COOPERATIVE CONTROL STRATEGIES AND DECEPTION IN ADVERSARIAL SYSTEMS

of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy COOPERATIVE CONTROL STRATEGIES AND DECEPTION IN ADVERSARIAL SYSTEMS By Zachariah E. Fuchs August 2012 Chair: Pramod P. Khargonekar Major: Electrical and Computer Engineering The protection of vulnerable, high-value assets has been a challenge throughout history. These high-value targets may be fixed, mobile, or in the cloud. In any case, it is necessary to deploy and effectively utilize defensive assets in an attempt to neutralize an attack if it occurs, or make the prospect of further engagement so unappealing that the attackers stand down and retreat. My research focused on two aspects of applying game theoretic methods to the development of defensive strategies. The first aspect of my research utilized tools from differential game theory to develop cooperative, defensive control strategies against a superior, mobile attacker. It was shown that through cooperation, the defending agents can combine their resources and make engagement more costly to the attacker than if they acted independently. A hierarchical attack-retreat game was then developed that incorporates the concept of player intent. This game demonstrated that it is sometimes optimal for the defending player to cooperate with the attacker in order to encourage retreat. Although cooperation can encourage retreat, cooperating at the wrong time can actually assist an attacker in engagement. For this reason, a modified differential game with a value function constraint was developed to prevent the attacker from moving the state of the system to a region where engagement is optimal. The second aspect of my research focused on developing an analytic framework to quantitatively capture and describe concepts of deception. A generic two-player,