International Journal of Scientific & Engineering Research, Volume 5, Issue 10, October-2014 ISSN 2229-5518 IJSER © 2014 http://www.ijser.org University Examination Timetabling Using Tabu Search Hamid D. Lawal 1 , Ibrahim A. Adeyanju 2,* , Elijah O. Omidiora 2 , Oladiran T. Arulogun 2 and Oluyinka I. Omotosho 2 Abstract— University Examination Timetabling deals with scheduling examinations for students while satisfying specified constraints. Such constraints include avoiding overlap of courses offered by same students, ensuring fair spread of courses offered by a class and allocating suitable venues. This paper discusses an automated examination timetabling system for universities based on Tabu Search, a meta- heuristic technique. Tabu search starts with random generation of an initial solution which is typically sub-optimal. It then gradually optimises this solution by exploring the search space but avoids unnecessary exploration by keeping a list of recently visited areas in a Tabu list. Three versions of the timetabling system (labelled Sys_A, Sys_B & Sys_C), having varying penalties for violating constraints specified as hard or soft, were evaluated. Sys_A used equal penalty on all constraints, Sys_B used equal penalty on all hard constraints and a lower equal penalty on all soft constraints while Sys_C used different penalties for hard constraints depending on their perceived importance and the lowest penalty on soft constraints. The experimental dataset consisted of 153 courses with varying class sizes for a total of 5550 students to be scheduled within 25 days using 15 venues of different capacities. The data was obtained from the Faculty of Engineering and Technology, LAUTECH, Nigeria. The time taken to generate a timetable within 1000 maximum iterations and weighted relative error of generated timetables were used as evaluation metrics. The least error (best result) was obtained with Sys_B, having equal penalty on all hard constraints and a lower equal penalty on all soft constraints though with a second best (lower) simulation time. Index Terms— Constraints satisfaction, Hard and soft constraints, Scheduling, Tabu search, Timetabling , University examination ——————————  —————————— 1 INTRODUCTION imetabling is a combinatorial problem that commonly arises in higher institutions [1] and essentially concerned with scheduling a certain number of examinations in a given number of time slots in such a way that no student is having more than one examination at a time. Timetabling problems are in the set of NP-hard problems [2] [3]. Assigning examinations to days and timeslots within the day are also subject to constraints. These constraints are divided into two types: hard constraints and soft constraints; and may vary from one institution to another. The hard constraints are the compulsory ones that cannot be violated while soft constraints are necessary to improve the quality of a timetable but not compulsory [4]. The manual solution of the timetabling problem usually requires several days of work and the solution obtained may be unsatisfactory in some respect; therefore, considerable attention has been devoted to automated timetabling [5] [6]. Genetic Algorithm, Simulated Annealing [7] [8], Memetic Algorithm, Tabu Search and Ant Colony Optimisation [9] are among the main approaches for solving the timetabling problem intelligently. In this paper, we propose a simple but effective approach to solving timetabling problem using Tabu search. Section 2 discusses some related work while Section 3 describes the Tabu search approach to Examination Timetabling. Our experimental methodology and discussion of results are given in Sections 4 and 5. The paper is concluded in Section 6 with pointers to preferable extensions to the current work. 2 RELATED WORK Examinations Timetabling Problem (ETP) is the problem of assigning courses to be examined, candidates and examination rooms to time slots while satisfying a set of constraints. It is an important issue in higher institutions and is known to be a highly constraint-based problem [10] [11]. This problem is typically characterized as a constraint satisfaction problem that is complex in nature and very difficult to solve. To overcome this problem, higher institutions need an automated examination timetabling scheduler that is robust, flexible (can accommodate new courses and venues), conflict-free (satisfies virtually all the specified constraints) and generates a time table within few minutes. Ant Colony Optimisation and Simulated Annealing were used by [12] and [13] respectively while [14] [15] proposed Tabu Search as approaches to solving ETP. Though the Tabu search approach is similar to the one used in this paper, hard and soft constraints were not differentiated in previous works. Chu & Fang [16] worked on the Genetic Algorithm (GA) and Tabu Search in timetable scheduling and compared the performance of these two techniques based on the quality of the exam timetable and the time spent in producing the timetable. Their results show that TS produced better timetable than GA, but search time spent in TS is less than that of GA while GA produces several different near optimal solutions simultaneously. T ———————————————— * Corresponding Author: I.A. Adeyanju. Email: iaadeyanju@lautech.edu.ng 1. Department of Computer Science, Federal Training Centre, Ilorin, Kwara State Nigeria. Email: hamidlawal2@gmail.com 2. Department of Computer Science and Engineering, Ladoke Akintola University of Technology, Ogbomoso, Oyo state, Nigeria Emails: {iaadeyanju|eoomidiora|otarulogun|oiomotosho}@lautech.edu.ng