Transcript Chapter 1

Chapter 1
Optical Networking:
Principles and Challenges
1
Course Information





Email: [email protected] or
[email protected]
研究室：機械大一樓 104室
電話：7232105-7047
Course Web Site: http://120.107.172.220/
http://deron.csie.ncue.edu.tw
2016/4/2
[email protected]
2
教師時間
Meeting
Meeting
2016/4/2
[email protected]
3
評分標準




期中考 30%
期末考 30%
論文報告2篇 20%
程式設計1題 20%
4
Outlines















1.1 Introduction
1.2 Telecom Network Overview
1.3 Telecom Business Models
1.4 Roles of Three Fields in Optical Networking
1.5 Cross-Layer Design
1.6 TE vs . NE . vs . NP
1.7 What is an Optical Network?
1.8 Optical Networking: Need + Promise = Challenge!
1.9 xDM vs . xDMA
1.10 Wavelength-Division Multiplexing (WDM)
1.11 WDM Networking Evolution
1.12 WDM Network Constructions
1.13 WDM Economics
1.14 Sample Research Problems
1.15 Road Map - Organization of the Book
5
1.1 Introduction


We need to be ready with the appropriate
technologies and engineering solutions to
meet the growing bandwidth needs of our
information society.
Optical networking using wavelengthdivision multiplexing (WDM)
6
1.2 Telecom Network Overview
Access network
Metropolitan-area
network

Backbone network
backbone
regional
Local
7
The access network






Enable the end-users to get connected to the rest of the
network infrastructure.
spans a distance of a few kilometers (<20 km)
Current solution for access are dial-up modems, higherspeed lines (such as T1/E1), digital subscriber line (DSL),
and cable modem.
However, the access network continues to be a bottleneck,
and users require (and are demanding) higher bandwidth to
be delivered to their machines.
Passive optical networks (PONS) based on inexpensive,
proven, and ubiquitous Ethernet technology (and referred
to as EPON) seem an attractive proposition for this market
segment.
PON technology in general, and EPON in particular, will be
studied in Chapter 5.
8
metro-area network




The metro-area network typically spans a metropolitan
region,
covering distances 20~200 km.
Given the deep-rooted legacy of SONET/SDH ring
networks
 SONET = Synchronous Optical Network;
 SDH = Synchronous Digital Hierarchy
 multi-wavelength versions of these rings are being
deployed for our metro networks.
Important characteristics of optical metro networks will
be discussed in Chapter 6.
9
Backbone network




The backbone network spans long distances, e.g., each
link could be a few hundreds to a few thousands of
kilometers in length (>200km).
set up to provide nationwide or global coverage.
Most telecom backbone networks are deployed today as
an interconnection of "stacked" SONET/SDH rings,
 the fibers support multiple wavelengths using WDM
transmission equipment; however, by "tying“ together
several wavelengths on different fiber segments, one
can create logical rings, and these rings can "meet"
one another at some junction nodes,
Backbone network will be discussed in Chapters 17 and
18.
10
1.3 Telecom Business Models

Skip
11
1.4 skip
12
1.5 Cross-Layer Design


The need for tight coupling between network
architectures and device capabilities.
Without a sound knowledge of device capabilities and
limitations, one can produce architectures which may be
unrealizable; conversely, research on new optical
devices, conducted without the concept of a useful
system, can lead to sophisticated technology with limited
or no usefulness.
13
14
1.6 (TE) vs. (NE) vs. (NP)

Traffic Engineering (TE) vs. Network Engineering
(NE) vs. Network Planning (NP)
 TE: "Put the traffic where the bandwidth is."
 NE: "Put the bandwidth where the traffic is."
 NP: "Put the bandwidth where the traffic is
forecasted to be.“
15
Traffic Engineering (TE)




