Hopfield neural networks for routing in all-optical networks

This paper proposes a routing algorithm based on Hopfield Neural Networks (HNN) for all optical networks. To our knowledge, it is the first RWA algorithm based on HNN that considers the physical layer impairments. The parameter selection is carried out by a Multi Objective Particle Swarm Optimization using crowding distance and roulette wheel and considers as performance criteria the blocking probability of the optical network and the number of iterations employed by the HNN algorithm to find the lightpath. The HNN algorithm presented better results than other well known algorithms, with a low number of iterations for the HNN convergence and a low blocking probability.