See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/272792217

On Survivability of Dynamically Provisioned Wavelength-Routed Networks

Conference Paper · December 2007

CITATIONS READS
0 1,028

3 authors, including:

Partha Goswami D. Datta

Indian Institute of Technology Kharagpur Indian Institute of Technology Kharagpur
10 PUBLICATIONS 65 CITATIONS 74 PUBLICATIONS 879 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Long Reach RSOA based Passive Optical Networks View project

book project View project

All content following this page was uploaded by Partha Goswami on 26 February 2015.

The user has requested enhancement of the downloaded file.

 On Survivability of Dynamically Provisioned Wavelength-Routed Networks
P. Goswami, S. K. Ghosh and D. Datta
Indian Institute of Technology, Kharagpur
E-mail: partha@cc, skg@sit, ddatta@ece {.iitkgp.ernet.in}

alleviated in this process, thanks to the ongoing

Abstract
developments in optical multiplexing, demultiplexing and
Optical networks employing wavelength-routing technique switching technologies. However, the electronic networking
have emerged as the potential candidate for next- devices, such as, IP routers, ATM (asynchronous transfer
generation backbone networks. Given the wide range of mode) switches, SDH (synchronous digital hierarchy) or
services to be offered by the high-capacity optical SONET (synchronous optical network) terminals would
backbones, optimal provisioning of optical resources and remain connected to the WADMs/OXCs to provide the
network survivability against the occurrences of failures desired networking services.
have become major issues to be considered in the network A typical mesh-connected physical topology of
design. The survivability schemes of such networking wavelength-routed networks consisting of OXCs is shown
infrastructures can be realized in two basic forms, viz., in Fig. 1, while in Fig.2 the configuration of a typical OXC
restoration-based and protection-based schemes or an is illustrated. As shown in Fig.2, OXCs with the help of
appropriate combination of both. In a real-life network WDM devices (optical demultiplexers/multiplexers and
setting, both resource provisioning against time-varying switches) and transponders (optical transceivers) can
bandwidth demands and failure recovery schemes need to optically bypass or add/drop a specific wavelength at a
be realized in a dynamic manner. This article provides a given intermediate node in a network, governed by the
brief overview of the survivability measures for dynamically traffic demands between node pairs (source and destination
provisioned optical backbone networks employing nodes).
wavelength-routing technology.

I. Introduction

In today’s networking infrastructure, the ever-increasing

Internet traffic has already superceded the existing voice
traffic. The large volume of Internet traffic, with its
connections having much wider geographical reach and
longer durations than voice traffic, has enhanced the need
for high-speed wide-area backbone network more than ever
before. While responding to this demand, national
networking backbones all over the world have gone through
wide deployment of optical networking setup, mostly using
point-to-point wavelength-division multiplexing (WDM)
technology. However, with the point-to-point WDM
technology, the unprecedented growth in IP traffic makes it Figure 1 Wavelength-routed optical WDM network
difficult for IP routers to handle the network traffic with
hop-by-hop packet processing and forwarding at The end-to-end optical connectivity between various node
intermediate nodes. The burden of routers in electronic pairs in wavelength-routed networks needs to be
domain can be significantly reduced by utilizing some established by choosing a set of lightpaths, with lightpaths
emerging optical technologies, which will enable the implying a specific wavelength of light propagating over a
intermediate nodes in a network to carry out some of the given route between a node pair. The topology formed by
routing functionalities in optical domain itself. Such these lightpaths, referred to as logical topology in the
networks, generally known as wavelength-routed literature, might turn out to be significantly different from
networks, would employ wavelength routing with the use of its underlying physical topology (fiber connectivity) [1-8].
wavelength add-drop multiplexers (WADM) for ring-based One needs to use an efficient logical topology design
fiber connectivity, typically used in metropolitan networks, (LTD) methodology for such a network to choose a set of
and optical cross-connects (OXC) for mesh-based fiber direct (i.e., single-hop) lightpaths between a given set of
connectivity, generally preferred for wide-area backbone node pairs. With a given logical topology designed from
networks [1-4], wherein the transit traffic at intermediate the traffic demands between node pairs, the route and the
nodes can be bypassed in optical domain itself. Thus, the wavelength of each lightpath over the physical topology are
problem of hop-by-hop packet forwarding (in electrical determined by using a judicious routing and wavelength
domain) by IP routers at intermediate nodes can be assignment (RWA) algorithm.
 II. Design Methodologies for Wavelength-Routed
Networks

