A Large Scale, High Resolution Agent-Based Insurgency Model
Abstract
Recent years have seen a large growth in research developed aimed at modeling the intricate social-cultural climate in which we operate. In particular, agent-based models have garnered considerable attention for their ability to simulate heterogeneous actors operating in complex environments. This paper introduces an agent-based insurgency model capable of simulating hundreds of thousands of agents with complex cognitions and behaviors. Unlike previous agent-based models of civil violence, this work includes the use of a hidden Markov process for simulating agent decision-making, the ability to actively or passively aid insurgency or counterinsurgency efforts, and the incorporation of environmental constraints on agent actions. To significantly improve performance and scale the simulation to hundreds of thousands of agents, the distributed processing capability of a graphics processing unit was utilized. With this model, we are able to show several emergent behaviors such as localized outbursts of insurgent activity, dynamics representative of punctuated equilibrium, and tipping points leading to major insurgent activities. In addition, data based on actual events in Ramadi, Iraq in between 2003 and 2008 was generated and correlated with simulation results to demonstrate how these models can portray realistic insurgent environments.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 30, 2013
- Accession Number
- ADA591808
Entities
People
- Bradley Mitchell
- Kenneth Mccullough
- Lucas A. Overbey
- Samuel Yaryan
Organizations
- Naval Information Warfare Systems Command