An experimental testbed for assessing consistency in the distributed simulation of spatiotemporal environments

The enhancements of synthetic environments have led to the effective use of behavioral reasoning models and physics-based models used in the simulation of autonomous bodies in distributed simulation. The combat simulation trainers have used synthetic environments to improve training effectiveness. The synthetic environments have been recently extended to support dynamic changes to the environment database. The sharing and changing of dynamic databases in a distributed environment has led to interoperability problems among networked simulators.; The lack of mechanisms to address consistency of dynamic environments is one of the causes of simulator interoperability problems. Achieving strict consistency implies maintaining global causal ordering of all of the events and changes to the databases. Problems in maintaining consistency occur because each simulator in the distributed simulation is allowed to make changes to the database without imposing restrictions to maintain causal ordering. The differences in the local clocks and the communication latencies between simulators create the possibility of inconsistency. This research addresses consistency at an event level that is close to the architectural constructs of the distributed simulation. Spatio-temporal environments that have a temporal and spatial dependence generally reflect the properties of dynamic environments and are more closely analyzed in this research.; This research examines the theoretical developments in parallel and distributed simulation on clocks and timing of events in order to address consistency of the distributed simulation for spatio-temporal environments. This research helps in developing an experimental testbed and new mechanisms to address consistency. Metamodels are used to represent abstract behaviors that would be found in combat simulations and are used in building a simulation testbed. The performance measures necessary to measure consistency have not been well-defined by previous researchers. This research applies several measures of consistency in the experimental testbed without employing the visual displays of the simulators. The simulation testbed is used in a designed experiment. The resulting regression models demonstrate how the testbed can be used to gain insights into factor settings that should lead to the design and development of simulation architectures for the distributed simulation with improved consistency.; Specifically, results obtained from the experiment indicate that with a little amount of lookaheads and limited time-stamp ordering of messages, consistency can be improved among distributed simulators.