A Comparison of Several Evolutionary Heuristics for the Frequency Assignment Problem

The Frequency Assignment Problem (FAP) is a very important problem of practical interest. This work compares several evolutionary approaches to this problem, based both in the forma analysis framework, and in the decoder paradigm. All approaches are studied from the point of view of two different quality measures of assignments: the number of distinct frequencies, and the frequency span. It is shown that using decoders as embedded heuristics is more adequate than performing a direct search in the feasible solution space. Furthermore, despite the apparent symmetry of the problem, a recombination operator based on multiple-emitter-to-frequency preservation performs better than focusing on multiple-frequency-to-emitter preservation.