The U.S. is increasingly dependent upon a complex framework of interdependent networks and systems that provides a continual flow of goods and services essential to the defense and economic security of the United States.

The interdependencies among infrastructure elements create vulnerabilities that are different, and potentially more damaging, than those found when considering individual infrastructures. These must be understood and then managed to minimize the impact of deliberate disruptions, human error, system complexity, or natural disasters on the surety of the U.S. infrastructure system. Although there are numerous models of individual infrastructure elements (e.g. telecommunications, transportation, electrical power systems) that can be used to predict the consequences of potential disruptions within an individual infrastructure element, there are currently no models available of the complex interdependencies that exist among the individual infrastructure elements. Sandia can provide modeling and analysis of interdependencies through application of its outstanding capability in simulation of complex systems. Sandia's modeling expertise provides a much needed capability to support mapping of critical nodes in the infrastructure system and a method to quantify the physical and economic consequences of threats to the security of the national infrastructure system.

The world's fastest computers, capable of running detailed simulations of a variety of complex systems, are now operating at Sandia.

Efforts are currently underway to develop computer simulation tools to predict, in real time, the consequences of disruptive events on the nation's critical infrastructures. The modeling approach utilizes an agent-based methodology to predict critical infrastructure interactions. This simulation technology capitalizes on recent technological advances in evolutionary learning algorithms and massively parallel computing. Interactions among infrastructure elements are modeled individually by smart agents, one for each interaction. This modeling protocol can utilize thousands of agents to model very complex systems and offers several advantages over traditional modeling techniques for modeling disruptions or shocks to interdependent infrastructure systems. For example, unlike analytic models, functional forms of the model's endogenous relationships are not required. For problems where macro-scale information is sparse or non-existent, as is the case for infrastructure interdependencies, agent-based models can utilize existing rich sources of micro-level data to develop interaction forecasts. The product being developed relies upon development of improved analytic modeling methods, dynamic object-oriented programming, improved visualization, and highly scalable simulation modeling methods that will leverage Sandia's high performance computing capability and will produce meaningful systems-level models.

Questions and Comments || Acknowledgment and Disclaimer