in the
transmitters as well as the receivers). TDM, on the other hand, uses relatively simple and less
costly digital logic circuits. Another disadvantage of FDM[8] is that in many practical
communication systems, the power amplifier in the transmitter has nonlinear characteristics
(linear amplifiers are more complex to build), and nonlinear amplification leads to the creation of
out-of-band spectral components that may interfere with other FDM channels. Thus, it is
necessary to use more complex linear amplifiers in FDM systems.
In many communication systems, a single, large frequency band is assigned to the system
and is shared among a group of users. Examples of this type of system include:
1. A microwave transmission line connecting two sites over a long distance. Each site has a
number of sources generating independent data streams that are transmitted
simultaneously over the microwave link.
2. AM or FM radio broadcast bands, which are divided among many channels or stations.
The stations are selected with the radio dial by tuning a variable-frequency filter.
3. A satellite system providing communication between a large number of ground stations
that are separated geographically but that need to communicate at the same time. The total
bandwidth assigned to the satellite system must be divided among the ground stations.
4. A cellular radio system that operates in full-duplex mode over a given frequency band.
The earlier cellular telephone systems, for example AMPS, used analog communication

methods. The bandwidth for these systems was divided into a large number of channels.
Each pair of channels was assigned to two communicating end-users for full-duplex
communications.
2.2.3 Sub Career Multiplexing:
The utilization of substantially higher frequency microwave subcarriers multiplexed in
the' frequency domain before being applied to intensity modulate a high speed injection laser
source has generated significant interest. Such microwave subcarrier Multiplexing (SCM)
enables multiple broadband signals to be transmitted over single-mode fiber and appears
particularly attractive for video distribution systems. In addition, with SCM[8] conventional
microwave techniques can be employed to subdivide the available intensity modulation,
bandwidth in a convenient way. The result is a useful multiplexing technique which does not
require sophisticated optics or source wavelength. Either digital or analog modulation of the
subcarriers can be utilized by up converting to a narrowband channel at high frequency
employing either, amplitude, frequency or phase shift keying (i.e. ASK, FSK or PSK), and either
amplitude, frequency or phase modulation (i.e. AM, FM or PM) respectively: For digital signals,
FSK has the advantage of being simple to implement, both at the modulator and demodulator,
whereas for analog video signals the modulation of the high frequency carrier (up conversion) is
often carried out using either AM-VSB (vestigial sideband) or FM techniques. In both cases, the
multicarrier signal is formed by frequency division multiplexing (FDM) of the modulated
microwave subcarriers in the electrical domain prior to conversion to an 'intensity modulated
optical signal.
2.2.4 Wave Length Division Multiplexing:
The concept was first published in 1970, and by 1978 WDM systems were being realized
in the laboratory. The first WDM systems combined only two signals. Modern systems can
handle up to 160 signals and can thus expand a basic 10 G bit/s system over a single fiber pair to
over 1.6 T bit/s.

WDM systems are popular with telecommunications companies because they allow them
to expand the capacity of the network without laying more fiber. By using WDM and optical
amplifiers, they can accommodate several generations of technology development in their optical
infrastructure without having to overhaul the backbone network. Capacity of a given link can be
expanded simply by upgrades to the multiplexers and de multiplexers at each end.
This is often done by use of optical-to-electrical-to-optical (O/E/O) translation at the very
edge of the transport network, thus permitting interoperation with existing equipment with
optical interfaces.
Most WDM systems operate on single-mode fiber optical cables, which have a core
diameter of 9 µm. Certain forms of WDM can also be used in multi-mode fiber cables (also
known as premises cables) which have core diameters of 50 or 62.5 µm.
Early WDM systems were expensive and complicated to run. However, recent
standardization and better understanding of the dynamics of WDM systems have made WDM
less expensive to deploy.
Optical receivers, in contrast to laser sources, tend to be wideband devices. Therefore the de
multiplexer must provide the wavelength selectivity of the receiver in the WDM system..
WDM technique is again divided into two types
They are:
1. Coarse Wave Length Division Multiplexing (CWDM)
2. Dense Wave Length Division Multiplexing (DWDM)
conventional/coarse (CWDM) and dense (DWDM). Conventional WDM systems
provide up to 8 channels in the 3rd
transmission window (C-Band) of silica fibers around
1550 nm. Dense wavelength division multiplexing (DWDM) uses the same transmission window
but with denser channel spacing. Channel plans vary, but a typical system would use 40 channels
at 100 GHz spacing or 80 channels with 50 GHz spacing. Some technologies are capable of
12.5 GHz spacing (sometimes called ultra dense WDM). Such spacings are today only achieved
by free-space optics technology.