Since the goal of TE is to "put the traffic where the bandwidth is,"
TE is essentially a "routing problem," where the traffic to be
routed could be packets, packet flows, or bandwidth chunks (i.e.,
circuits).
Routing and assigning appropriate bandwidth to packet flows and
circuits is also referred to as bandwidth provisioning, or
provisioning for short.
Thus, TE is an "online,“ dynamic problem whose decision-making
time is very quick, perhaps on the order of milliseconds.
The typical performance metric used to evaluate a TE algorithm
is



"blocking probability" (by (implicitly) assuming that the network is operating
at "steady-state")
volume of control overhead,
convergence time (to reduce routing instability), etc.
16
Network Engineering (NE)




As a network continues its operation, and as traffic on it builds up
(perhaps asymmetrically), certain parts of the network may
become more congested due to increasing traffic, and these
parts may need "help" in the form of additional capacity to relieve
the congestion.
The decision-making time is perhaps on the order of weeks or
months.
Thus, a typical performance metric for a NE problem could be
"exhaustion probability“ which determines when a current
network, given a traffic-growth pattern, will run into capacity
exhaust.
This (NE) is a very realistic problem in our operational networks;
and, unfortunately, it has been underestimated in the academic
research literature.
17
Network Planning NP



The NP description is almost the same as that of NE,
except that an additional phrase "forecasted to be" is
appended.
NP corresponds to the planning (i.e., design) of a
network from scratch, with a decision-making timescale
of perhaps a few years.
A sample NP problem is the following:
 Given a set of traffic demands between various pairs
of nodes (which is also called a "traffic matrix"), design
the network for minimum cost, i.e., determine how
much capacity to put on each link of the network, as
well as routing of traffic through the network links.
(Note that the typical performance/optimization metric
18
for a NP problem is "cost.")
Network Planning NP



As an example, the cost could be the sum of the
(bandwidth) cost of all the links. In the brief NP problem
description (above), no statement was made about the
connectivity between network nodes (i.e., "network
topology").
By default, the network topology (or graph) may be given.
But an additional dimension to the NP problem would be to
also determine the topology (while achieving minimum
cost).
This "dual" problem can be stated as follows: Given the
traffic demands, and the maximum cost (including perhaps
the topology and capacity of each link), determine how to
establish the demands so that the network throughput (in
terms of carried demands) is maximized
19
Summary of TW, NW, NP
20
1.7 What is an Optical Network?
21
Optical network



The links require “transmission equipment,” while
the nodes require “switching equipment.”
An optical amplifier can simultaneously amplify all of
the signals on multiple wavelength channels
(perhaps as high as 160) on a single fiber link,
independent of how many of these wavelengths are
currently carrying live traffic.
However, many attempts at developing all-optical
switches have indicated that optical switching is still
in its infancy.
22
Optical network


Thus, an optical network is not necessarily all-optical:
the transmission is certainly optical, but the
switching could be optical, or electrical, or hybrid
Also, an optical is not necessarily packet-switched.



It could switch circuits (Chapters 7-12, 15-16), or
sub-wavelength-granularity bandwidth pipes (Chapters 1314), or
"bursts," where a burst is a collection of packets (Chapter
18).
23
PMO and FMO




As an example, consider that two users located at the two coasts of
the USA, need to exchange some large files.
Under present mode of operation (PMO) of today's data networks,
a simlple "traceroute" from Davis to Boston indicates that the file
transfers may encounter 20 router hops, at each of which there
exists the possibility of buffering (and hence delay), as well as loss
(due to buffer overflow).
In future mode of operation (FMO) of data networks, by exploiting
the underlying support from optical-networking technologies, one
should be able to "dial up" a fat bandwidth pipe (of appropriate
capacity and duration) to complete the task. It is not necessary that
all such applications need to accomplished in "one hop.“
20 hops down to 3 or 1 (not necessary)
24
Separated control network
plane



Note that signaling in IP networks is "in-band," so (short)
control packets may have to contend with (long) data
packets for transmission bandwidth.
As data entities that need to get transferred over our
networks get longer, control packets are expected to
encounter more contention for bandwidth.
Thus, one can create a separated control network by
setting aside a wavelength (or a sub-wavelength
granularity bandwidth chunk using a traffic-grooming
principle) on each link for this purpose, so that control
packets have their own dedicated network.
25
1.8 Optical Networking: Need +
Promise = Challenge!


