Timetabling at German Secondary Schools: Tabu Search versus Constraint Programming

In recent years many solution approaches for different school timetabling problems (TTP) have been tried, among them tabu search, simulated annealing, genetic algorithms, and constraint programming [8]. However, most of the solution methods developed so far have been tested by means of only few (mostly only a handful) problem instances [6]. Further, until today there are nearly no reports on tests which subject different methods to a comparative analysis on the basis of the same benchmark instances and thus provide empirical evidence for the (relative) suitability of solution approaches. In this paper, a tabu search algorithm [4] for timetabling at German secondary schools of the Gymnasium type is presented. The TSA is subjected to an extensive test including 1500 problem instances. The instances have been introduced by MARTE [6] and used for the test of his constraint programming [5] method. Therefore, the results obtained with the TSA and the CP method are finally compared here.