2.2.5 Coarse Wave Length Division Multiplexing (CWDM):
Coarse wavelength division multiplexing (CWDM) is a method of combining
multiple signals on laser beams at various wavelengths for transmission along fiber optic cables,
such that the number of channels is fewer than in dense wavelength division multiplexing
(DWDM) but more than in standard wavelength division multiplexing (WDM).
CWDM systems have channels at wavelengths spaced 20 nanometers (nm) apart,
compared with 0.4 nm spacing for DWDM. This allows the use of low-cost, un cooled lasers for
CWDM. In a typical CWDM system, laser emissions occur on eight channels at eight defined
wavelengths: 1610 nm, 1590 nm, 1570 nm, 1550 nm, 1530 nm, 1510 nm, 1490 nm, and 1470
nm. But up to 18 different channels are allowed, with wavelengths ranging down to 1270 nm.
The energy from the lasers in a CWDM system is spread out over a larger range of
wavelengths than is the energy from the lasers in a DWDM system. The tolerance (extent of
wavelength imprecision or variability) in a CWDM laser is up to ± 3 nm, whereas in a DWDM
laser the tolerance is much tighter. Because of the use of lasers with lower precision, a CWDM
system is less expensive and consumes less power than a DWDM system. However, the
maximum realizable distance between nodes is smaller with CWDM.
WDM, DWDM and CWDM are based on the same concept of using multiple
wavelengths of light on a single fiber, but differ in the spacing of the wavelengths, number of
channels, and the ability to amplify the multiplexed signals in the optical space. EDFA provide
an efficient wideband amplification which is used in DWDM For CWDM wideband optical
amplification is not available, limiting the optical spans to several tens of kilo metres.
2.2.6 Dense Wave Length Division Multiplexing (DWDM):
Dense Wavelength Division Multiplexing (DWDM) is a technology that allows multiple
information streams to be transmitted simultaneously over a single fiber at data rates as high as
the fiber plant will allow (e.g. 2.4 G b/s). The DWDM approach multiplies the simple 2.4 G b/s

system by up to 16 times, giving an immense and immediate increase in capacity—using
embedded fiber! A sixteen channel system (which is available today) supports 40 G b/s in each
direction over a fiber pair, while a 40 channel system under development will support 100 G b/s,
the equivalent of ten STM–64/OC–192 transmitters.
Dense wavelength division multiplexing (DWDM) refers originally to optical signals
multiplexed within the 1550 nm band so as to leverage the capabilities (and cost) of erbium
doped fiber amplifiers (EDFAs), which are effective for wavelengths between approximately
1525–1565 nm (C band), or 1570–1610 nm (L band). EDFAs were originally developed to
replace SONET/SDH optical-electrical-optical (OEO) regenerators, which they have made
practically obsolete. EDFAs can amplify any optical signal in their operating range, regardless of
the modulated bit rate. In terms of multi-wavelength signals, so long as the EDFA has enough
pump energy available to it, it can amplify as many optical signals as can be multiplexed into its
amplification band EDFAs therefore allow a single-channel optical link to be upgraded in bit rate
by replacing only equipment at the ends of the link, while retaining the existing EDFA or series
of EDFAs through a long haul route. Furthermore, single-wavelength links using EDFAs can
similarly be upgraded to WDM links at reasonable cost.

