Vulnerability of complex networks under three-level-tree attacks.

We investigate vulnerability of complex networks including model networks and real world networks subject to three-level-tree attack. Specifically, we remove three different three-level-tree structures: RRN (Random Root Node), MaxDRN (Max Degree Root Node) and MinDRN (Min Degree Root Node) from a network iteratively until there is no three-level-tree left. Results demonstrate that random network is more robust than scale-free network against three tree attacks, and the robustness of random network decreases as the 〈 k 〉 increases. And scale-free network shows different characteristics in different tree attack modes. The robustness of scale-free is not affected by the 〈 k 〉 parameters for RRN, but increases as the 〈 k 〉 increases for MinDRN. The important thing is that MaxDRN is the most effective in the three tree attack modes, especially for scale-free network. These findings supplement and extend the previous attack results on nodes and edges, and can thus help us better explain the vulnerability of different networks, and provide an insight into more tolerant real complex systems design. [ABSTRACT FROM AUTHOR]