Distributed Interactive Simulation (DIS) provides a means to create realistic, complex, virtual environments by linking together simulations of various types at multiple locations, and structuring the information exchange between the entities that comprise the simulation. The DIS methodology has demonstrated significant potential for creating virtual worlds that can simulate highly interactive activities. However, as the DIS architecture becomes more complex to support demands for more realistic simulation environments, the data transfer and processing capabilities of the supporting networks and participating simulations become limiting factors in constructing seamless, virtual environments. This document proposes a framework for a Dual Agent Architecture that enables large scale exercises without creating excessive demands upon network or the participating entities. Three key features combine to distinguish the proposed Dual Agent Architecture from other approaches. First is the notion of a fidelity hierarchy and how it can be employed to structure traffic on a network. Second is the use of multiple agent types to reduce the complexity of both the source and receiver of PDU traffic. And finally, the use of efficient multicast protocols and routers to minimize network traffic. The Dual Agent Architecture provides for two agent classes, the local agent and the regional agent The local agents function as gateways for simulators on the network, implementing the appropriate multicast protocols and creating fidelity based pdu's on behalf of each client simulator. Regional monitor the information exchange between entities located on the simulated exercise play field and serve as information clearing houses to ensure that the information needs of their client entities are met.
Kelley, Barry L., "A Dual Agent Approach To Distributed Interactive Simulation Complexity" (1998). Link Foundation Modeling, Simulation and Training Fellowship Reports. 53.