Photonic Network Communications, 7:1, 17±35, 2004
# 2004 Kluwer Academic Publishers. Manufactured in The Netherlands.
Preferred Link Based Distributed Adaptive Routing in Wavelength
Routed Optical Networks*
S. Dharma Rao, Chava Vijaya Saradhi, C. Siva Ram Murthy*
Department of Computer Science and Engineering, IIT Madras 600036, India
E-mail: dharma@cs.iitm.ernet.in; saradhi@i2r.a-star.edu.sg; murthy@iitm.ernet.in
Received February 14, 2003; Revised and Accepted July 25, 2003
Abstract. As the WDM technology matures and the demand for bandwidth increases, dynamic provisioning of lightpaths at the WDM layer
becomes an important and challenging problem. In this paper, we consider the problem of dynamic routing and wavelength assignment in
wavelength-routed optical networks. The conventional approach to this problem is to select a route from a set of candidate routes, which has a
common wavelength available on all the links of the route. In this paper, we propose a distributed algorithm which selects a route based on the
state of the network (called preferred link approach). In this approach, a route is selected link by link based on a preference value given to each
of the links. We propose three different heuristic functions for calculating the preference of the links, depending on the cost and congestion on
the links. We evaluate our routing algorithm in terms of call acceptance ratio, cost of the path, hop length, and call setup time. Our experimental
results suggest that our algorithm not only out performs the existing methods with respect to average call acceptance ratio, but, also improves the
fairness among different hop connections, which is an important result in the case of WDM optical networks.
Keywords: wavelength division multiplexing, distributed control, adaptive routing, and preferred link approach
1 Introduction
Ever increasing demand for bandwidth for applica-
tions such as remote information access, video-on-
demand, video conferencing, and other multimedia
applications, has led to the development of intelligent
all-optical networks. These networks employ wave-
length division multiplexing (WDM) (data is
transmitted through different wavelengths of an
optical ®ber) and wavelength routing (routing
different wavelengths) [1]. A WDM network consists
of wavelength cross-connects (WXCs) interconnected
by point-to-point ®ber links in an arbitrary mesh
topology. In these networks, a connection is estab-
lished using a lightpath, which is uniquely identi®ed
by a physical route and a wavelength. The require-
ment that the same wavelength must be used on all the
links along the chosen route is known as the
wavelength continuity constraint [1,2]. Two light-
paths can use the same wavelength, if the paths are
link disjoint, or if they use different ®bers on the same
physical link (wavelength reuse). If a lightpath is
established between any two nodes, traf®c between
these nodes can be routed without requiring any
intermediate opto-electrical conversion and buffering.
Traf®c demand can be either static or dynamic. In a
static lightpath establishment (SLE), traf®c demand
between node pairs is known a priori and the goal is to
establish lightpaths so as to optimize a certain
objective function (maximizing single-hop traf®c,
minimizing congestion, minimizing average weighted
hop count, etc.). The nodes together with the set of
lightpaths at the optical layer form a virtual topology
[3].
The dynamic lightpath establishment (DLE) prob-
lem is concerned with establishing lightpaths with an
objective of increasing the average call acceptance
ratio (or equivalently reducing the call blocking
*Author for correspondence.