METAHEURISTICS VERSUS SPECTRAL AND MULTILEVEL METHODS APPLIED ON AN AIR TRAFFIC CONTROL PROBLEM

摘要

In this paper, different metaheuristics applied on an air traffic control problem. This problem is a graph partitioning problem. It can be solved by classical methods which are spectral and multilevel methods. State-of-the-art public-domain graph partitioning packages, CHACO and METIS are used to resolve it. A comparison between results return by these packages and metaheuristics implementations is made for different objective functions of the literature. Metaheuristics used are simulated annealing, ant colony and a new one called fusion fission developed in the LOG laboratory. Experimental results show that metaheuristics find better results than classical packages.