Parallel Artificial Bee Colony Optimisation for Solving Curricula Time-Tabling Problem
The paper presents parallel computational model for solving the curricula time tabling problem using hybrid metaheuristics approach that combines population based artificial bee colony optimization and trajectory based simulated annealing for local optimization. The suggested solution is targeted for multicomputer high performance architecture and exploits both shared memory and distributed memory parallel computational models utilizing fine grained thread level parallelism and master-slave message passing flat model. The experimental evaluation shows good scalability of the solution quality due to better diversification and local exploration of the search space when increasing the number of processes and threads. The speedup of the parallel computational model also scales almost linearly in respect to both the parallel workload and the machine size. In addition a web based application is developed to ease the schedule construction, editing, visualization and usage in educational institutions.