A Distributed Computing Framework for Parallelization of Coevolution in Multi-sided Conflicts

Abstract

A large number of military simulation systems dealing with warfare and games with coordinated missions can be described as multi-sided conflicts. These systems usually involve groups sharing various alliances and relationships, each pursuing a range of different goals. A coevolutionary approach in modeling the dynamics of such complex systems allows all sides of the conflict to evolve their strategies or courses of action. The coevolutionary approach used in Sheherazade, a multi-sided simulation environment for Stability and Support Operations (SASO), allows each side to evolve strategies in turns against other strategies captured from the other sides. In such systems, the greater the number of plans that are evaluated, the better the final alternatives are likely to be. To improve the speed and efficiency in generating strategies, a distributed computing modular framework based on the coevolutionary approach used in Sheherazade was designed and implemented to provide the following salient features: provide a more natural model of how different sides interact in a conflict, parallelize the generation of strategies by the different sides, and improve performance by utilizing network computing capabilities.