Life in our increasingly information-dependent society
requires that we have access to information at our finger
tips when we need it, where we need it, and in whatever
format we need it.
Internet and ATM networks - unfortunately, don't have
the capacity to support the foreseeable bandwidth
demands.
26
Fiber
Fiber optic technology









huge bandwidth (nearly 50 terabits per second (Tbps),
low signal attenuation(衰減) (as low as 0.2 dB/km),
immunity to electromagnetic interference,
high security of signal because of no electromagnetic
radiation, so difficult to eavesdrop,
no crosstalk and interferences between fibers in the
same cable,
low signal distortion(扭曲),
low power requirement,
low material usage, small space requirement, and low
cost.
high electrical resistance, so safe to use near highvoltage equipment or between areas with different earth
potentials.
28
Solving Problem


Our challenge now is to turn the promise of
optical fiber technology to reality to meet our
information networking demands of the
foreseeable future.
Solving Problem



Network lag.
Not enough bandwidth today
Exponential Growth in user traffic.
29
opto-electronic bandwidth
mismatch

Given that a single-mode fiber's potential
bandwidth is nearly 50 Tbps, which is nearly
3-4 orders of magnitude higher than
electronic data rates of a few gigabits per
second (Gbps), every effort should be made
to tap into this huge opto-electronic
bandwidth mismatch.
30
Solution in Optical Network

In an optical communication network, this
concurrency may be provided according to
either



wavelength or frequency [wavelength-division
multiplexing (WDM)],
time slots [time-division multiplexing (TDM)], or
wave shape [spread spectrum, code-division
multiplexing (CDM)].
31
Why not TDM or CDM?



Optical TDM and CDM are somewhat
futuristic technologies today.
Under (optical) TDM, each end-user should
be able to synchronize to within one time slot.
The optical TDM bit rate is the aggregate rate
over all TDM channels in the system, while
the optical CDM chip rate may be much each
higher than user's data rate.
32
Why not TDM or CDM?


Both the TDM bit rate and the CDM chip rate
may be much higher than electronic
processing speed, i.e., some part of an end
user's network interface must operate at a
rate higher than electronic speed.
Thus, TDM and CDM are relatively less
attractive than WDM, since WDM — unlike
TDM or CDM — has no such requirement.
33
1.9 xDM vs. xDMA



We have introduced the term xDM where x = {W, T,
C} for wavelength, time, and code, respectively.
Sometimes, any one of these techniques may be
employed for multiuser communication in a multiple
access environment, e.g., for broadcast
communication in a local-area network (LAN)
Thus, a local optical network that employs
wavelength-division multiplexing is referred to as a
wavelength-division multiple access (WDMA)
network; and TDMA and CDMA networks are
defined similarly.
34
Basic Concept

WDM is the ability to combine



Multiple sources of data using
Multiple wavelengths (colors) of light on
One strand of fiber cable
Source 1
Source 2
Source 3
Source 4
Its Analog Transmission
Attenuation
Dispersion
Nonlinearity
Reflectance
Transmitted data waveform
Waveform after 1000 km
Fiber Types ...
Multi-mode fiber allows
multiple modes of light to
propagate along its length at
various angles. Typically:
62.5/125um, 50/125um
Single-mode fiber allows a
single mode of light to
propagate along its core
efficiently. Typically:
8/125um, 8.3/125um,
9/125um
1.10 WDM



