Defeating Lag in Network-Distributed Physics Simulations

Shared worlds for distributed games, simulations, and VR/AR rely on intuitively “good-enough” depictions of shared world state currently limited by the belief that it is impossible to maintain identical shared state across a real-time distributed network. This prevents repeatable, verifiable results for decision-making support, safety-related and/or equipment-in-the-loop simulations, or multi-user virtual/augmented reality. A network-distributed simulation architecture is presented which maintains consistent distributed state with known maximum bounds on the duration of transient state divergence due to external input. The key concept is computing simulation events slightly early so that they arrive just-in-time to other nodes in the distributed system. This works for all events where the data resides fully within the simulation model. A procedure is provided to eliminate any state divergence caused by external inputs, which to date has been considered impossible. Providing consistent distributed dynamic shared state enables repeatable, verifiable results for all distributed simulation applications.