Building a Science of Cyber-Security Games
Abstract
Securing machines attached to the internet against adversaries is now turning out to be an incredibly important problem for the Army and the nation. There has been recent interest in the Cyber Security community on the use of Game Theory to design strategies (especially address randomization and IP address change) that would increase the cost for adversaries to attack. However, how to design good operational risk models is an open problem. The Offeror will address the question of what is an appropriate operational risk model, and what kinds of real-time strategies can be used to defend in a fast paced environment. Given that the overarching objective of the proposed work is to lay the foundations for an integrated theory and associated models and tools that improve our understanding of interactions between human users and cyber attack surface, the Offeror will build on the emerging science of security games, a subarea of game theory that we have founded, to lead to a new science of cybersecurity games. The Offeror will bring to bear a unique combination of computational and behavioral game theory, as well as machine learning, to advance solutions to solving large scale games that will likely emanate from modeling human behaviors, and the associated uncertainty, vis-a-vis their interaction with computing and cyber systems. In particular the Offeror will collaborate with ARL scientists and complement the work of ARL Cyber Security CRA in validating his solutions.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 12, 2017
- Source ID
- W911NF1510515
Entities
People
- Milind Tambe
Organizations
- Army Contracting Command
- United States Army
- University of Southern California