3. Performance Analysis of DWDM
Here we are analyzing the performance of DWDM technique using its one of the
application named as NEMO under water neutrino telescope, other than NEMO two more
experiments under water neutrino telescope experiments are used and named as ANTARES[2]
and KM3NET[3].
Neutrinos are omnipresent in nature such that in just one second, tens of billions of them
"pass through every square centimetre of our bodies without us ever noticing. Despite this, they
are extremely "difficult to detect" and may originate from events in the universe such as
"colliding black holes, gamma ray bursts from exploding stars, and violent events at the cores of
distant galaxies," according to some speculation by scientists. There are three types of neutrinos
or what scientists term "flavors": electron, muon and tau neutrinos, which are named after the
type of particle that arises after neutrino collisions; as neutrinos propagate through space, the
neutrinos "oscillate between the three available flavours. Neutrinos only have a "smidgen of
weight" according to the laws of physics, perhaps less than a "millionth as much as an
electron. Neutrinos can interact via the neutral current or charged current weak interactions.
In a neutral current interaction, the neutrino leaves the detector after having transferred
some of its energy and momentum to a target particle. If the target particle is charged and
sufficiently light (e.g. an electron), it may be accelerated to a relativistic speed and consequently
emit Cherenkov radiation, which can be observed directly. All three neutrino flavors can
participate regardless of the neutrino energy. However, no neutrino flavor information is left
behind.
In a charged current interaction, the neutrino transforms into its partner lepton (electron,
muon, or tau). However, if the neutrino does not have sufficient energy to create its heavier
partner's mass, the charged current interaction is unavailable to it. Solar and reactor neutrinos
have enough energy to create electrons. Most accelerator-based neutrino beams can also create
muons, and a few can create taus. A detector which can distinguish among these leptons can
reveal the flavor of the incident neutrino in a charged current interaction. Because the interaction

involves the exchange of a charged boson, the target particle also changes character (e.g.,
neutron → proton).
KM3NeT, an acronym for Cubic Kilo meter Neutrino Telescope, is a future
European research infrastructure which will be located at the bottom of the Mediterranean Sea. It
will host the next generation neutrino telescope in the form of a water Cherenkov detector with
an instrumented volume of more than five cubic kilo meter distributed over three locations in the
Mediterranean KM3NeT will search for neutrinos from distant astrophysical sources
like supernova remnants, gamma-ray bursts, supernovae or colliding stars and will be a powerful
tool in the search for dark matter in the universe. Its prime objective is to detect neutrinos from
sources in the Galactic Centre. An array of tens of thousands of optical sensors will detect the
faint light in the deep sea from charged particles originating from collisions of the neutrinos and
the water. The facility will also house instrumentation for other sciences like marine
biology, oceanography and geophysics for long term and on-line monitoring of the deep sea
environment and the sea bottom at depth of several kilo metres.
In 2010, a design study of the infrastructure was concluded with a Technical Design
Report, which was made public in 2011. It is anticipated that the neutrino telescope will contain
in the order of 12000 pressure-resistant glass spheres attached to about 600 detection units -
vertical structures with a height of almost one kilometer. Each glass sphere will contain
31 photomultiplier tubes and will be connected to shore via a high-bandwidth optical network.
At the shore, a farm of computers will perform the first data filter in the search for the signal of
cosmic neutrinos.
For the design of the neutrino detector, KM3NeT builds on the experience of three pilot
projects in the Mediterranean Sea: the ANTARES detector, the NEMO experiment and the
NESTOR Project. ANTARES was completed on 30 May 2008, and is the largest neutrino
telescope in the northern hemisphere. In the southern hemisphere, at Antarctica, the Ice Cube
Neutrino Observatory has recently been completed. Together, Ice Cube and KM3NeT will view
the full sky and form a global neutrino observatory.

ANTARES is the name of a neutrino detector residing 2.5 km under the Mediterranean
Sea off the coast of Toulon, France. It is designed to be used as a directional Neutrino
Telescope to locate and observe neutrino flux from cosmic origins in the direction of
the Southern Hemisphere of the Earth, a complement to the southern hemisphere neutrino
detector Ice Cube that detects neutrinos from the North. The name comes from Astronomy with
a Neutrino Telescope and Abyss environmental research project; the acronym also being the
name of the prominent star Antares. And nemo is the neutrino detector used as the same as
anatares and km3net. A DWDM link or module is also used in it we assess the performance of it
using BER and OSNR . we use a board designed for the nemo experiment is FCM.
To test the behavior of the optical network, we used a board designed for the NEMO
experiment that hosts a DWDM link for data transmission, as explained in Section 3.1; such a
link was designed as a pluggable module to permit its reuse in other projects. In Section 3.2, the
experimental test-bench setup to assess the performance of the network is described. In Section
4, the results of this characterization will be shown, and, finally, the conclusions are drawn in
Section 5.
3.1 FLOOR CONTROL MODULE BOARD
The principal indicator applied to measure the performance of the DWDM network,
whose layout is being explained in Section III, is the bit error ratio (BER), i.e., the ratio between
the number of wrong bits received to the total number of transmitted bits. To transmit and
receive bits over the optical channels under test and calculate the BER, the board developed for
the NEMO experiment, the so-called Floor Control Module (FCM), has been used. The FCM
hosts a plug-in module which is based on a transceiver chip-set and a DWDM laser. Fig. 1 shows
the FCM board hosting the DWDM link board.

