Dynamic routing and wavelength assignment algorithms in wavelength division multiplexed translucent optical networks Yabin Ye a, * , Teck Yoong Chai a , Tee Hiang Cheng a,b , Chao Lu a,b a Institute for Infocomm Research, Unit 230, Innovation Centre, Block 2, 18 Nanyang Drive, Singapore 637723, Singapore b School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore Available online 19 May 2006 Abstract In dynamic wavelength division multiplexed (WDM) translucent optical networks, lightpaths are setup and torn down dynamically, and some of the transceivers in optical nodes could be free. These spare transceivers can also be used for regeneration or wavelength conversion. In order to setup lightpaths in the translucent network, first an adaptive routing algorithm has been proposed, in which K edge-disjoint paths are selected first and then one of these K paths is chosen by considering not only the number of free wavelengths in the fibers and the length of the path, but also the number of free transceivers in the nodes along the path. After a path has been selected for a connection request, an auxiliary graph is generated for this path to assign wavelengths and determine the nodes for regeneration or wavelength conversion. The auxiliary graph for this path is generated by considering the transparent length limitation. If a least-weighted route can be found in the auxiliary graph, it means that the connection request could be successfully set up along this path. Through extensive simulation in typical networks, it is proved that the proposed routing and wavelength assignment algorithm can achieve better blocking performance than other algorithms. Ó 2006 Elsevier B.V. All rights reserved. Keywords: Routing and wavelength assignment; Translucent optical networks; Regeneration; Wavelength division multiplexed 1. Introduction Wavelength division multiplexed (WDM) optical net- work is the most possible candidate for the backbone of future communication networks. If an optical network is so designed that an optical signal can always travel from the source to the destination without regeneration, the net- work is said to be transparent. The advantages of eliminat- ing regeneration are obvious since the expensive regenerators and hence cost of the network can be saved. However, due to various impairments in optical fiber trans- mission (amplified spontaneous emission noise from opti- cal amplifiers, chromatic dispersion, polarization mode dispersion, nonlinearities, etc.) and crosstalk (introduced by optical switches, demultiplexers and multiplexers, opti- cal filters, etc.) in optical nodes, the transmission length of a lightpath is limited. Realistically, a lengthy end-to- end lightpath has to be set up in a multi-hop manner, i.e., the lightpath needs regeneration at some intermediate nodes, dividing the lightpath into two or more fragments such that the length of each fragment is less than the trans- parent length (‘Reach’). This type of optical networks in which each node routes some lightpaths transparently while subject others to go through regeneration is known as translucent optical networks [1]. Extensive works have been done for translucent optical networks design and different approaches have been pro- posed. One approach for a translucent optical network is to divide a large-scale optical network into several islands of optical transparent domains. Within the same island, a lightpath can reach any node without regeneration. For communications across islands, regeneration nodes at the 0140-3664/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.comcom.2006.04.013 * Corresponding author. Tel.: +65 67922485; fax: +65 67959842. E-mail addresses: yye@ntu.edu.sg (Y. Ye), tychai@ntu.edu.sg (T.Y. Chai), ethcheng@ntu.edu.sg (T.H. Cheng), eclu@ntu.edu.sg (C. Lu). www.elsevier.com/locate/comcom Computer Communications 29 (2006) 2975–2984