Simulating Network Robustness for Critical Infrastructure Networks

We examine the robustness of critical infrastructure networks in the face of terrorist attack, using a simulation experiment that incorporates link capacity; and an extension of data farming which we call network farming. Our results show that symmetrical designed networks generally outperform randomly generated networks, although ring-like structures are very vulnerable. Under targeted attacks, most networks begin to fail when the number of attacks is equal to the node connectivity. Examining the distribution of real-world terrorist attacks, we show that these can be modelled by a Poisson statistical distribution, leading to recommendations for the node connectivity required at different threat levels.