Using a nature inspired technique to train a dynamic IA-RWA algorithm

In this work we add a training phase to an Impairment Aware Routing and Wavelength Assignment (IA-RWA) algorithm so as to improve its performance. The initial IA-RWA algorithm is a multi-parametric algorithm where a vector of physical impairment parameters is assigned to each link, from which the impairment vectors of candidate lightpaths are calculated. The important issue here is how to combine these impairment parameters into a scalar that would reflect the true transmission quality of a path. The training phase of the proposed IA-RWA algorithm is based on an optimization approach, called Particle Swarm Optimization (PSO), inspired by animal social behavior. The training phase gives the ability to the algorithm to be aware of the physical impairments even though the optical layer is seen as a black box. Our simulation studies show that the performance of the proposed scheme is close to that of algorithms that have explicit knowledge of the optical layer and the physical impairments.