A memetic algorithm for channel assignment in wireless FDMA systems

A new problem encoding is devised for the minimum span frequency assignment problem in wireless communications networks which is compact and general. Using the new encoding, which reduces search space dramatically over previous problem encodings, an optimization algorithm is developed which combines a genetic algorithm global search with a computationally efficient local search method from the literature. This memetic algorithm is shown to be more effective than six previous approaches in the literature on a suite of established test problems. Further, it shown that the integration of the global search with the local search is important; neither component by itself is nearly as effective.