Traffic Grooming in WDM Networks
Kevin Su
University of Texas at San Antonio
9/22/2003
Kevin Su ([email protected])
1
WDM Technology
WDM: stands for wavelength division multiplexing, it is a
technology that divides the bandwidth of an optical fiber into many
non-overlapping wavelengths, so that multiple communication
channels can operate simultaneously on different wavelengths.
 Each piece of equipment which sends an optical signal has an
illusion that it has its own fiber.
 Increases the transmission capacity of optical fibers.
 Allows simultaneously transmission of multiple wavelengths
within a single fiber.
(up to 320 wavelengths per fiber; per wavelength, 10Gb/s, OC-192,
today; expected to grow to 40Gb/s, OC-768, soon)

9/22/2003
Kevin Su ([email protected])
2
WDM Technology





SONET Add/Drop Multiplexer (SADM): can be used to
aggregate lower rate stream from different end-users into a single
high-rate SONET stream in Time Division Multiplexing (TDM)
fashion.
Optical cross connect (OXC): is a network element that perform
wavelength switching and/or wavelength conversion.
Wavelength switching: switch traffic from a wavelength of an
input fiber link to the same wavelength of any outgoing fiber link.
Wavelength conversion: traffic from a wavelength of an input
fiber link can be switched to any wavelength in the outgoing fiber
link.
Lightpath: a traffic route that using same or different wavelength
without optical-electronic-optical conversion in WDM optical
networks.
9/22/2003
Kevin Su ([email protected])
3
Wavelength Switching / Conversion
fiber2
1
2
OXC
fiber1
1
2
fiber3
1
2
9/22/2003
Kevin Su ([email protected])
4
Traffic Grooming

Motivation:



Problem Formulation:



Gap between capacity of a WDM Channel (OC-48, or OC-192, or OC768) and bandwidth requirement of a typical connection request (e.g. STS1, OC-3, OC-12 etc)
In order to use network efficiently, low-speed traffic streams need to be
efficiently multiplexed, or “groomed” onto high-speed lightpaths
Given a network configuration and a set of connection requests with
different bandwidth granularities, such as OC-12.
Determine how to set up lightpaths to satisfy the connection requests.
Category:


9/22/2003
Static Case (static traffic): Set of connection requests can all be given in
advance
Dynamic Case (dynamic traffic): connection requests are given one at a
time
Kevin Su ([email protected])
5
Network Topology

SONET Ring:
SONET ring network is currently the most widely deployed
optical network infrastructure.

Mesh Network:
Due to the increase of Internet traffic, more WDM networks
would be deployed in general mesh topology to meet this
demands. Thus more work needs to be done in the mesh
networks grooming problem.
9/22/2003
Kevin Su ([email protected])
6
Traffic Grooming Example
9/22/2003
Kevin Su ([email protected])
7
Traffic Grooming Example
9/22/2003
Kevin Su ([email protected])
8
Traffic Grooming Example
9/22/2003
Kevin Su ([email protected])
9
Modeling Traffic Grooming Problem

Static Case:


The problem is usually formulated as an Integer Linear Program (ILP)
problem and get different optimal solution according to different goals.
Unfortunately, for large networks it is computationally infeasible to solve
the ILP problem. Therefore, many heuristic algorithms were proposed.
Dynamic Case:

9/22/2003
Usually problem is divided into 4 sub-problems. Using different
algorithms to solve different sub-problems. Recently one integrated
algorithm was proposed to solve the 4 problems altogether.
Kevin Su ([email protected])
10
Static Case (ILP)
9/22/2003
Kevin Su ([email protected])
11
Subproblems in Traffic Grooming

(1) Determine the virtual topology that consists of lightpaths;


NP-hard
(2) Routing the lightpaths over the physical topology;

NP-hard

(3) Performs wavelength assignment to the lightpaths;
(2) and (3) together are RWA problem

(4) Routing the traffic on the virtual topology.
9/22/2003
Kevin Su ([email protected])
12
Future Challenges


Grooming with Protection
Provide two routing path for each connection requests, one is
primary traffic stream path (TSP), the other is link-disjoint backup
traffic stream path (TSP). 1:1 protection (dedicated),
1: m (shared backup TSP).
Multicast Traffic Grooming
The objective is to construct multicast trees (or light-trees) that
optically carry the multicast traffic from the source to the
destination nodes, which will reduce the cost of network.
9/22/2003
Kevin Su ([email protected])
13
Protection
9/22/2003
Kevin Su ([email protected])
14
Multicast Traffic Grooming
9/22/2003
Kevin Su ([email protected])
15
Multicast Traffic Grooming
• Session 1: Source = A; Destination = {B, C}; Traffic demand = 1
unit;
• Session 2: Source = B; Destination = {C}; Traffic demand = 2 unit;
• Session 3: Source = A; Destination = {F}; Traffic demand = 1 unit;
• Routing the demands using an SMT requires 7 ADMs and two
wavelengths, as shown in Figure 1. However,
• using the routing shown in Figure 2 costs just 6 ADMs and one
wavelength
9/22/2003
Kevin Su ([email protected])
16
References
• Ruda Dutta and George N. Rouskas, Traffic Grooming in WDM
Networks: Past and Future, IEEE Network 2002
• Sashisekaran Thiagarajan, Arun K.Somani Traffic Grooming for
Survivable WDM Mesh Networks OptiComm 2001
• Ahmed E. Kamal, Raza Ul-Mustafa, Multicast Traffic Grooming in
WDM Networks, OptiComm 2003
• Hui Zang, Canhui Ou, Biswanath Mukherjee Path-Protection
Routing and Wavelength Assignment (RWA) in WDM Mesh
Networks Under Duct-Layer Constraints, IEEE/ACM Transactions
on Networking, April 2003
9/22/2003
Kevin Su ([email protected])
17