摘要:A network attack is a set of network elements that are disabled by an adversary. The goal for the attack is to produce the most possible damage to the network in terms of network connectivity by disabling the least possible number of network elements. We show that the problem of finding the optimal attack in a serverless network is NP-Complete even when only edges or nodes are considered for disabling. We study a node attack policy with polynomial complexity based on shorter paths and show that this attack policy outperforms in most cases classical attacks policies such as random attack or maximum degree attack. We also study the behavior of different network topologies under these attack policies.