Heterogeneity-Oriented Immunization Strategy on Multiplex Networks

Abstract

Many real-world complex systems can be treated as multiplex networks and there have constantly been unwanted diffusion (e.g., computer viruses, rumors, and epidemics) running on top of them. These type of network risks often lead to the global economic burden every year. Centrality-based immunization is an important approach to reduce the cost of preventing such unexpected massive outbreaks, for its effectiveness in cutting off the dissemination paths to delay the propagation. However, most of the current strategies on multiplex networks only focus on the topological structures when evaluating the influence of nodes, and the heterogeneity of individual behaviors has been less addressed. This paper proposes a heterogeneity-oriented (HO) immunization strategy for multiplex networks based on heterogeneous features of nodes. Specifically, the HO strategy treats nodes as independent agents, and the behaviors of them are defined and quantified in each layer. After coupling with the topological factor, this strategy is able to characterize the importance of nodes which can further be used for pre-immunization to delay the detrimental propagation. To testify the effectiveness, plenty of experiments are conducted based on a multi-agent email model. The results on large real-world and synthetic multiplex networks show that our strategy outperforms the existing representative strategies and effectively delay the propagation.