On the Use of Genetic Algorithms to Fine-tune OSPF Routing

This paper revisits the problem of Traffic Engineering (TE) to evaluate the performance improvements achieved by destination-based TE when OSPF routing is fine-tuned to achieve route optimisation. We build upon the framework provided by evolutionary algorithms to present a TE model where link weights are computed by a genetic algorithm and used as fine-tuning parameters in destination-based path selection. The TE model is based on a hybrid scheme combing offline estimation of the link weights and online path selection. We apply the newly proposed TE model to compute the routing paths for the traffic offered to a 23-node test network. We compare the correlation between the link weights found by the genetic algorithm and the recently proposed Least Path Interference algorithm (LPIA) and evaluate the impact of this correlation on the network performance. We evaluate the routing performance of the TE model using the NS packet level simulator.