The first innovative solution to set up WDM networking

based on the lightpaths was proposed in [5]. It may be
noted that, although setting up direct lightpath between
each node pair using OXCs reduces electronic processing at
intermediate nodes, it might not be feasible due to the
limited resources (upper limit on number of optical
transceivers, number of available wavelengths etc.). On the
other hand, provision of OEO (optical-electronic-optical)
conversion at OXCs of intermediate nodes, enables more
traffic grooming at intermediate nodes (although with
higher electronic costs), leading to efficient bandwidth
utilization of fibers. Consequently, performing LTD/RWA
with a limited number of lightpaths and routing the entire
traffic over the designed lightpaths become an optimization
Figure 2 WDM-based OXC architecture problem, wherein some node pairs may need to
communicate between themselves using multiple lightpaths
In a wavelength-routed backbone network, the failure of a (multihop) with OEO conversion at intermediate nodes.
network element, such as, fiber link, OXC etc. may cause The design optimization can be carried out with relevant
failure of several lightpaths, leading to massive data and performance metrics, such as, traffic congestion in
revenue losses. Extensive studies have been carried out in lightpaths, propagation delay, average hop-count, bit-error
the recent years to make these backbone networks rate etc. [10]. In particular, the design of the logical
survivable against such incidents of failure and, there are topology for network-wide congestion minimization for an
several approaches to ensure the network survivability [9]. acceptable propagation delay can be formulated as a mixed-
Service availability in such high-capacity networks needs to integer linear programming (MILP) problem [1]. However,
be assured through a combination of equipment reliability in general, MILP is found to be computationally intractable
and the implementation of various network survivability for large network, and hence this task is usually split into
approaches. Survivability refers to the ability of the two subproblems: first step - finding a set of feasible direct
network to transfer the interrupted services on to spare lightpaths between a set of candidate node pairs and routing
network capacity to circumvent a point of failure in the them through the OXCs of appropriate intermediate nodes
network. Such failures can involve fiber cuts and node- and fiber links with specific wavelength assignments;
related failures. Fibers cuts primarily occur due to human second step - routing of entire network traffic (including
errors or natural disasters leading to link failures, and node node pairs both with and without direct lightpaths) over the
failures occur due to component failure within equipment lightpaths designed in the first step. Both the subproblems
or entire node shutdown caused by accidents such as fire, are solved using appropriate heuristic algorithms [6-7],
flood, earthquake etc. Node equipment or component wherein the solutions of the first subproblem (i.e.,
failure can be dealt with standby redundancy. However, the acceptable lightpaths with specific wavelengths through
whole node shutdown is difficult to recover, as it requires specific fiber links and nodes) are supplied to the second
huge spare capacity. Nevertheless the occurrence of such subproblem as inputs to obtain a near-optimal result.
events is rare and can be handled by applying technique of
recovery from link failures by rerouting the traffic around
the failed nodes. The monitoring and maintenance of links III. Survivability Measures in Wavelength-Routed
is more difficult than that of nodes since links generally Networks
span over a long distance with variable geographical
conditions, such as mountain, underwater or in a densely- The above design methodologies for wavelength-routed
populated city. Extensive studies have been carried out in networks need to be supplemented with additional
the recent years to make wavelength-routed backbone resources and functionalities, so that the traffic streams
networks survivable against such incidents of failure and, carried by all the designed lightpaths can survive (i.e., are
there are several approaches to ensure the network not disrupted) in the event of network failures. Survivable
survivability [9]. In this article, we provide a brief network architectures are in general based either on
overview of the various possible schemes to provision dedicating back-up resources in advance (referred to as
dynamically the wavelength-routed backbones with protection-based survivability), or on dynamic restoration
appropriate survivability measures against fiber-link in the event of a failure [9], [11-13]. In protection-based
failures. survivability, the disrupted network service in a WDM
network can be brought back by utilizing the dedicated all the connections that traverse the failed link are rerouted
network resources, such as, link-disjoint back-up lightpaths around that link, and the source and destination nodes of
(similar protection schemes, viz., automatic protection the connections remain oblivious to the link failure.
switching and self-healing rings are already in use for
single-wavelength networks, e.g., SDH/SONET rings). In
dynamic network restoration, the spare capacity available
in the network can be utilized for restoring services
affected by the failure. Generally dynamic restoration
schemes are more efficient in utilizing network resources
and capacity, while dedicated restoration schemes offer
guaranteed restoration capability with faster recovery time.
Both protection as well as restoration schemes can be
applied for a complete lightpath between a node pair or for
each fiber link of a given lightpath, leading to path Figure 5 Link protection/restoration
protection/restoration and link protection/restoration IV. Dedicated and Shared Protection Schemes for
schemes. In the following we describe these
Network Survivability
protection/restoration schemes in brief.
Protection schemes adopted in wavelength-routed networks
In path protection, as shown in Fig.3, backup resources
can employ dedicated or shared backup lightpaths. In
are reserved during lightpath setup (applicable for both for
dedicated-path protection, the resources along a backup
static provisioning during initial network setup process, or
path are dedicated for only one connection and are not
online dynamic provisioning), while in path restoration
shared with the backup paths for other connections. For
backup lightpaths are discovered dynamically after the
each lightpath request, first primary path and wavelength
occurrence of link failure. In path restoration, when a link
are searched. After successful searching of primary path
fails (Fig.4), the source and destination nodes of each
and primary wavelength, backup path (which should be
lightpath/connection that traverses the failed link are
link-disjoint of the primary path) and dedicated backup
informed about the failure via messages from the nodes
wavelength is searched. If wavelengths are available in
adjacent to the failed link, and subsequently the alternate
both primary and backup path then RWA is successful for
paths are dynamically explored to support the disrupted
that lightpath request, as shown in Fig. 6.
traffic.

