6 A Collaborative Framework for Distributed Multiobjective Combinatorial Optimization

This paper states a collaborative framework for the distributed multiobjective optimization of combinatorial problems. The proposed framework is completely agnostic to the specific specialized metaheuristic used. Thus, it is able to use different hybrid strategies using two or more metaheuristics in a collaborative fashion. Besides, the designed framework uses a central repository of nondominated solutions. The solutions are further processed in different nodes (machines) and later go back to the central repository. On the other hand, once the metaheuristic has converged to a new solution its quality is checked, and if it is a non-dominated solution then it is stored in the central repository to be used by other nodes (possibly executing a different metaheuristic) as a new starting point. Lastly, we tested the proposed framework using metrics from the specialized literature. Results show a consistent improvement of the Pareto Front as the number of nodes is increased.