Human Behavior Representation in Physical Security Systems Simulation

Abstract

Physical security systems are designed to prevent access to a facility by intruders, detect the presence of intruders, or facilitate the capture of intruders once they are detected. These systems generally include a combination of physical barriers, human guards, and sensor-based detection systems such as video surveillance systems. Because of the complex interactions between guard, intruder, and neutral entities as well as the interactions between these entities and the environment, analysis of these systems is very difficult and is often limited to static ""line of sight"" and ""field of view"" models designed to help with camera placement. Existing simulation-based analysis methodologies include only crude and often hard-coded implementations of human behaviors for the guard, intruders, and neutrals. This limits the analysis capabilities of these systems. In response,this research develops a computational framework that supports realistic computer characters (or agents) that can operate within physical security system simulations. The outputs of these simulations can then be used to analyze the effectiveness of the tested physical security system configurations and to design more effective physical security systems. The proposed computational framework is comprised of three components: a spatial model, a temporal model, and a representation of the application domain. As the spatial model, a conceptual data model named Hierarchical Graph Representation for Scenes (HIGHRES) is developed to formally represent the static features of the environment in a simulation-friendly structure. A Behavior-Intuition Framework for Realistic Agents (ABIRA) is devised as a temporal model to realistically model the decision making activities of the agents. A retail store security system is selected as the sample application domain to demonstrate the capabilities of the proposed framework and furthermore, to validate the behavior emerging from the proposed computational models. The primary contribution of this work is twofold: a generic, extensible computational framework to emulate realistic human decision making and the integrated physical security systems simulation framework. This integrated simulation framework is capable of conducting simulation experiments to analyze the effectiveness of different physical security configurations that are comprised of both the physical security measures themselves and the security policies that manage them.