Figure 3 Path protection: primary and backup paths

Figure 6 Dedicated-path protection

In shared-path protection, the resources along a backup

path may be shared with other backup paths. As a result,
backup wavelengths are multiplexed among different
failure scenarios (which are not expected to occur
simultaneously), and therefore, shared-path protection is
more capacity-efficient when compared with dedicated-
path protection. For each lightpath request, first primary
path and primary wavelength are identified. Next backup
path (which is fiber link-disjoint of the primary path) and
Figure 4 Path restoration: link-failure messages sent to all source backup wavelength (which may be used by another backup
and destination nodes of the connections traversing the failed link. lightpath, but the corresponding primary lightpaths need to
be link-disjoint) are identified. If wavelengths are available
In link protection, backup resources are reserved around in both primary and backup paths, then RWA is successful
each link during connection setup, while in link restoration for that lightpath request, as shown in Fig.7 (note that,
the end nodes of the failed link dynamically discover a Backup 1 and Backup2 lightpaths are assigned the same
route around the link. In link protection/restoration (Fig. 5), wavelength over the fiber link between Node 1 and Node 2
– on the other hand, in Fig.6 for dedicated protection, these annual lease. However, the actual bandwidth needed by an
wavelengths are different). ISP for a specific link at any instant is very dynamic and
only occasionally reaches the full amount of leased capacity
(based on worst-case traffic).
The Internet2 network is a typical example of U.S.
research and education community with a dynamic,
innovative and cost-effective hybrid optical and packet
network. Fig. 8 shows the Internet weather map [14] of
Internet2 Network [15] on a map of the United States,
which shows the current (1st Nov 2007 16:15:12 UTC) level
of traffic in each link along with link utilization. On a
typical link of this network, bandwidth demand is high
during daytime hours, while, at night, less bandwidth is
Figure 7 Shared-path protection
necessary as shown in Fig. 9 (measured on 1st November
2007 at router interface so-0/0/0.0 of Atlanta Router for a
In dedicated-link protection, at the time of connection
period of last 7 days). If an ISP can track these changes in
setup, for each link of the primary path, a backup path and
bandwidth demand in real time, it can dynamically lease
wavelength are reserved around that link and are dedicated
only the bandwidth necessary, and reduce its operating
to that connection. In general, it may not be possible to
expenditure. Thus large-scale ISPs require geographically
allocate a dedicated backup path around each link of the
distributed IP Routers to be connected by high-bandwidth
primary connection and on the same wavelength as the
pipes (lightpath for optical backbone) with quickly-
primary path. The dedicated-link protection utilizes
reconfigurable capacities.
wavelengths very inefficiently; hence dedicated-link
protection is not actively considered in realistic networking
setups. In shared-link protection, the backup resources
reserved along the backup path may be shared with other
backup paths. As a result, backup wavelengths are
multiplexed among different failure scenarios (which are
not expected to occur simultaneously), and therefore
shared-link protection is more capacity-efficient when
compared with dedicated-link protection. It may be added
that, in link restoration, the end nodes of the failed link
participate (as in path restoration) in a distributed algorithm
to dynamically discover a route around the link. If no routes
are available for a broken connection, then the connection
is dropped.
Figure 8 Internet weather map of Internet2 Network
V. Dynamic Lightpath Provisioning with
Survivability Measures

