While the strategic interaction between a system and an adversary can evolve over time, the use of game theoretic formulation to model such interaction as in economics is misleading. This project attempts to study how to model such interactions in realistic settings. In this work, we develop a “Science of Security” based on control and dynamical systems. We develop a control-theoretic framework that will allow us to characterize the actions and impact of the adversary as well as the system under attack. Using our framework, we develop non-equilibrium dynamics driven adaptive defense.