Fig.3.1. Picture of the FCM board with the DWDM link plugged-in. The link interface and the test-bench
controls are implemented by a software program running on an embedded microcontroller synthesized in the
programmable logic.
In order to make the design of the DWDM e/o transceiver independent from a specific
optical component, the optical layer complies with the DWDM Small Form Pluggable (SFP)
Multi Source Agreement (MSA) [4]. This standard specifies the transceiver’s package outlines,
pin function definitions, and optical and electrical interfaces and characteristics. The module
form factor is much smaller than previous MSA versions permitting the mating with a 20-pin
connector soldered on the printed circuit board. This connector is designed so that the device is
hot-swappable, allowing an easy replacement in case of fault or if a different wavelength is
required. The control Logic is LVTTL compatible, and the high-speed interface is in ac-coupled
differential Current Mode Logic.
The transceiver is monitored by an internal microcontroller that communicates through
an I2C port. We have successfully tested two SFP MSA-compliant transceivers from Sumitomo,
the SCP9EA8-GL-CN-F330 and F440. The transceiver has multi rate capability starting from
622.08 Mb/s up to 2.7 Gb/s, has a maximum output power of 4 dBm, and a typical reach of 100
km; the input dynamic range at full data rate comprises between 9 and 28 dBm.

The link design is based on a pair of transmitter and receiver devices produced by PMC Sierra,
the HDMP1032/1034 G-Link chip-set, which provides a fixed and deterministic latency
transmission protocol, i.e., the clock recovered by the receiver is phase-locked to the transmitter
clock, with data rates that range from 400 Mb/s to 1.4 Gb/s. The user bus consists of a 16-bit data
word plus some control signals. The G-Link protocol adds four control flags, called the C-Field,
which add many features: dc-balance control through word inversion; possibility of sending
control, data, or idle words; insertion of the Conditional-Invert Master Transition (CIMT) in
each transmitted word; error checking. For a detailed description of the complete set of features,
see. The total amount of data transmitted is 20-bit multiplied by the clock frequency of the
parallel bus, which ranges from 20 to 70 MHz. During all the tests, we fixed this frequency at 40
MHz, which yields 800 Mb/s of total bit rate, as it is in the NEMO experiment. The receiver in
turn extracts the 800-MHz serial clock from the bit stream with a Clock and Data Recovery
(CDR) module, which divides the line clock down to 40 MHz and filter sit with a phase locked
loop (PLL), providing both the parallel word data and the control flags extracted from the C-field
Fig. 3.2. Diagram of the FCM blocks, including the DWDM link and the FPGA functionalities.
Fig.3.2 shows the block diagram of the DWDM link module and of the functions
implemented inside the field programmable gate array (FPGA). In this test, the 40-MHz parallel
clock is obtained by multiplying a local 4-MHz clock with an on-board PLL. The FPGA
implements a full-duplex bit error ratio tester (BERT) consisting of a transmitter and a receiver
section. After an initial delay that allows the receiver to start up, the BertTx module starts

sending to the serialize parallel 16-bit words, which are read from a read-only memory (ROM)
made up with FPGA’s internal RAM. The ROM is preloaded with random words to minimize
systematic errors in bit recovery.
As soon as the transmission starts, the BERT receiver module starts receiving data and
control flags decoded by the deserializer and compare the received words to the same content of
the ROM used by the transmitter. After the receiving process is started by reception of the first
valid word, then it is not stopped anymore assuming that for each clock transition a word has
been transmitted and must be received even in case of protocol errors: This procedure prevents
from losing the lock between BERT transmitter and receiver. An error counter is assigned to
each bit of the 20-bit line pattern and is incremented in case of mismatch.
A Xilinx Micro Blaze embedded processor has been instantiated in the FPGA to manage
the test-bench. The processor initializes the laser and the transmission with the correct startup
sequence, monitors transceiver functionalities, starts the BERT transmitter and a receiver
module, reads back the results from the BERT receiver, resets the bit error counters, and shows
errors number after a predefined time interval. Interaction with the user is accomplished through
an asynchronous serial port, which sends the system status and accepts commands from a
terminal running on a host machine.
3.2 DWDM LINK TEST BENCH
In this section, the test bench for evaluating the performance of the optical network is
described. The proposed setup closely reflects the network layout used in the NEMO experiment,
as explained in detail in [6]. For the sake of simplicity, the test bench network has less nodes
than those needed in the NEMO architecture, but this choice does not affect performance as soon
as adjacent channels move away from the one under test.

