Logical process partitioning in distributed simulation using genetic algorithms

This paper presents a method to partition models in logical processes in the context of distributed simulation. The proposed method uses genetic algorithms to decide on the viability and the partitioning technique most indicated. The input parameters to the genetic algorithm are information about the model (number of elements, communication, arrival and service taxes), and the architecture where the simulation is executed. As result, we have the number of logical processes and their mapping on the distributed environment. Two models were used to evaluate the proposed method.