Parallelizing PBIL for Solving a Real-World Frequency Assignment Problem in GSM Networks

Frequency planning (also known as frequency assignment problem -FAP-) is a very important task for current GSM operators. The problem consists in trying to minimize the number of interferences caused when a limited number of frequencies has to be assigned to a quite high number of transceivers. In this work we focus on solving this problem for a realistic-sized, real-world GSM network using a parallelized version of the PBIL (population-based incremental learning) algorithm. Therefore, we have parallelized the PBIL algorithm fixed to the FAP problem using cluster computing. The analysis of the results proves that we have reached a double goal: on the one hand, with the parallelized version of the algorithm, its execution time is reduced down to the optimum values; and on the other hand, we prove that using a distributed island model applied to PBIL, the results for the network-planning are better than the ones obtained with the sequential version.