Pheromone evaluation in Ant Colony Optimization

In ant colony optimization (ACO) artificial ants communicate by laying synthetic pheromone along the edges on their path through a decision graph. This attracts following ants so that they are likely to search in the same region of the search space. The problem of how pheromone information should be evaluated in ant systems is studied in this paper. The standard approach for pheromone evaluation is that, for every decision, ants use only the local pheromone values corresponding to the possible outcomes of this decision. We show that the optimization behaviour of an ant algorithm can be improved substantially when the ants use more pheromone information than just local pheromone values. This clearly indicates that the problem of pheromone evaluation is an important topic for ant colony optimization that deserves attention.