Transcript Wavelength switching

Introduction
Chapter 2
1
Outlines
• Telecommunication networks
• Optical communications
• WDM network testbed and product
comparison
2
2.1 Telecommunication Networks
3
Network design
• A telecom network is typically constructed level-bylevel, i.e. hierarchically.
• Traditionally, AT&T networks have used a 5-level
hierarchy, but the current telecom networks uses only
three levels:
– dynamic routing tandem/international switching centre
– toll centre
– end-office.
• There are three switching modes in telecom networks:
– circuit switching
– packet switching
– cell switching.
4
Switching mode
• Circuit-switched networks
– offer connection-orientated services, where connections are set up endto-end before information transfer and resources are reserved for the
whole duration of connection.
– delays occur before and after information transfer,
– but there is no extra delay and no overhead during information transfer.
•
Packet-switched networks
– offer connectionless services in which there is no connection setup and
no resource reservation.
– offers connection orientated services in which virtual connections are
set up end-to-end before information transfer but there is no resource
reservation.
– Information transfers as discrete packets each of which has a varying
length.
5
Switching mode
• connectionless packet switching
– requires that each packet carries its global address of destination;
– no delay before information transfer, but during information transfer, packets
have to carry their header (overhead), expect packet-processing delays, and
may suffer queuing delays (when packets compete for joint resources).
• connection-orientated packet switching
– uses only a local address, i.e. the logical channel index.
• In connectionless packet switched networks, there is Cell-switching
networks offer connection orientated services, where virtual connections
are set up end-to-end before information transfer, and where resource
reservation is possible but not mandatory.
• Information transfers as cells, each of which has a fixed length and uses a
local address.
• In cell-switched networks,
– no delay before information transfer, but during information transfer,
– each cell carries a header (overhead), expects packet-processing delays, and
– may suffer from queuing delays (if resources are not reserved beforehand).
6
Comparison
7
2.2 Optical Communications
• light sources, LED and ILD are semiconductor devices that
can generate beams of lights when a voltage is applied.
– the light-emitting diode (LED)
• The LED is comparatively cheaper and has a longer operational life.
• can operate in a greater range of temperatures.
– the injection laser diode (ILD).
• operate on the laser principle and potentially produce higher data rates due
to smaller range of light frequencies generated and less dispersion.
• A cornerstone of optical networking is the creation of a tuneable laser
diode operating around 1.55 mm.
• ‘Tuneable’ means the same physical device (i.e. the laser diode) can be
used to generate different wavelengths of light.
8
Receiver
• An optical receiver is a semiconductor device that
detects the light and then generates a flow of
electricity.
• Optical receivers can be described in terms of these
parameters:
– receiver efficiency (i.e. the ratio of output current power to
input optical power),
– the range of optical wavelengths (over which the receiver
operates),
– response time (how quickly the receiver can react to
changes in the input optical power), and
– the noise level of the receiver.
9
Wavelength
• Three optical bandwidth regions with low
attenuation have been selected in fibre
transmission. They are centred on 0.85, 1.3
and 1.55 mm. The first wavelength band
10
11
2.2.1 Optical Communication
Impairments
• Attenuation
– The phenomenon that signal power gradually reduces over distance as
the signal propagates is known as attenuation.
– To allow for attenuation, the signal must have sufficient strength for the
receiver to detect that signal and therefore we must maintain a signal
level sufficiently higher than any noise in the signal channel.
– The signal power can be improved using amplifiers, for example, in the
case oflong distance transmission.
– In addition, attenuation is an increasing function of frequency.
– As indicated in Figure 2.2, the lowest attenuation in fibre, 0.2 dB/km,
occurs at 1.55 mm. Beyond that point, as wavelengths increase, the
attenuation starts to pick up and then becomes high quickly.
Attenuation in fibre is logarithmic as expressed below:
12
Dispersion
•
•
Dispersion is the effect that different frequency components of the transmitted
signal travel at different velocities in the fibre, arriving at different times at the
receiver.
Types of dispersion
– multimode distortion, occurs in multimode fibres, in which the signal is spread in time
because the velocity of propagation of the optical signal is not the same for all modes.
– polarisation dispersion, describes the light wave orientation. The different polarisations
of light travelling at different speeds through optical fibre can cause polarisation
dispersion
– chromatic dispersion. specifies the wavelength dependence of the velocity of
propagation (of the optical signal) on the bulk material of which the fibre is made.
•
•
•
Multimode distortion Polarisation.
The amount of dispersion is wavelength dependent. Dispersion is a problem
because it results in inter-signal interference if fibre lengths (before amplifiers) are
too long.
