Effective Graph Protection Method to Prevent the Spreading of Attacks in Networks

Abstract

Networks are fundamental models for representing and analyzing the structures of real-world systems. For instance, in social networks, nodes are used to represent users and edges represent the connection between users. Networks are also termed as graphs in the discrete mathematics language. One essential problem in networks is how to protect a limited number of nodes to prevent the spreading of malicious attacks or dangerous rumor in the networks, which is known as the graph protection problem. In this paper, an effective graph protection method called PowerShield is proposed which pre-emptively protects critical nodes prior to any incoming attacks. It combines connectivity and centrality criteria of the input graph. Connectivity criterion is measured by the principal eigenvector, i.e., the eigenvector corresponding to the largest eigenvalue of the adjacency matrix of the input graph. Centrality criterion is defined by the degree centrality which considers nodes having more neighborhood relations to be more important. Contrary to the existing state-of-the-art method which takes into account only the connectivity criterion, the proposed method combines both criteria and empirically improves the effectiveness of protection result.