A Graph Partitioning Game for Distributed Simulation of Networks

Distributed simulation of large and complex networks requires equitable partitioning of the network model. We propose an iterative partitioning scheme for distributed network simulation that is based on a game theoretic model of simulated network nodes acting as players. We first model the cost function for the nodes and prove that a Nash equilibrium exists for the non-cooperative game in pure strategies. We then propose an iterative algorithm based on our model and prove that it converges to one of the many Nash equilibria. We show with the help of simulations that such a design of cost function moves the system towards optimum partition of the network model as the algorithm converges. keywords: parallel network simulation, game theory, graph partitioning, Nash equilibrium.