Ant colony optimization routing mechanisms with bandwidth sensing

摘要

The study and understanding of the social behavior of insects has contributed to the definition of some algorithms that are capable of solving several types of optimization problems. In 1997 Di Caro and Dorigo developed the first routing algorithm for wired networks, called AntNet, using an approach which was inspired in the behavior of ant colonies. At each node, AntNet, similar to others Ant Colony Optimization (ACO) based algorithms, forward ants based in the amount of pheromones present in the links and in response to the node's queue lengths. In this paper, an adaptation of the e-DANTE algorithm for discrete problems, as an IP based routing mechanism, was implemented. We also propose the inclusion of a new parameter for the computation of paths for both the AntNet and the newly proposed algorithm: the available bandwith. Those methods were tested in ns-2 using two dense network architectures and their efficiency is compared with the original AntNet and a Link-State routing algorithm, when considering the transmission of competing traffic flows between distinct nodes.