Solving large-scale real-world telecommunication problems using a grid-based genetic algorithm

Francisco Luna (1) , Antonio Jesús Nebro (1) , Enrique Alba (1) , Juan José Durillo (1)

1 Computer Science

This paper analyzes the use of a grid-based genetic algorithm (GrEA) to solve a real-world instance of a problem from the telecommunication domain. The problem, known as automatic frequency planning (AFP), is used in GSM networks (Global System for Mobile communications) to assign a number of fixed frequencies to a set of GSM transceivers located in the antennae of a cellular phone network. Real data instances of the AFP are very difficult to solve due to the NP-hard nature of the problem, so combining grid computing and metaheuristics comes out as a way to provide satisfactory solutions in a reasonable amount of time. GrEA has been deployed on a grid with up to 300 processors to solve an AFP instance of 2,612 transceivers. The results not only show that significant running times reductions are achieved, but that the search capability of GrEA outperforms clearly that of the equivalent non-grid algorithm.

Mots clés: Physical Sciences

Francisco Luna, Antonio Jesús Nebro, Enrique Alba, Juan José Durillo. Solving large-scale real-world telecommunication problems using a grid-based genetic algorithm. Engineering Optimization, 2008, 40 (11), pp.1067-1084. ⟨10.1080/03052150802294581⟩. ⟨hal-00545359⟩