Targeted attack on correlated interdependent networks with dependency groups

Many different cyber–physical infrastructure systems can be described as interdependent networks. Attacking important infrastructures deliberately leads to catastrophic influences on the interdependent systems. In this paper, we propose a toy model to describe the targeted attack on interdependent networks with groups. Through a percolation theory analysis, we find that attacking hubs is more likely to destroy the giant cluster. For homogeneous artificial interdependent networks, attacking nodes with any degrees cannot alter the phase transition. However, for heterogeneous correlated interdependent networks, the giant cluster decreases discontinuously (continuously) with the fraction of initial failed nodes when nodes with large (small) degree are more likely to be attacked. The theory can well predict the numerical simulation results.