Fig. 3.3 Block diagram of the test-bench.
The NEMO experiment will require more than 40 channels per fiber, and the data rate per
channel will not exceed 800 Mb/s. The aim of the DWDM optical data transport network is pro-
viding many point-to-point logical channels sharing the same physical medium. The DWDM
network will introduce optical losses of about 30 dB. A distributed Raman amplifier is in-
troduced in the network in order to compensate for the insertion loss of the DWDM filters and
the 100-km fiber, thus improving the system margin. Each transceiver transmits a serial data
stream that modulates an optical carrier at 800 Mb/s. A bidirectional DWDM is achieved
allocating the channels on the 100-GHz ITU standard frequency grid over two separated bands.
Separation mux and de mux of the different directions is realized by an optical wideband
circulator. Single-channel mux and de mux is realized by Add/Drop passive optical filters.
The Add/Drop components are realized as thin-film optical filters having a bandwidth of
25 GHz. The Add/Drop filters exhibit a maximum insertion loss for the dropped or added channel
of 1.1 and 0.4 dB for the pass-through channels. The circulator component is polarization-
insensitive and exhibits a maximum insertion loss of 0.6 dB. The pictorial view of the
experimental setup is shown in Fig. 3. The measurements have been taken in the network
configuration shown in Fig. 4 for three DWDM channels at 194.3, 194.4, and 194.5 THz.
The bit stream generated by the first FCM is transmitted through fiber. The DWDM
aggregate of three adjacent channels is realized feeding the output of three FCM transceivers
into the optical network by a chain of three Add filters. The DWDM aggregate is later passed

through the Raman amplifier module (the IPG Fiber tech RLT-1-1427-1-1455-FW) and fed
into the 100-km fiber trunk, which has been chosen to comply with the ITU G.655 standard for
single-mode nonzero dispersion shifted fiber. At the fiber trunk output, the aggregate is de
multiplexed by a chain of three Drop filters. The central channel is passed through the Agilent
8164B Light wave Measurement System, which acts as programmable optical attenuator, before
it is fed into the receiver. The attenuation can be set very precisely, and the output power is
measured by the instrument itself. The attenuator has been set to give a fixed output power
during the measurement time window in order to avoid the instability of the power level fed into
the FCM transceiver. The instrument can be tuned at the wavelength of the DWDM transmitter
under test.
The optical path is split at the attenuator output with a 50/50 ratio in order to measure it
with the OSA, the EXFO FTB-5240. Data stream is sent and received from the same FCM,
which extracts clock and data from the bit stream. The received data is used by the BERT
receiver module inside the FPGA to measure the BER. Both the operations of the FCM board
and the measurements results are supervised by a host PC that communicates with the
embedded processor through the RS232 connection. The optical power at the variable attenuator
output has been regulated, and once the receiver loses the link [asserting the Loss Of Signal
(LOS) flag, the BER has been measured, increasing the optical power level fed to the receiver
input.

4. TEST RESULTS
The measurements described in this section aim at evaluating how the BER increases
when the optical signal is attenuated, worsening the signal-to-noise ratio (SNR). The minimum
optical power allowed at the receiver section of the 194.4-THz transceiver is 37.2 dBm: Below
this value, the receiver loses the link and asserts the LOS flag. To regain the lock, the optical
power must be greater than 34.5 dBm: This behavior implies an hysteresis mechanism
implemented by the e/o transceiver. The output power of the laser has been measured by the
Agilent 8164B and is 1.05 dBm, i.e., about 1.3 mW. Hence, the total link budget is about 38.2
dB, largely exceeding what is stated in the laser data sheet. The reason for this behavior is that
the link budget is measured at the highest rate, i.e., 2.7 Gb/s; roughly, when the rate is doubled,
the budget decreases of 3 dB, hence transmitting at 800 Mb/s will yield 6 dB.
4.1. DWDM Optical Spectrum
Spectral measurements have been taken during the BER analysis at 194.4 THz. The
calculation of the optical signal-to-noise ratio (OSNR) for the DWDM channel has been done
measuring the Drop filter in-band power and the noise floor. The measured noise floor is 74.93
dBm. The spectrum of the 3 DWDM channels at the circulator output is shown in Fig. 5. The
power
Fi.4.1. Three-channels WDM spectrum with 100-GHz spacing.

