Ant algorithms for the exam timetabling problem

Abstract

Scheduling exams at universities can be formulated as a combinatorial optimization problem. Basically one has to schedule a certain number of exams in a given number of time periods so that a predetermined objective function is minimized. In particular, the objective function penalizes schedules where students have to write exams in consecutive periods or even in the same period. Ant colony approaches have been demonstrated to be a powerful solution approach for various combinatorial optimization problems. This paper presents two ant colony approaches for the exam timetabling problem, a Max-Min and an ANTCOL approach. Using the Toronto benchmark test cases from the literature, both algorithms arc compared to other timetabling heuristics. Finally, the Max-Min and ANTCOL algorithms are compared using the same set of test cases. In spite of some shortcomings, the ANTCOL approach turned out to be a worthwhile algorithm, among the best currently in use for examination timetabling. © Springer-Verlag Berlin Heideltaerg 2007.