Multipurpose Genetic Algorithm Based in Pareto Optimality to Design Logical Topologies in Reconfigurable WDM Networks

Abstract

We propose a novel multipurpose genetic algorithm based in Pareto optimality to design logical topologies for wavelength-routed optical networks with the aim of minimizing both the global average end-to-end delay and the average end-to-end delay of the most delayed traffic flow. The algorithm determines which nodes should be connected by means of lightpaths, solves the routing and wavelength assignment problem, and routes the traffic in the logical topology. When compared with another multipurpose genetic algorithm based in random selections of two fitness functions (one for each parameter to be optimized), the novel algorithm leads to reductions of up to 5% in both parameters