Uma implementação em paralelo para decomposição de benders aplicada a sistemas eixo raio com múltipla atribuição

Abstract

Hub and Spoke systems, is a important research area in localization theory. This occur, because of these systems are very used in logistics problems, e.g., telecommunication networks and transport of passenger and load.To serve the demand of each pair source destination, basically, the Hub and Spoke system replaces direct connections between the pairs for a hubs network. These hubs group the traffic sharing the transportation medium. To get the best hubs configuration is necessary efficient methods, because this problem, hubs allocation, is a NP-problem. In this work was developed an parallel implementation of the Benders Decomposition method for the uncapacitated multiple allocation hub location problem. In our implementation we use the Skorin-Kapov model. The parallel implementation of Benders Decomposition for hub and spoke problem is not known in literature. The results show that the parallel approach is applicable and more efficient that nonparallel one. The experiments reveals that the parallel algorithm had a time execution 70% minor when compared with the nonparallel one.