One way to reduce dispersion is to increase the distance between the light pulses,
but this lowers the signalling rate and so the overall data rate.
13
Nonlinear effects
• When the optical power within an optical fibre is small such as
in low bit-rate systems, fibre can be regarded as a linear
medium, i.e. the loss and Refractive Index (RI) of the fibre are
independent of the signal power.
• RI is a property of the fibre core and determines how fast the
light travels in the fibre. As the advent of WDM systems and
the increasing demand of higher bit rates, the amount of
optical power within the fibre increases.
• When power levels reach a fairly high level in the system, the
impact of nonlinear effects arises since both the loss and RI
depend on the optical power of the signal in the fibre.
14
Nonlinear effects
• Nonlinear effects include Kerr and Scattering effects.
• Kerr effects refer to the relationship between the
refractive index and the light intensity of the signal.
This can result in:
– self-phase modulation, where a wavelength can spread out
into adjacent wavelengths;
– cross-phase modulation, where different wavelengths
spread out into each others; or
– four-wave mixing, where several wavelengths interact to
create a new wavelength.
15
Scattering effects
• refer to the signal loss and stimulation due to the
contact between light and fibre.
• Scattering effects include Raman Scattering and
Brillouin Scattering.
– Raman Scattering
re-emits a longer wavelength due to the loss of energy,
while the latter ccurs because of the generated acoustic
waves.
– Nonlinear effects generally are nondesirable since they
make it hard to read signals, and it is possible to
misinterpret signals at the receiver. However, Raman
Scattering boosts signal power so it is useful or
amplification.
16
Crosstalk
• Crosstalk represents the undesired coupling of a signal to
another optical signal.
• Crosstalk in fibre transmission is also known as optical
coupling.
• Interchannel crosstalk occurs when the two interfering optical
signals have different wavelengths.
• Intrachannel crosstalk can take place when two light sources
are transmitting using the same wavelength (or very close
wavelengths) and a small amount of light of the first signal
‘leaks’ to the second’s receiver.
• Crosstalk can also happen when there are multiple paths for an
optical signal. In that case, light is leaked into alternate path(s).
The branching signal also reaches the receiver and therefore
causes receiver confusion.
17
2.2.2 Optical Switching
• Switching domains: There are several types of optical
switching technology classified by switching domains.
• The classification is also related to the switching traffic
granularity (i.e. the size of the signal that a carrier needs to
switch):
– Fibre switching: switches an incoming fibre including all the
wavelength channels on it to an outgoing fibre.
– Wavelength band switching: switches a set of wavelengths on an
incoming fibre to an outgoing fibre.
– Wavelength switching: switches an individual wavelength to a
wavelength on an outgoing fibre.
– Subwavelength switching: in the case of aggregated traffic, switches
subwavelength payloads onto the outgoing fibre, e.g. TDM slots.
– Space switching: switches a signal from one input port to several
different (possibly all) output ports.
– Time switching: each input port is given a time slot for admitting a
signal. Time switching is used in conjunction with other switching
techniques.
18
Client interface vs. physical transport
interface
•
Client interfaces
– represent the boundary between the WDM network and external
networks.
– provides an interface from the WDM provider network to a client
network.
– may request specific client signal formats, e.g. SONET/SDH signals.
•
•
•
•
•
Client interfaces that are located at the WDM network edge-switch are directly
connected to a switching fabric. An add port can insert a client signal into the
WDM network through the switching fabric.
The client signal is switched to a corresponding multiplexer for wavelength
multiplexing and then amplified before being sent through the fibre to the receiver.
A drop port can ‘drop’ a wavelength channel from the WDM network.
The incoming wavelengths go through de-multiplexing, fabric switching, and then
the drop port.
Depending on the interface card supported at the drop port, the wavelength is
converted into a specific client signal format.
19
Wavelength continuity vs.
wavelength interchange
• In WDM optical networks, optical cross-connect (OXC) requires
wavelength switching, which connects a specific wavelength on an
incoming fibre to the same wavelength on one or more ongoing
fibres. Wavelength switching for an incoming signal depends not
only on the availability of the specific optical frequency on the
outgoing fibres but also on the capability of the fabric supporting
wavelength interchange.
• Therefore, wavelength continuity and interchange are the two
conditions to describe the wavelength at the fabric physical transport
interface.
• Wavelength continuity means lack of wavelength interchange, i.e.
the same wavelength or frequency is required in the signal for endto-end transmission. Supporting wavelength interchange not only
can reduce the network bandwidth waste but also helps with
contention for specific wavelengths.
20
1R, 2R, and 3R regeneration
• The degree of transparency depends on the type of signal
regeneration at the fabric.
• 1R
– simply just relays the signal by amplification without retiming and
reshaping.