Wavelength-division multiplexing (WDM) is an approach that
can exploit the huge opto-electronic bandwidth mismatch by
requiring that each end-user's equipment operate only at
electronic rate, but multiple WDM channels from different endusers may be multiplexed on the same fiber.
Under WDM, the optical transmission spectrum is carved up
into a number of non-overlapping wavelength (or frequency)
bands, with each wavelength supporting a single
communication channel operating at whatever rate one desires,
e.g., peak electronic speed.
WDM devices are easier to implement since, generally, all
components in a WDM device need to operate only at
electronic speed; as a result, several WDM devices are
38
available in the marketplace today, and more are emerging.
39
ITU recommended Bands





E: 1360-1460 nm
S: 1440-1530 nm
C: 1530-1565 nm
L: 1565-1625 nm
U: 1625-1675 nm
Light Spectrum
Approximate Attenuation
of Single Mode fiber cable
Infrared
Visible
700
900
1100
“O” Band ~ 1270-1350 nm
“E” Band ~ 1370 - 1440 nm
“S” Band ~ 1470 - 1500 nm
“C” Band ~ 1530 - 1565 nm
“L” Band ~ 1570 - 1610 nm
1300
1500
1700 nm
1.10.1 ITU Wavelength Grid





There is a strong need for the standardization of WDM systems
so that WDM components and equipments from difference
vendors can inter-operate with one another.
Thus, industry standards for wavelengths have been developed
under the leadership of the International Telecommunications
Union (ITU) [ITU02b].
A standard set of wavelengths, called the ITU grid, has been
defined to coincide with the 1550-nm low-loss region of the fiber.
Specifically, this grid is anchored at a frequency of 193.1 THz
(which corresponds to a wavelength of 1552.52 nm).
There is a 100-GHz grid, which means that spacing between
adjacent channels is 100 GHz, which corresponds approximately
to 0.8-nm wavelength channel spacing around the anchor
frequency.
42
WDM-routed networks

Optical signal and wavelength:
Resonate/suppress
WDM and TDM
TDM Time Division
Multiplexing
WDM Wavelength
Division
Multiplexing
ITU wavelength grid
45


For denser packing of channels, a 50-GHz
grid has also been defined around the same
reference frequency of 193.1 THz [ITU02b].
The 50-GHz grid is obtained by adding a
channel exactly half way between two
adjacent channels of the 100-GHz grid.
Continuing this process, a 25-GHz grid can
also be defined, and it can support 600
wavelengths [ITUOZb].
46
WDM


Thus, by allowing multiple WDM channels to coexist
on a single fiber, one can tap into the huge fiber
bandwidth, with the corresponding challenges being
the design and development of appropriate network
architectures, protocols, and algorithms.
WDM devices are easier to implement since,
generally, all components in a WDM device need to
operate only at electronic speed; as a result, several
WDM devices are available in the marketplace today,
and more are emerging.
47
Development of WDM


Since 1990
Several Conference:



Country:


ICC: IEEE International Conference on
Communications
OFC: Optical Fiber Communications
U.S., Japan, Europe
WDM: backbone, global coverage.
48
1.10.2 A sample WDM
Networking Problem



End-users in a fiber-based WDM backbone network
may communicate with one another via all-optical
(WDM) channels, which are referred to as lightpaths.
A lightpath may span multiple fiber links, e.g., to
provide a "circuit-switched" interconnection between
two nodes which may have a heavy traffic flow
between them and which may be located "far" from
each other in the physical fiber network topology.
Each intermediate node in the lightpath essentially
provides an all-optical bypass facility to support the
lightpath.
49
WDM network




Complete graph, N nodes, N(N-1)links.
The number of links is increased with the number of nodes.
Technological constraints dictate that the number of WDM
channels that can be supported in a fiber be limited to W.
RWA Problem (routing and wavelength assignment):
 given a set of lightpaths that need to be established on the
network, and given a constraint on the number of
wavelengths, determine the routes over which these
lightpaths should be set up and also determine the
wavelengths that should be assigned to these lightpaths
so that the maximum number of lightpaths may be
established. .
Lightpaths that cannot be set up due to constraints on routes
and wavelengths are said to be blocked, so the
corresponding network optimization problem is to minimize
this blocking probability.
50
wavelength-continuity
constraint