levels shown in BER measurements figures: This difference is most probably due to a defect of
the input connector of the OSA that could not be fixed during the measurements. The spectrum
shows no trace of nonlinear interactions between the three channels: This is mainly due to the low
output power level at the transmitter and the wide frequency span between channels compared to
their modulation bandwidth.
The spectrum of the 194.4-THz channel at the receiver input is shown in Fig. 6. The
spectrum shows the very good behavior of the Drop filter, which acts almost as a perfect pass
band filter.
4.2 BER vs Optical Power at the Receiver
As stated in Section II, the main parameter used to evaluate the performance of the data
transmission system is the BER. Along with the BER, the confidence level must also be indicated
in order to assess up to which degree the result can be considered significant. For the calculation
of the confidence level, we will use the following classical formula, which is based on the
binomial distribution of errors.

Fig. 4.2. Measured BER as a function of optical power at the receiver at two different DWDM wavelengths.
Fig. 4.3. Measured BER as a function of optical SNR at two different DWDM wavelengths.
Fig. 4.2 shows the BER measured as a function of the received optical power using the
pseudorandom payload. Measures have been repeated for two transceivers working at two
different frequencies, 194.4 and 193.3 THz, coded as F440 and F330. The test setup for the
193.3-THz frequency is the same as the one shown in Fig. 4, where the adjacent channels are
193.2 and 193.4 THz. The BER remains below 10 for a received power level higher than 36.6
dBm for both frequencies. When the optical power approaches the LOS state, the BER reaches
for F440 transceiver and for F330. The errors were equally distributed between the
20 bits of the CIMT pattern.
Fig. 4.3 shows the measured BER as a function of the measured OSNR, for the same
data set of Fig. 4.1 An instability in the BER curves is observed for both transceivers as the
power level approaches the LOS state; this behavior could be due to an instability of the output

power of the Raman pumps affecting the overall amplifier gain. These values have been
measured more than once to show that they are not due to measurement errors, even though this
can be considered just a qualitative statement without a deep statistical analysis.
The performance gets worse with respect to the system without amplification, and
this is reasonable because of the higher noise level introduced by the Raman amplifier into the
receiver bandwidth.
The negligible effect of the interfering signals was expected: The 100-GHz spacing
between adjacent wavelengths and the small modulated bandwidth make the channels
completely independent from each other as shown in Figs. 5 and 6.
Fig. 4.4 Comparison between calculated BER and measured BER as a function of optical SNR (linear) for
F440 transceiver.
In order to quantify the noise contribution due to the Raman amplifier, the OSNR
measurements should take into account the in-band noise level. The add/drop filters in the
network sup- press the noise between optical channels. The OSA offers two methods to estimate
the level of the noise present at the channel wavelength: the linear interpolation and the in-band
methods. The former uses interpolation of the noise level measured on both sides of the signal
peak to estimate the noise. The latter uses a series of scans having different polarization states to
estimate the noise level at the wavelength of the channel; during the series of scans, the fiber is

coiled twice to for a 3.5-cm-wide loop that is rotated about 30 for each scan. Using the linear in-
terpolation method, the noise level is underestimated, resulting in a misinterpretation of the
OSNR values. In-band method offers a better noise level estimate.
The measured in-band noise level for the F440 transceiver is 48.38 dB, thus leading to
mean OSNR of 11.50 dB; the measurement is done by the OSA over a 12.5-GHz band around the
channel frequency. In order to calculate the BER from the measured OSNR, we have the following
empirical approximation of the classical formula found in literature.
(2)
For a BER Of , an OSNR of 14.39 dB is required at the input of the ideal receiver. For a
real receiver, a correction factor should be introduced to take into account the ratio between the
3-dB bandwidth of the optical filter at the receiver input and the bit rate of the signal. We consider
a 12.5-GHz optical bandwidth and 0.8-GHz modulated bandwidth for the NRZ pseudorandom
signal; the calculated correction factor is
(3)
The OSNR penalty due to the nonideal receiver is 2.45 dB, and the resulting corrected
OSNR is 13.95 dB. Introducing the OSNR corrected values, the BER values are calculated using
the simplified formula.
Fig. 4.3 shows the calculated and the measured BER, with statistical error bars, as a
function of the measured OSNR including the penalty induced by the non ideal transceiver F440.
For error bars calculation, a binomial error distribution has been assumed. The calculated BER
remains below for a received OSNR higher than 13.88 dB. F330 transceiver shows a
behavior analogous to F440.