When a wavelength-routed network is deployed in a real-

life network setting, one would need to update the
provisioned lightpaths dynamically with time, as governed
by the arrival of time-varying traffic demands and
occurrences of failure. The LTD and RWA problems, as
discussed in Section II, are basically static lightpath
establishment (SLE) problems applicable for initial network
setup, whereas a real-life network already provisioned with Figure 9 Bandwidth demand in an Internet backbone link
an initial design using SLE must be able to respond
In wavelength-routed networks handling the above
dynamically to the arrival of bandwidth demands and
situation would have to use an optical client layer for
failure messages. In today’s business model, a network
efficient resource provisioning, wherein an optical client
operator owning the physical telecom infrastructure (optical
node would submit the requests for lightpaths to be setup as
fibers), charges its customer, typically an Internet service
needed. In a dynamic situation, a lightpath has to be set up
provider (ISP), a constant bandwidth rate, based on the
for each connection request as it arrives, and the lightpath
maximum bandwidth needed by the ISP for a period of
would be released after some finite amount of time. The
time. Usually, the interfaces of two adjacent IP routers are
problem of lightpath establishment in a network with
interconnected by leased lines (namely, a circuit of fixed
dynamic traffic demands is called the dynamic lightpath
bandwidth) with a long holding time, usually based on an
establishment (DLE) problem. Vigorous investigations are link-state information needs to carry also optical resource
being made in this direction all over the world leading to information, including wavelength and bandwidth
the design of DLE schemes using efficient LTD/RWA availability, shared risk link group (SRLG), physical layer
algorithms and appropriate signaling protocols for the constraints, and link protection information among others.
distribution of control messages and network state This information is then used to build and update the
information [16-18]. optical network traffic engineering database as (shown in
In order to execute the DLE schemes as discussed in the Fig 10) which guides the route selection algorithm.
above, optical backbone networks can be conceived to be 3. Lightpath management: Lightpath management in Fig.10
having two distinct logical planes, viz., data plane and is concerned with setup and teardown of lightpaths as well
control plane, operating over the same physical topology. as coordination of protection switching in case of failures.
In data plane data would be flowing through lightpaths as Once a lightpath is selected a signaling protocol must be
per the network needs, which will in turn be controlled by invoked to setup and manage the connection. Two
the exchange of appropriate control messages in the control protocols have currently been defined to signal a lightpath
plane of the network, realized using some dedicated setup; RSVP-TE [23] and CR-LDP [24]. RSVP-TE is
resources, viz., time slots etc. in some specific wavelengths based on the resource reservation protocol (RSVP), with
over appropriate fiber links. Currently most work on appropriate extension to support traffic engineering where
defining control plane protocols in optical network takes as CR-LDP is an extension of Label distribution protocol
place under the auspices of IETF, reflecting a convergence (LDP) augmented to handle constrained-based routing.
of optical networking and IP communities to developing The typical control plane elements are shown in Fig 10,
technology built around a single common frame-work for which are considered to be independent of each other and
controlling both IP and optical network element [19]. There thus separable. This modularity allows each component to
are three components of the control plane that are crucial to evolve independent of others or to be replaced with a new
setting up lightpaths within the optical network and thus and improved protocol. As the optical networking and IP
relevant to traffic engineering as shown in Fig 10. communities come together to define standards, the
constraints and the new realities (namely the explosion in
the number of channels in the network) imposed by the
Topology
and
Updates Traffic optical layer and WDM technology is affecting our long
Resource Discovery Engineering held assumption regarding issues such as routing, control,
Database
discovery and so on which have been developed for mostly
Topology and
resource opaque electronic network.
Route information Optical Node Model: Network control, as discussed in the
information
foregoing, can be classified as centralized or distributed
Lightpath Route control. In centralized control, the route computation / route
Management Computation
control commands are implemented and issued from one
place. Each node communicates with a central controller
Signaling Protocol RWA algorithms / and the controller performs routing and signaling on behalf
traffic engineering of all other nodes. In a distributed network control each
node maintains partial or full information about the network
Figure 10 Control Plane Components state and existing connections. Routing and signaling are
performed independently at each node. Therefore
1. Topology and resource discovery: The main purpose of coordination between nodes is needed to alleviate the
discovery mechanism, as shown in Fig. 10, is to problem of connections. An optical node consists of an
disseminate network state information including resource OXC and an optical network control plane as shown in Fig.
use, network connectivity, link capacity availability, and 11. Between two neighboring nodes preconfigured control
special constraints. The link management protocol (LMP) channel is provisioned which may be in-band or out-of-
[20] has been proposed to perform neighbor discovery and band to provide IP connectivity for transmission of control
link monitoring. LMP is expected to run between messages such as routing or signaling messages. The
neighboring OXCs and can be used to establish and switching function in an OXC is controlled by
maintain control channel connectivity, monitor and verify appropriately configuring the cross-connected fabric.
data link connectivity and isolate link fiber or channel Conceptually this leads to the setting up a crossconnect
failures. table whose entries have the form as <input port, input
2. Route computations: This component in Fig. 10 employs wavelength, output port, output wavelength>, indicating
RWA algorithms and traffic engineering functions to select that the data stream entering on a given input wavelength at
an appropriate root for the requested lightpath. There are a specific input port will be switched to the specified output
currently several efforts underway to extent open shortest wavelength in the specified output port.
path first (OSPF) to support Generalized Multi-Protocol
Label Switching (GMPLS) [21] and Traffic Engineering
(TE Extensions to OSPF Version 2) [22]. In particular,
 OXC2
Control 1 Control 2

