Network flow algorithms for routing in networks with wavelength division multiplexing

Abstract

Wavelength-division multiplexing adds one more dimension to routing in telecommunication networks through identification of origin-destination (OD) pairs by wavelength assignment. However, only partial logical direct connectivity is possible owing to the availability of a limited number of wavelengths. One of the existing solutions to this problem is based on fixed wavelength assignment and multihop OD pair connections. The resulting network model involves a multiplicity of traffic flows corresponding to the number of wavelengths used, with simultaneous pairwise OD traffic requirements. This multicommodity network problem is well known in combinatorial optimisation as a hard problem in terms of its computational complexity. A two-stage heuristic approach is investigated using existing network flow algorithms to achieve near-optimal network capacity allocation through consideration of the number of wavelengths, the topology and the possible traffic pattern in a given network. The resulting computations illustrate that the wavelength assignment problem at the first stage depends highly on the pattern of projected traffic and that more complex routing algorithms than shortest-path produced routings with a lower number of wavelength translations and more uniformly distributed traffic loads.