Reconfiguration Heuristics for Logical Topologies in Wide-Area WDM Networks

Hironao TAKAGI  Yongbing ZHANG  Hideaki TAKAGI  

IEICE TRANSACTIONS on Communications   Vol.E89-B   No.7   pp.1994-2001
Publication Date: 2006/07/01
Online ISSN: 1745-1345
DOI: 10.1093/ietcom/e89-b.7.1994
Print ISSN: 0916-8516
Type of Manuscript: PAPER
Category: Fiber-Optic Transmission for Communications
lightpath,  logical topology,  reconfiguration,  wavelength-division multiplexing,  optical networks,  

Full Text: PDF(351.3KB)>>
Buy this Article

Wavelength division multiplexing (WDM) technology offers the capability of building wide-area networks with high speed. Reconfigurability is a key feature of a WDM network that enables the network logical topology to change dynamically in response to the changing traffic patterns. There are two important issues involved in the reconfiguration of a network logical topology. One is how to determine the new logical topology corresponding to the current topology. It needs to consider a trade-off between the performance of the new target topology and the cost of the topology transition from the current topology to the new one. The other is how to determine the transition sequence from the current topology to the new one. It needs to control the disruption to the network as less as possible during the reconfiguration process. In this paper, we focus on the latter problem and propose several heuristic algorithms that reconfigure logical topologies in wide-area wavelength-routed optical networks. Our reconfiguration algorithms attempt to control the disruption to the network as less as possible during the reconfiguration process. For this purpose, a lightpath is taken as the minimum reconfiguration unit. The proposed algorithms are evaluated by using an NFSNET-like network model with 16 nodes and 25 links. The results show that very simple algorithms provide very small computational complexity but poor performance, i.e., large network disruption, and that an efficient algorithm provides reasonable computational complexity and very good performance. More complex algorithms may improve performance somewhat further but have unrealistically large computational complexity.