Distributed Interactive Simulation (DIS) is both a protocol for networking simulators and a style for implementing Department of Defense training simulator networks. Follows-on to DIS such as ADS (Advanced Distributed Simulation) or other architectures that adopt the style will have to overcome the same networking and communication problems now being faced by DIS. This paper discusses these problems in an abstract, mathematical context so that the conclusions will be applicable to DIS and its heirs.
DIS is defined as a formal system which can be extended to cover any conceivable simulator networking architecture. This formalism is used to prove several theorems about the computational capability of formal DIS networks. These theorems demonstrate that quite simple DIS networks can attain computational universality. Of significance for the DIS standards process is the result that the specification of protocols for DIS networks of Turing machines can only approach completion asymptotically. Limitations are also found in the ability to validate a simulation as a result of mismatches in simulation host speeds.
When DIS networks simulate physical processes, a relationship between the maximum simulated physical velocities of entities and the propagation delays is derived. Another relationship is found between the network propagation delays and the maximum rate of state change of network entities. These maxima are only a problem for networks with very fast movers or with very long propagation delays.
To keep the presentation suitable for a general audience, the formalism and the conclusions are presented in the form of a Socratic dialog. The talk will consist of the dialog as staged by the two authors. One will play the inventor of DIS, Thorpus, and the other his skeptical critic, Skepticus.