Abstract: To solve the problem that how to find important nodes accurately and quickly in complex and huge AS-level networks, a valuable research of centralization was carried out. With three common measurement such as degree centrality, closeness centrality, betweenness centrality, core centrality was proposed to act as an indicator to measure high-core nodes combination. The importance of a node within a network was characterized as the extent of which the network has been destroyed by deleting the node. The applicability of the closeness centrality was not good as degree centrality and betweenness centrality for Internet AS-level. There was high similarity in degree centrality and betweenness centrality when the proportion of attack nodes was smaller than 0.5%. Core centrality was applied to finding out the communities which were made up of high-degree and closely-connected nodes within networks.