Principal Investigator: Milind Tambe
Abstract:
Our research has been at the forefront of applying computational game theory techniques for security. It has led to a wide range of actual deployed applications of game theory for security. Early transition successes of our work included ARMOR at LAX airport for scheduling checkpoints and canine patrols, IRIS for the US Federal Air Marshal Service to deploy air marshals on US air carriers, GUARDS developed for the TSA for randomizing security activities to protect airport infrastructure, PROTECT for the US Coast Guard to schedule randomized patrols within ports, deploying escort boats for the US Coast Guard to protect ferries, and TRUSTS for the Los Angeles Sheriffs Department to schedule multi-operation patrolling (fare evasion, counter-terrorism and crime) on LA area metro trains. These systems focus on the game-theoretic method of providing efficient randomization of security plans and processes. Casting the problem as a Bayesian Stackelberg game, they obtain randomized strategies for security agents; one of the fundamental advances in all these systems then is to provide the fastest algorithms known-to-date to solve such games. The strength of this research is the marriage of strong theoretical game-theoretic foundations with practical applications, and the virtuous cycle of theory and practice to benefit from each other. These deployments of applications in the real-world have led to significant interest from the media and other potential users/customers, and substantial research. We have continued applying our game theoretic techniques to new counterterrorism domains which introduce novel research challenges. For example, we are now working on a project known as DARMS with the TSA for dynamic passenger screening at airports. While the DARMS application is specific to airport screening, the underlying game model that has been developed has broad ramifications and is applicable to any domain where a security agency has to strategically process, i.e., screen, a sequence of either people or objects in order to identify potential threats without unduly impeding the flow of those people or objects. Beyond counterterrorism, we have also started to extend our techniques and applications into everyday security domains such as fish and wildlife conservation and urban crime prevention. These everyday security applications require new adversary models for an adversary who is less strategic in planning and more flexible in execution. For fish and wildlife conservation, we have been working on an application called ARMOR-FISH that is in use by the United States Coast Guard to help prevent illegal, unreported, and unregulated fishing. By understanding the distribution of critical fish stocks as well as the way in which illegal fishermen make decisions, we can then optimize aircraft patrols to mitigate illegal fishing and ensure the economic and ecological stability of commercial fisheries. For urban crime, we are working with the University of Southern California Department of Public Safety to schedule patrols to help reduce crime around the university campus. A challenge there is being able to predict the levels of future crime based on previous levels of crime as well as previous levels of patrols.
SOW: