Event-based pheromone modification strategy for Ant Systems applied to dynamic TSP

The aim of this paper is to present an Ants System approach to handle the dynamics of Travelling Salesman Problem (TSP). The dynamics in a TSP instance is caused by changing travel times between cities. To avoid long travelling times a fast reaction to the changes is essential. We propose a special event-based pheromone modification strategy intended to handle these dynamics. With respect to the strength of the events, pheromone information will be modified. The possibility of positive forecasting of a path to be a part of possibly good solutions, supply ants a rapid recognition of alternative routes. Our specialized strategy is compared to three other strategies by an experimental test for a dynamic TSP.