Case Study: The University Course Timetabling Problem

: Introduction

The University Course Timetabling Problem (UCTP) is a combinatorial optimization problem [1] faced by every university of the world. The basic definition of the UCTP states that a number of university courses have to be assigned within the time slots of the university time table with a number of additional preferences being satisfied are maximized. There are also many other time table problems which are similar to the UCTP problem like examination time table, school time table, employee time table, hospital time table, etc. These all problems are similarly hard to solve. The general UCTP problem is an NP-hard problem because there are many sub-problems associated with it with additional constraints. Since organization of courses

The problem of UCTP has a given set of n events E = {e1, . . . ,en}which has to be scheduled in a given set of timeslots T={t1 , . . . , tk} (k = 42, 6 days of 7 timeslots each), a given set of certain capacity rooms R = {r1, . . . , r|R|}in which all events can occur, a set of students S who have to attend the particular events, and a set of features F which are satisfied by the rooms and are required by the events. Each student is assigned to a list of events by the university. A timetable is feasible if all of the events are assigned to a particular timeslot and to a particular room while satisfying all of the following hard

Rediscovery of ILS is done by many authors and they has given many different names for ILS like iterated descent, large-step Markov chains, iterated Lin- Kernighan, chained local optimization, and combinations of these. The ILS metaheuristic relies upon the multiple runs of a given local search algorithm iteratively. The major difference between ILS and RRLS (Random Restart Local search) is in choosing the starting points; i.e. in RRLS, the starting point is chosen randomly but in ILS, starting point is chosen based upon the previous best solution found. A perturbation has been applied to the previous best solution which gives a new solution which has to be improved again by the use of local search procedure. The performance of the ILS algorithm depends very significantly on the perturbation