Optimizing Sensor Allocation against Attackers with Uncertain Intentions: A Worst-Case Regret Minimization Approach

This paper focuses on the optimal allocation of multi-stage attacks with uncertainty in attacker’s intention. We model attack planning problem using a Markov decision process and characterize intention finite set reward functions–each represents type attacker. Based this modeling, we employ paradigm worst-case absolute regret minimization from robust game theory develop mixed-integer linear program (MILP) formulations for solving minimizing sensor strategies two classes attackdefend interactions: one where defender attacker engage zero-sum another they non-zero-sum game. demonstrate effectiveness our algorithm stochastic gridworld example.