Gonzalo Joya, Departamento de Tecnología Electrónica, University of Malaga, Spain
Francisco Sandoval, , Departamento de Tecnología Electrónica, University of Malaga, Spain
In this work we propose a set of modifications to the AntNet routing algorithm
developed by Di Caro, G., and Dorigo, M., and its application to the evaluation and
management of a telecommunication network. AntNet is a distributed and adaptive
routing algorithm, based on mobile agents (Ant Colony paradigm) which indirectly
interact through changes in their environment (stimergy). In order to obtain a higher level
of performance we briefly suggest the following modifications: A non-rigid ant
generation is proposed. We reach this dynamism by imposing a maximum number of ants
in the network at any moment. Under that limit, the generation rate will be given by the
mean delay time of the packets. Traffic introduction, as the possibility of the ant to wait a
threshold time to access a node, so if that time is exceeded, the ant returns to the source
node, only updating the probabilities (reflect how good is that node for a certain path) of
the nodes passed to that moment. The nodes not updated are then, in an indirect way,
penalized. As to the application of the routing algorithm to a real world problem the
possible functions will be: a) a pre-simulation of a network with the purpose of estimate
its viability; b) evaluation and management of an existing network detecting and
correcting possible anomalies, obtaining a more balanced traffic.