Title

High Performance Army Combat Models Operating Within The Open Unified Technical Framework

Keywords

Army modeling and simulation; Forces modeling and simulation; High performance computing; Manycore; Multicore; Open Unified Technical Framework; Parallel and distributed modeling and simulation; WarpIV Kernel

Abstract

The U.S. Research, Development and Engineering Command (RDECOM) Army Research Laboratory (ARL) Simulation and Training Technology Center (STTC) is exploring the use of parallel and distributed computing within the Open Unified Technical Framework (OpenUTF) for future Army M&S programs. An experimental effort was funded in late 2010 to study modeling issues, scalability, and overall execution speed for core models that are necessary to represent ground combat scenarios. The core models researched, developed, and prototyped in this effort were: (1) representation of ground terrain, geographical locations, and logistical pathway links between location; (2) singleobjective and multiple-objective graph representations and single-objective routing algorithms; (3) troop (platoon) motion along pathway links represented within the battlefield; (4) combat weapon-fire engagement models; and, (5) rule-based and emotion-based cognitive thought processes used to represent human decision-making during combat engagements. The goal is to examine the performance of parallel and distributed simulations executing representative Army engagement models when operating on modern high performance computing facilities such as the STOKES IBMbased high performance computing facility at the University of Central Florida. The initial prototype effort represented the battlefield as a rectangular theater with artificially generated terrain. A graph describing geographical locations (as graph nodes) and potential movement costs along pathways connecting them (as graph edges) was overlaid on top of the terrain. A user-specified number of red and blue troops (platoons) were placed at various locations on the battlefield and then given, as their mission, to "take" one or more different locations. To accomplish their mission, each troop moved from location to location along the recommended routes specified by the routing algorithm. Red and blue troops stopped movement to fight when their paths crossed. The engagement model exchanged weapon-fire based on the combat posture recommended by the cognitive model. Each troop maintained its own internal cognitive model so decisions were based on knowledge of self-state, perception of the enemy, non-organic abiotic environmental features, and performance characteristics of the individual troop. Factors affecting the fighting posture included: skill, saltiness, number of healthy soldiers, number of weapons and ammunition available, value of critical assets or supplies being transported, etc. This study examined several run time performance factors for measuring scalability. The number of troops was varied to determine scenario scalability as a function of entity size. The simulation was benchmarked under a variety of shared-memory and message-based computing architectures. A Phase II effort is being planned to enhance each of these models to provide more realism, optimize performance, and provide better visualization of the scenario as it unfolds over time. The goal is to demonstrate feasibility of realistic Army models operating within the OpenUTF.

Publication Date

12-1-2011

Publication Title

Fall Simulation Interoperability Workshop 2011, 2011 Fall SIW

Number of Pages

430-440

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

Socpus ID

84865500025 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/84865500025

This document is currently not available here.

Share

COinS