Effects of Combat Simulation Variance on Course of Action Development
Abstract
The U.S. Army Research Laboratory (ARL) is exploring the use of combat simulations in the development of military courses of action (COAs). We currently have the capacity to accept an automatically generated COA and produce a set of simulated results to compare using measures of effectiveness (MOEs). In a previous study, we evaluated a prototype COA generator by playing out its recommendations in the combat simulation Modular Semi-Automated Forces (ModSAF). Two large scenarios were played several times within ModSAF running on ARL's high performance computers to establish an empirical distribution of outcomes. An unexpected finding was the high variability, the so-called pure error variance, observed in the ModSAF results. Some variability is expected; but even after transformation to mitigate the instability of a ratio measure, the magnitude was surprising. In this paper we investigate sources of variability within ModSAF's successor, One Semi-Automated Forces (OneSAF), which provides flexibility to the user regarding simulated characteristics for units, terrain, weather, rules of engagement, and others. Changing input parameter settings introduces another form of variability expressed in signal effects formed from possible combinations of those parameters. The direction and magnitude of those effects is explored in consideration of pure error variance for representative cases.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2001
- Accession Number
- ADA468162
Entities
People
- Barry Bodt
- Eric Heilman
- Janet O'may
- Richard Kaste
Organizations
- United States Army Research Laboratory