5. CONCLUSION
In this paper, we evaluated the performance of a DWDM optical network measuring the
BER and the OSNR in the final network configuration with three DWDM channels on the same
medium. The measured BER is in the worst case for values of optical power at the
receiver close to the LOS state. As a consequence of the introduction of the Raman amplifier,
the OSNR does not drop as the signals travel along the optical path as it would have in a system
where a discrete amplifier is placed at the receiver input. Moreover the introduction of optical
filters increases the system performance due to the reduced amount of optical noise fed to the
receiver.
The results show a high optical power budget around 38 dB m. This optical power budget
allows to operate the DWDM network with a system margin up to 8 dB. The OSNR measured
with the linear interpolation method by the OSA is an over estimated value. The in-band noise
measurement method should be introduced for a correct OSNR measurement, and corrections
should be done to account for the non ideal behavior of the receiver. A good agreement between
calculated and measured BER has been found adopting the approximated formula.
The system designer can rely on the in-band OSNR measurements performed by the OSA
to foresee the performance of the system and optimize the DWDM network architecture in order
to maintain the required OSNR to meet the target system BER. The next steps will be the BER
measure for more wavelengths and the addition of the counter propagating wavelengths on the
same optical network. The oscillating trend of BER points close to the LOS state should also be
confirmed, enriching the data set. The behavior of the system for higher data rates must be
also evaluated to verify the receiver input range at full speed.

REFERENCES
[1] E. Migneco, S. Aiello, M. Ambriola, F. Ameli, I. Amore, M. Anghi- nolfi, and A.
Anzalone, ―Status of NEMO,‖ VLVvT2, vol. 567, no. 2, pp. 444–451, Nov. 2006, NIM-
A. Elsevier.
[2] Antares Collaboration, ―The data acquisition system for the ANTARES neutrino
telescope,‖ NIM-A, vol. 570, no. 1, pp. 107–116, Nov. 2007.
[3] KM3NeT, ―Km3net Web site,‖ 2011 [Online]. Available: http://www. km3Net.org/
[4] Agilent Technologies, Blaze Network Products, E2O Communications, Inc.,
ExceLight Communications, Finisar Corporation, Fujikara Technology America Corp.,
Hitachi Cable, Infineon Technologies Corp.,IBM Corp., Lucent Technologies, Molex,
Inc., Optical Communication Products, Inc., Picolight, Inc., Stratos Light- wave, and
Tyco Electronics, ―Small form factor pluggable (SFP) transceiver multisource
agreement (MSA). Cooperation agreement for small-form factor pluggable
transceivers,‖ [Online]. Available: http://www.schelto.com/SFP/SFP%20MSA.pdf
[5] PMC-Sierra, Inc., PMC-Sierra, HDMP-1032AG/1034AG transmitter/ receiver chip
set Data Sheet 1, Nov. 2006.
[6] A. D’Amico, NEMO Collaboration, ―Design of the optical Raman amplifier for
the shore station of NEMO phase 2,‖ Nucl. Instrum. Methods Phys. Res. Sec. A,
Accelerators, Spectrometers, DetectorsAssoc, Equip,, vol. 626–627, no. Supplement 1,
pp. S173–S175, 2011.
[7] ―Statistical confidence levels for estimating error probability,‖ Maxim Eng. J., vol.
37, p. 4, 2007.
[8] G. P. Agrawal, Fiber-Optic Communications Systems. New York: Wiley, 2002.

Dense wavelength division multiplexing....

Recomendados

Recomendados

Más contenido relacionado

La actualidad más candente

La actualidad más candente (20)

Destacado

Destacado (20)

Similar a Dense wavelength division multiplexing....

Similar a Dense wavelength division multiplexing.... (20)

Último

Último (20)

Dense wavelength division multiplexing....