Multi-objective optimization based on ant colony optimization in grid over optical burst switching networks

A multi-objective task scheduling approach for grid over optical burst switching (GOBS) networks is proposed. It takes into account the selection of both computational resource and network resource and is able to simultaneously satisfy three objectives representing the requirements of grid users and resource providers, namely completion time, payment and load balancing. An ant colony optimization algorithm (ACO) for this multi-objective GOBS optimization problem is designed. The convergence and diversity preserving of the algorithm is compared with the nondominated sorting genetic algorithm (NSGA-II) through three performance metrics. Simulations are carried out to compare grid and network performance of these two algorithms. Grid scheduling performance comparison between single-objective optimization and multi-objective optimization based on ACO is also taken by simulations.