Parallel processing implementation at graph coloring to build software of courses timetabling

Abstract

Many factor that affected schedule arrangement made the problem became more complicated. Searching of timetabling solution using optimization process is a real example for a complex computation. One of the ways to solve the problem of computation complexity is changing sequential computation paradigm into parallel computing ones. This research aimed to analyze and to apply parallel processing for building a courses timetabling software for the Department of Electrical Engineering, State Islamic University of Sultan Syarif Kasim Riau. The research was begun with the description of course timetabling problem in the department. Some of solution prerequisite for hard and soft constraints has been determined. Afterward, the reconstruction of conflict course graph was done. The process was done simultaneously with the process of rooms to be assembled, graph that was formed has included its room candidate. Then, conflict course graph was colored to become non-conflict course graph. The coloring task was calculated by parallel computing using multithread classes in Java programming language. The results were achieved in this research are the process of graph to be formed in an adjacency matrix, and coloring of the graph was done successfully both in sequential and multithread computations. Solving the process into two threads, the execution time become faster than sequential ones with the speed up is 3.14 times, and the efficiency of processor is 78.51%.