Simulation networking is no longer new or novel. Heterogeneous, multi-fidelity networks have been successfully demonstrated using either proprietary protocols such as SIMNET, or Distributed Interactive Simulation (DIS) protocols. As the technology for simulation networking has matured, it has resolved some major issues. For example, we now have a standard for the exchange of information between networked simulations (EEE-1278-1993). There has been very little work done toward prediction and accurate measurement of simulator network loading, and little significant work has been published concerning the implications of network loading toward the overall network fidelity and the successful transfer of training. Implicit in the underlying structure of the DIS is an assumption that network performance is purely an issue of applying appropriate technology to support a particular set of objectives. However, network loading imposes limitations upon these objectives and it is unclear what effect unexpected network performance has upon meeting a particular set of objectives.

This paper addresses the problem of predicting network loading in a heterogeneous, multi-fidelity simulation network. It discusses the issues associated with heterogeneous networks and multi-fidelity simulation. Using objective data obtained from a variety of networked exercises (both DIS and non-DIS) for context, this paper discusses the detailed issues involved in measuring network loading. Finally, it makes some recommendations for the future.