On the Optimal Interdiction of Transportation Networks

We consider the optimal interdiction problem in transportation networks as a game in which an attacker acts as the player who goes first and, subject to budget constraints, fails nodes (partially or fully) at time zero so as to maximize the total travel time of the mass. A centralized network operator then acts as the player who goes second and, subject to the system's dynamics, routes the mass so as to minimize its total travel time. We prove that the attacker's best action is to find the most consequential nodes and employ his resources to fail them fully, so that the optimal attack is both sparse and binary. We then propose an algorithm to numerically solve the optimal interdiction problem, and demonstrate the utility of our approach through illustrative examples.