Computational Psycho-Social Models of Terrorist Threat

Not Available With the emergence of the Internet, information systems are increasingly vulnerable to cyber warfare. Existing approaches to information systems security focus on component level vulnerabilities that are already known to exist. What is needed is an approach that targets strategic vulnerabilities at a system-wide level. In order that such an approach be applicable to attacks that may be devised in the future, adversaries need to be modeled in a relatively generic fashion. Further, the approach should be capable of suggesting countermeasures and of providing the system security officer with a visualization of the tradeoffs. The central contribution of the proposed work will be the development of a tool, called Mitigator, that will model adversary threats and suggest countermeasures to offset identified vulnerabilities. It will help the system designer and security administrator to predict where red teams will attack the system and to handle the tradeoff between risk, performance, cost and functionality. To provide these capabilities, we propose a market-based multi-agent approach based on time-tested game theoretic concepts from economics. We will demonstrate the effectiveness of this approach by applying it to DARPA's Advanced Logistics Program (ALP).