Monte Carlo techniques for design of wavelength-routed all-optical networks

We address the problem of routing and wavelength assignment (RWA) in wavelength division multiplexed networks without wavelength converters. Our objective is to minimize the number of wavelengths required to establish a prescribed list of point-to-point connections across a given optical topology. We propose a class of stochastic Monte Carlo heuristics and compare their performance with two deterministic algorithms reported in previous publications. The Monte Carlo algorithm outperforms its competitors and gives surprisingly good results when compared to precomputed wavelength lower bounds. It is a very simple technique, and provides compelling evidence that the stochastic ingredient should be part of a good RCA heuristic.