Control
OXC3
OXC1
Control

Working
OXC1 OXC2 Lightpath Control
Data
OXC4
OXC5
Optical Network Optical Network
Node 1 Node 2 Protection
Lightpath

Figure 11 Optical node model Figure 12 Dedicated (1:1) link protection (control plane exchanges
GMPLS-based realization of protection and restoration shown in the above include all the relevant messages, viz., RSVP
schemes: The development of a common control plane for path refresh message, Resv message, LMP messages etc.)
both optical and electronic networks will need the relevant
functionalities realized through appropriate signaling, As shown in Fig.12, in order to provide dedicated (1:1) link
routing, and link management protocols to enable protection, failures are localized before the switchover can
intelligent fault management. At the connection level, occur. Once a failure has been localized between the two
GMPLS-based fault management would consist of four adjacent nodes, the upstream node can initiate local span
primary steps: detection, localization, notification and protection by sending an RSVP path refresh message. Use
mitigation. Fault detection should be handled at the layer of path refresh messages is a distinct feature of RSVP that
closest to the failure; for optical networks this is the allows intermediate nodes to refresh the state of an LSP.
physical (optical) layer. One measure of fault detection at This feature allows for a switchover from the primary to the
the physical layer is detecting loss of light (LOL); other backup channel. Note that the benefit of exchanging the
techniques based on optical signal-to-noise ratio (OSNR), shared-protection configuration in advance (using LMP) is
bit-error rate (BER), dispersion, crosstalk, and attenuation that it minimizes the potential backup channel (label)
are still being developed. conflict when protection switching takes place. When the
Fault localization requires communication between nodes downstream node receives the path message, it verifies the
to determine where the failure has occurred (e.g., parameters, updates the signaling state, and either responds
SDH/SONET alarm indication signal (AIS) is used to with a Resv message with a new label or generates an error
localize failures between spans). One interesting message.
consequence of using LOL to detect failures is that, LOL Link Restoration: As mentioned earlier, link restoration
propagates downstream along the connection path, and is designed to react to failures quickly in an exploratory
therefore all downstream nodes may detect the failure. The mode. This in turn involves dynamic resource allocation
LMP protocol includes a fault localization procedure and route calculation, and therefore takes more time to
designed to localize failures in both transparent (all-optical) switch to an alternate path than protection techniques.
and opaque (optoelectrical) networks. This is done by Similar to protection techniques described previously,
sending LMP channel fail messages between adjacent restoration can be implemented at the source or an
nodes over a control channel (maintained separately from intermediate node, once the responsible node has been
the data-bearing channels, as mentioned earlier). Once a notified. In case of link restoration, where traffic is
failure has been detected and localized, failure information switched via an alternate route around a failure, a new path
is notified to all concerned and protection and/or restoration is selected at an intermediate node, as illustrated in Figures
are used to mitigate the failure. Some example of such 13a & 13b. This involves passing the traffic through
failure recovery (for dedicated protection) procedure would additional transit nodes. Link restoration may be beneficial
be worthwhile to discuss as follows. for connections that span multiple hops and/or large
Link protection: As mentioned earlier, link protection is distances because the latency incurred for failure
carried out between two adjacent nodes and involves notification may be significantly reduced. In this case only
switching to a backup channel or link when a failure occurs segments of the connection are rerouted instead of the
(Fig.12). As a part of the GMPLS-routing extensions [21] entire path. Link restoration, however, may break traffic
the link protection type (LPT) is advertised so that span engineering requirements if a strict-hop explicit route is
protection can be used in route calculation. Once a route is defined for the connection. Furthermore, the constraints
selected, the connection is signaled using RSVP-TE or CR- used for routing the connection must be forwarded so that
LDP, which includes a protection bit vector indicating an intermediate node doing link restoration is able to
which LPTs are acceptable for the connection. calculate an appropriate alternate route.
 discovered (from the spare capacity in the network) upon
OXC2
the occurrence of a failure. The studies reported in the
literature indicate that shared protection schemes provide
Notify
significant savings in capacity utilization over dedicated
OXC1
OXC3 protection schemes. Furthermore, the restoration schemes
Control
are in general more capacity-efficient as compared to the
Working protection schemes, however with slower recovery process.
Lightpath Control
It is also important to note that, if two fiber links fail in the
Notify
OXC4 network at the same time, then the number of connections
OXC5 that are dropped under shared-path or dedicated-path
protection schemes would be more than the number of
connections dropped under shared-link protection. Thus,
with a given network setting, one needs to choose
Figure 13a Failure of working lightpath judiciously a specific scheme or a combination, governed
by the available resources, needed speed of service
recovery, and the number of possible link failures occurring
OXC2 concurrently in the network. Finally, it may be worthwhile
to indicate that, in a realistic network setting, different users
may have different needs, and also different parts of a
network may have different failure characteristics, thereby
OXC1
OXC3 necessitating adequate investigations on the differentiated
Control survivability methods. Moreover, traditionally the notion of
network service has been binary, i.e., it is either available
Control (as contracted) or not. But the notion of degraded service
also needs to be addressed, so that even if some parts of the
OXC4
network are down, service can still be provided at a reduced
OXC5 quality, if possible. Investigations on multiple and large-
Working scale correlated failures (or attacks) are also going to be of
Lightpath future interest [25].

