Combat Simulation Trajectory Management.
Abstract
This project tested the feasibility of explicitly tracking multiple outcomes for random events in combat simulations. A simple force on force combat simulation resembling Eagle was built that creates new states and trajectories when selected random events occur, in order to maximize coverage of the simulation outcome space. This multitrajectory simulation concept was implemented in C++ using a class structure that hides the complexity of spawning new trajectories, so that the functional code programmer's task is nearly the same as for stochastic simulation. The challenge is that a random choice method called once must return multiple times, once for each trajectory, if an event is resolved in a multi-trajectory fashion. The prototype simulation was tested for 4 unit and 40 unit scenarios, with a variety of management strategies selected to manage (and ration) trajectory proliferation. The results showed superior information gained for a given commitment of computational resources, compared to stochastic simulation. There remain some software issues in the use of local variables. The project was unable to address scaling issues to larger sized scenarios, which will require more sophisticated management heuristics.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 14, 1996
- Accession Number
- ADA313648
Entities
People
- John B. Gilmer Jr.