Modeling Human Decision Making and Agent-Based Modeling of C3 Architectures in Warfare Assessment Models
Navy SBIR 2007.3 - Topic N07-196
SPAWAR - Ms. Linda Whittington - email@example.com
Opens: August 20, 2007 - Closes: September 19, 2007
N07-196 TITLE: Modeling Human Decision Making and Agent-Based Modeling of C3 Architectures in Warfare Assessment Models
TECHNOLOGY AREAS: Information Systems, Battlespace, Human Systems
The technology within this topic is restricted under the International Traffic in Arms Regulation (ITAR), which controls the export and import of defense-related material and services. Offerors must disclose any proposed use of foreign nationals, their country of origin, and what tasks each would accomplish in the statement of work in accordance with section 3.5.b.(7) of the solicitation.
OBJECTIVE: To improve the representation of human decision making in constructive campaign-level warfare assessment models. To integrate agent-based modeling and simulation frameworks with warfare assessment models to provide capabilities for sophisticated C3 behavior modeling.
DESCRIPTION: Current traditional warfare assessment models have been criticized for their simplistic modeling of command and control behavior, resulting in often inappropriate reactions to complex tactical situations. These models typically employ rule-based decision trees to represent military doctrinal decision making and rules of engagement. While this approach is both repeatable and traceable in model-based analysis (a plus) the palette of triggers and reactions must be sufficiently rich to accurately capture behavior that appears "human". This is where many models fall short. Updating the rules by which simulated military commanders make decisions requires a thorough understanding of the various triggers, reactions, and caveats a tactical situation presents. These can vary by warfare area, by level of command, by degree of training, and by threat condition/tactical situation. Agent-based modeling and simulation systems developed using concepts from genetic algorithms, the game of Life as developed by John Conway, and other artificial intelligence techniques, implemented in frameworks such as SWARM, Ascape, and RePast, have been identified as high-potential techniques that can be effectively used to address the perceived deficiencies in C3 modeling.
Beyond improving the implementation of rule-based military decision making, human behavior in campaign level warfare assessment models should be upgraded to allow for more complex reactions such as recognizing enemy intent from intelligence and sensor reports, learning from past experiences, deciding to withdraw after a certain threshold of own-force losses is sustained, and making intelligent allocations of assets to task in response to a perceived group of threats of a certain composition. Currently these complex decisions are typically scripted into models, or a model is run with human decision makers in-the-loop. Both these approaches have drawbacks and do not allow for robust and seamless constructive modeling analysis.
Recent work has results in several middleware solutions that can be used as a means for integration. Frameworks such as the System for Parallel Agent Discrete Event Simulation (SPADES) provide potential solutions towards integrating agent based models and simulation, and traditional warfare models. This effort would address the identification and further development of such integrating frameworks.
PHASE I: Using the OPNAV warfare assessment scenarios as a baseline context, identify current deficiencies in modeling tactical decision making and propose improvements drawing from military doctrine, tactics, and current rules of engagement. Propose how doctrine and tactics might change given future systems and FORCEnet improvements that might exist in the timeframe of the analysis. Explain how these improvements might change modeling results and provide a better assessment tool for OPNAV QDR analysis.
PHASE II: Building on deficiencies identified but not addressed in Phase I, design and implement advanced command and control/ decision making logic into a campaign-level warfare assessment tool. Explain how these upgrades would address key failings in the current OPNAV assessments. Use the new features in a study to demonstrate the improved capability.
PHASE III: Explore alternative methods for modeling decision making in constructive simulation, to include tactical algorithms, value-driven methods for Course of Action analysis, optimization (genetic algorithms, linear programming), and agent-based representations. Propose which approach is appropriate for certain decisions, commanders, or situations. Base these recommendations on published military doctrine/tactics/rules of engagement, interviews with military commanders, and exercise observations. Design and implement these features, and collect the required data. Demonstrate the new features in an OPNAV-level assessment and collect subject matter expert reviews of the new representation of commander behavior.
PRIVATE SECTOR COMMERCIAL POTENTIAL.DUAL-USE APPLICATIONS: The resulting product would provide valuable insight into the complex field of modeling human decision making, which would be of use to many DoD modeling and simulation tools. Further application in the area of automating commercial processed through the use of artificial intelligence based on this research would also have potential.
REFERENCES: 1. An Assessment of the Current State-of-the-Art in Modeling Command and Control Processes and Systems: A Survey of Current and Planned Models with Recommendations for Future R&D, William K. Stevens and Colleen M. Gagnon, prepared for The Advanced Concepts, Technologies, and Information Strategies (ACTIS) Directorate, National Defense University (NDU), 30 September 1997.
2. Modeling Human and Organizational Behavior: Application to Military Simulation, Richard W. Pew and Anne S. Mavor, National Academy Press, Washington, DC, 1998.
3. SPADES --- A Distributed Agent Simulation Environment with Software-in-the-Loop Execution, Patrick Riley and George Riley. In Winter Simulation Conference Proceedings, pp. 817–825, 2003.
KEYWORDS: Human Decision making; Campaign level modeling; simulation; tactics; C4ISR; ABM; Agents; Assessment