Figure 13b Restored working lightpath References

1. R. Ramaswami and K. N. Sivarajan, Optical Networks: A
Practical Perspective, Morgan Kaufmann, Second Edition,
VI. Conclusion 2004.
2. B. Mukherjee, Optical Communication Networks, McGraw-
Hill, 1997.
Dynamically provisioned wavelength-routed backbone 3. S. Dixit (Ed.), IP over WDM: Building Next Generation
networks can potentially transport several Gbps of data on Optical Internet, John Wiley & Sons, 2003.
each lightpath over the fiber-optic physical topology. 4. K. M. Sivalingam and S. Subramaniam (Ed.), Emerging
However, the high capacity of each lightpath in a fiber link Optical Network Technologies: Architectures, Protocols and
has the associated problem that a link failure can potentially Performance, Springer, 2004.
lead to the loss of a large amount of data and revenue. All 5. I. Chlamtac, A. Ganz and G. Karmi, “Lightpath
such failures must be dealt with promptly and efficiently. A Communications: An Approach to High Bandwidth Optical
key requirement for such networks is to develop of a WAN’s,” IEEE Transactions on Communications, Vol. 40,
common control plane for both optical and electronic No.7, July 1992, pp. 1171-1182.
6. B. Mukherjee, D. Banerjee, S. Ramamurthy and A. Mukherjee,
networks with the features of signaling, routing, and link “Some Principles for Designing a Wide-Area WDM Optical
management protocols to enable intelligent fault Network,” IEEE/ACM Transactions on Networking, Vol. 4,
management. This article provides a brief overview of the No. 5, October 1996, pp. 684-696.
survivability measures for dynamically-provisioned optical 7. R. Ramaswami, and K. N. Sivarajan, “Design of Logical
backbone networks employing wavelength-routing Topologies for Wavelength-Routed Optical Networks,”
technology. IEEE Journal on Selected Areas in Communications, Vol.
In case of link failures of a WDM backbone network, 4, No. 5, June 1996, pp. 840-851.
there are two basic survivability paradigms: path 8. H. Zang, J. P. Jue, and B. Mukherjee, “A Review of Routing
protection/restoration and link protection/restoration. In and Wavelength Assignment Approaches for Wavelength-
Routed Optical WDM Networks,’’ Opt. Net. Mag., vol. 1, no.
path and link protection schemes, backup paths and 1, Jan. 2000, pp. 47–60.
wavelengths are reserved in advance at the time of
connection setup. Path and link restoration schemes are
exploratory in nature wherein backup paths or links are
 9. S. Ramamurthy, L Sahasrabuddhe and B. Mukherjee,
“Survivable WDM Mesh Networks,” IEEE/OSA Journal of
Lightwave Ttechnology, Vol. 21, No.4, April 2003, pp. 870-
883.
