Handling Dynamic Networks Using Evolution in Ant-Colony Optimization

Abstract

Ant Colony Optimization (ACO) is a popular meta-heuristic for solving combinatorial optimization problems. ACO uses the concept of ants foraging for food to find good solutions to these types of problems. ACO has been successfully applied to many problems, from the traveling salesman problem (TSP), to the problem of network routing. However, it has been pointed out that ACO does not perform as well as other heuristics in very dynamic problems. At first, this statement seems strange but a close look reveals that the nature strategy that inspires the ACO meta-heuristic has an important element that is lacking in ACO: evolution. This paper proposes a new algorithm, named Evolutionary Ant Colony Optimization (EACO), that combines ACO with elements of traditional Genetic Algorithms (GA), namely: selection, recombination, and mutation. Individual ants are endowed with a genotype that is allowed to evolve through generations of the population. In doing this, the EACO algorithm adds another element of optimization to the ACO algorithm that allows the individual agents (ants) in the algorithm to improve their behavior over several generations. Our results demonstrate that EACO can indeed overcome the hurdles faced by the original ACO.