In this regard, note that, normally, a lightpath
operates on the same wavelength across all
fiber links that it traverses, in which case the
lightpath is said to satisfy the wavelengthcontinuity constraint.
Thus, two lightpaths that share a common
fiber link should not be assigned the same
wavelength.
51
wavelength converter facility


However, if a switching/routing node is also
equipped with a wavelength converter facility,
then the wavelength-continuity constraints
disappear, and a lightpath may switch
between different wavelengths on its route
from its origin to its termination.
RWA problem: Routing and Wavelength
Assignment (RWA) problem
52
1.11 WDM Networking Evolution


Point-to-Point WDM Systems
When the demand exceeds the capacity in existing
fibers, WDM is turning out to be a more costeffective alternative compared to laying more fibers.



installation/burial of additional fibers and terminating
equipment (the "multifiber" solution);
a four-channel "WDM solution" where a WDM multiplexer
(mux) combines four independent data streams, each on
a unique wavelength, and sends them on a fiber; and a
demultiplexer (demux) at the fiber's receiving end
separates out these data streams; and
OC-192, a "higher-electronic-speed" solution.
53
Four channels of point-to-point WDM
54


The analysis in [MePD95] shows that, for distances
lower than 50 km for the transmission link, the
"multi-fiber" solution is the least expensive; but for
distances longer than 50 km, the "WDM" solution's
cost is the least with the cost of the "higherelectronic-speed" solution not that far behind.
WDM mux/demux in point-to-point links is now
available in product form from several vendors such
as IBM, Pirelli, and AT&T [Gree96]. Among these
products, the maximum number of channels is 160
today, but this number is expected to increase soon.
55
1.11.2 Wavelength Add/Drop
Multiplexer (WADM) or Optical
Add/Drop Multiplexer (OADM
Bar state
cross state
56
WADM

Architecture:




States:



DEMUX
A set of 2x2 switches (one switch per wavelength)
MUX
Bar state: If all of the 2 x 2 switches are in the "bar"
state, then all of the wavelengths flow through the
WADM "undisturbed."
Cross state: electronic control (not shown in Fig. 1.3),
then the signal on the corresponding wavelength is
"dropped" locally, and a new data stream can be "added"
on to the same wavelength at this WADM location.
More than one wavelength can be "dropped and
added" if the WADM interface has the necessary
hardware and processing capability.
57
1.11.3 Fiber interconnection
Device



passive star (see Fig. 1.11),
passive router (see Fig. 1.12), and
active switch (see Fig. 1.13).
58
passive star (see Fig. 1.11),



The passive star is a "broadcast" device, so a
signal that is inserted on a given wavelength
from an input fiber port will have its power
equally divided among (and appear on the same
wavelength on) all output ports.
"collision" will occur when two or more signals from
the input fibers are simultaneously launched into the
star on the same wavelength.
Assuming as many wavelengths as there are fiber
ports, an N x N passive star can route N
simultaneous connections through itself.
59
Passive Star
60
passive router (see Fig. 1.12),




A passive router can separately route each of
several wavelengths incident on an input fiber to the
same wavelength on separate output fibers
this device allows wavelength reuse, i.e., the same
wavelength may be spatially reused to carry multiple
connections through the router.
The routing matrix is "fixed" and cannot be changed.
Such routers are commercially available, and are
also known as Latin routers, waveguide grating
routers (WGRs), wavelength routers (WRs), etc.
Again, assuming as many wavelengths as there are
fiber ports, a N x N passive router can route N2
simultaneous connections through itself (compared
to only N for the passive star); however, it lacks the
broadcast capability of the star.
61
Passive Router
62
active switch (see Fig. 1.13).