• 2R
– amplifies and then reshapes the signal to remove noise and
dispersion without retiming during signal regeneration.
– offers transparency to bit rate but does not support different modulation
formats.
• 3R
– regenerates signals through amplification and reshaping and then
retiming by synchronising with the network clock.
– eliminates transparency to bit rates and framing formats completely
since the signal is reclocked.
– produces a cleaner signal at each regeneration node so 3R signals can
travel a relatively long distance ‘safely’.
– However, complex regeneration is expensive and time-consuming.
21
Arbitrary concatenation
• Concatenation and grooming are used for improving
bandwidth utilisation.
• A grooming switch has the ability to divide the signal
into smaller payload granularities and directs the
payloads to different ports.
• Arbitrary concatenation refers to the capability of
precise band- width provisioning that is not confined to
the standard SONET line rates.
• For example, a conventional OC-48 OADM only handles
OC-3 and OC-12, but with arbitrary concatenation, a
switch can combine, for example, exactly 17 STS-1s to
create an OC-17.
• Arbitrary concatenation is an important feature for a
WDM switch because it introduces flexibility and quality
of service to traffic with finer granularities.
22
wavelength and subwavelength
concatenation
23
ITU-T WDM system
24
2.2.3 Opaque vs. Transparent
Switching
• WDM switching functionality can be implemented in either the
electrical or the optical domain.
• electrical switching, known as opaque fabric
– the optical signal is terminated at the entry of the switching fabric,
– converted to electrical signals, and
– regenerated as an optical signal at the output of the fabric.
• All-optical switching, known as transparent fabric
– the original optical signal passes through the fabric without the electrical
and optical conversion.
• In a circuit-switched network, switching control is provisioned
through a shadowed data communication network.
• A transparent optical network refers to the capability of supporting
an end-to-end optical channel, upon which there is no O-E-O
conversion in the intermediate hops.
25
Comparison
• Opaque systems
– are attractive for applications where subwavelength
grooming and signal processing are required.
– The majority of vailable opaque switches support or
will support grooming at subwavelength granularity
and provide the ability to arbitrarily concatenate
lower-speed signals.
• Transparent systems
– are suitable for switching the entire wavelength and
groups of wavelengths at the network core.
26
Comparison
27
WDM Network Testbed and
Product Comparison
• Bellcore’s LAMBDANET and IBM’s Rainbow are the two early WDM
local area
• network testbeds.
• LAMBDANET
– Both TDM and WDM technology are used in the Bellcore’s
LAMBDANET
– each node has a fixed wavelength for each transmitter (using a
Distributed FeedBack laser, DFB) and an array of receivers (the size of
the array is equal to the number of nodes in the network). The incoming
wavelengths are separated using a grating demultiplexer.
– aims at simplicity and also supports multicasting.
– Each transmitter, in time slots, multiplexes the traffic destined to all
other nodes into a single wavelength.
– Each receiving node simultaneously receives all the traffic, buffers it,
and selects the traffic destined for it in the electronic domain.
– Two sets of experiments were performed, with 16 and 18 wavelengths,
running at 1.5 and 2.0 Gbps per wavelength, respectively.
28
IBM’s Rainbow network
• IBM’s Rainbow network
– originally designed to interconnect 32 IBM PS/2 workstations using 32 X
200 Mbps channels.
– was the first to demonstrate tunable components.
– The network uses fixed transmitters and tunable receivers.
– Each node has a DFB laser transmitter that is associated with a specific
wavelength channel.
– Before the transmitter sends data, the receiver needs to tune to the
transmitter wavelength.
– The synchronisation is completed in the connection setup process using
out-of-band signalling. If a receiver is idle, it will check all the
wavelength channels for a connection setup request.
– Once the receiver finds the channel, it sends an acknowledgement to
the corresponding transmitter.
– Thereafter, the channel is set up and ready for transmission.
29
All-Optical Networking (AON)
• The All-Optical Networking (AON) Consortium, formed by AT&T,
DEC, and MIT Lincoln Laboratory also under DARPA sponsorship.
• tried to develop architectures and technologies to exploit the unique
properties of fibre optics for advanced broadband networking
including both WDM and TDM.
• AON deployed a static wavelength routing testbed in the Boston
metropolitan area to demonstrate the feasibility and interaction of
architectures, optical technologies and applications.
• The testbed uses space switches to implement wavelength routers
and converters.
• It allows the establishment of semi-permanent physical circuits for
teleconferencing or other scheduled services.
• This is supported by DFB lasers with tuning times of tens of seconds.
30
Wdm NETWORK TESTBED
31
Testbed Comparison
32
Product Comparison
33