Towards the Integration of Multi-Attribute Optimization and Game Theory for Border Security Patrolling Strategies

Border security is a key element of national security policy for any sovereign nation. In the United States, the Border Patrol deploys thousands of agents integrated with technology (e.g., vehicles, cameras, sensors) and infrastructure (e.g., fences, checkpoints) to prevent illegal entry of people and goods into the country along vast land borders with Canada and Mexico. The problem of border security is incredibly complex, due to the diversity and volume of illegal activity that must be controlled, the variety of resources that can be deployed to secure the border, and the differences in circumstances, strategies, and tactics in different sections of the border. Our goal is to develop decision support tools to assist with intelligent allocation of resource in border security. Several recent lines of work have applied game-theoretic and adversarial reasoning to security domains, including deployed transportation security systems (Jain et al. 2010) and robot patrolling strategies (Gatti 2008; Elmaliach, Agmon, and Kaminka 2009). Our approach takes this adversarial reasoning framework and incorporates elements of multiobjective optimization. Game theory and other adversarial reasoning methods typically assume a single, well-known objective function for each player. However, in complex domains like border security decision-makers must carefully weigh the impact of decisions on different objectives. For example, should the defensive posture emphasize narcotics smuggling, illegal immigration, or terrorist threats? How should the impact of patrolling activities on the community and other costs be balanced against the benefits of law enforcement? By developing solution techniques that integrate multi-objective optimization and game-theoretic analysis we hope to provide more powerful tools to decision makers to explore the space of alternative solutions and their impact on various objectives.