The active switch also allows wavelength reuse, and it can
support N2 simultaneous connections through itself (like the
passive router).
But the active star has a further enhancement over the passive
router in that its "routing matrix" can be reconfigured on demand,
under electronic control.
However the "active switch" needs to be powered and is not as
fault-tolerant as the passive star and the passive router which
don't need to be powered.
The active switch is also referred to as a wavelength-routing
switch (WRS), wavelength selective crossconnect (WSXC), or
just crossconnect (XC) for short. (We will refer to it as a WRS in
this book.)
63
Active Switch
64
Wavelength Convertible Switch


The active switch can be enhanced with an
additional capability, viz., a wavelength may be
converted to another wavelength just before it
enters the mux stage before the output fiber (see
Fig. 1.6).
A switch equipped with such a wavelengthconversion facility is more capable than a WRS,
and it is referred to as a wavelength-convertible
switch, wavelength interchanging crossconnect
(WIXC), etc
65
1.11.4 Development of WDM
networks


The first generation of WDM: point-to-point physical links include
design and development of WDM lasers and optical amplifiers
(OAMP) [Liu02].
The second generation of WDM is capable of establishing
connection-oriented end-to-end lightpaths in the optical layer by
introducing optical add/drop elements (WADM or OADM) and optical
cross-connects (OXC).



The ring and mesh topologies can be implemented using these OADMs and
OXCs.
The lightpaths are operated and managed based on a virtual topology over the
physical fiber topology, and the virtual topology can be reconfigured dynamically
in response to traffic changes.
The technical issues of second-generation WDM include the development of
OADM and OXC, wavelength conversion, routing and wavelength assignment
(RWA), interoperability among WDM networks, network control and
management.
66
3rd generation

The third generation of WDM is expected to support a
connectionless optical network. The key issues include
the development of optical access network (such as
passive optical network (PON)), and optical switching
technologies, generically referred to as Optical "X"
Switching (OXS), where X = P (for packet), B (for burst),
L (for label), F (for flow), C (for cluster or circuit), etc.
Sorne of these techniques, namely, OPS and OBS will
be discussed in Chapters 17 and 18.
67
68
1.12 WDM Network Construction




1.12.1 Broadcast-and-Select (Local) Optical
WDM Network
A local WDM optical network may be constructed by
connecting network nodes via two-way fibers to a
passive star,
The information streams from multiple sources are
optically combined by the star and the signal power
of each stream is equally split and forwarded to all of
the nodes on their receive fibers. A node's receiver,
using an optical filter, is tuned to only one of the
wavelengths; hence it can receive the information
stream.
the passive-star can support “multicast”
services.
69
70
Passive-Star-Based Optical WDM LAN vs.
Centralized, nonblocking-Switch-Based LAN
Passive Star WDM has following advantages:


In the space-division-switch solution, the "switching
intelligence" is centralized. However, the passive star
relegates the switching functions to the end nodes If
a node is down, the rest of the network can still
function. Hence, the passive-star solution enjoys the
fault-tolerance ad-vantage of any distributed
switching solution, relative to the centralized-switch
architecture, where the entire network goes down if
the switch is down.
71
Passive Star WDM has following
advantages


it allows multicasting "for free." There are
some processing requirements with respect to appropriately coordinating the
nodal transmitters and receivers.
Centralized coordination for supporting
multicasting in a switch (also referred to
as a "copy" facility) is expected to require
more processing.
can be potentially much cheaper since it is
purely glass with very little electronics.
72
1.12.2 Wavelength-Routed
(Wide-Area) Optical Network



The network consists of a photonic switching fabric,
comprising "active switches" connected by fiber links
to form an arbitrary physical topology.
Each end-user is connected to an active switch via a
fiber link. The combination of an end-user and its
corresponding switch is referred to as a network
node.
Each node (at its access station) is equipped with a
set of transmitters and receivers, both of which may
be wavelength tunable. A transmitter at a node
sends data into the network and a receiver receives
data from the network.
73
74