10. B. Ramamurthy, D. Datta, H. Feng, J. P. Heritage and B.
Mukherjee, “Impact of Transmission Impairments on the
Teletraffic Performance of Wavelength-Routed Optical
Networks,” IEEE/OSA Journal of Lightwave Technology,
Vol.17, No.10, October 1999, pp. 1713-1723.
11. S. Ramamurthy and B. Mukherjee, “Survivable WDM Mesh
Networks, Part I - Protection” Proc. IEEE Infocom ’ 99, pp.
744-750.
12. S. Ramamurthy and B. Mukherjee, “Survivable WDM Mesh
Networks, Part II - Restoration” Proc. ICC ’ 99, pp. 2023-
2029.
13. L Sahasrabuddhe, S. Ramamurthy and B.Mukherjee “Fault
Management in IP-Over-WDM Networks: WDM Protection
Versus IP Restoration”, IEEE Journal on Selected Areas In
Communications, Vol. 20, No. 1, January 2002, pp. 21-33.
14. http://weathermap.grnoc.iu.edu/i2_jpg.html
15. http://www.internet2.edu/network/
16. Scott F. Gieselman, Narendra K. Singhal, and Biswanath
Mukherjee “Minimum-Cost Virtual-Topology Adaptation for
Optical WDM Mesh Networks”, IEEE International
Conference on Communications, (ICC 2005), Vol 3, pp.
1787- 1791, 16-20 May 2005.
17. A. Gençata, and B. Mukherjee, “ Virtual-Topology Adaptation
for WDM Mesh Networks Under Dynamic Traffic”
IEEE/ACM Transactions On Networking, Vol. 11, No. 2, April
2003, pp. 236- 247.
18. K. Shiomoto, E. Oki, W. Imajuku, S. Okamoto and N.
Yamanaka “Distributed Virtual Network Topology Control
Mechanism In GMPLS-Based Multiregion Networks” IEEE
Journal On Selected Areas In Communications, Vol. 21, No. 8,
October 2003, pp. 1254-1262.
19. B. Rajagopalan, J. Luciani and D. Awduche “IP over Optical
Networks: A Framework”, Request for Comments: 3717
March 2004.
20. “Link Management Protocol (LMP)”, Request for Comments:
4204, October 2005.
21. K. Kompella and Y. Rekhter, “OSPF Extensions in Support of
Generalized Multi-Protocol Label Switching (GMPLS)”,
Request for Comments: 4203, October 2005.
22. D. Katz , K. Kompella and D. Yeung “Traffic Engineering
(TE) Extensions to OSPF Version 2”, Request for Comments:
3630 Updates: 2370 , September 2003.
23. D. Awduche, L. Berger, D. Gan, T. Li, V. Srinivasan and G.
Swallow “RSVP-TE: Extensions to RSVP for LSP Tunnels”,
Request for Comments: 3209, December 2001.
24. B. Jamoussi, L. Andersson, R. Callon, R. Dantu, L. Wu, P.
Doolan, T. Worster, N. Feldman, A. Fredette, M. Girish, E.
Gray, J. Heinanen, T. Kilty and A. Malis “Constraint-Based
LSP Setup using LDP”, Request for Comments: 3212, January
2002.
25. Report of “US/EU Workshop on Key Issues and Grand
Challenges in Optical Networking” Brussels June 27-28, 2005
(http://networks.cs.ucdavis.edu/%7Emukherje/US-EU-wksp-
June05.html).

View publication stats