Risk Perception in Modeling Malware Propagation in Networks

We investigate the effects of risk perception in a SIS model for malware propagating in different types of networks such as regular, random and scale-free. We assume that the perception of the risk of being infected rely on the fraction of neighbors that are infected. The effects are mainly affected by two parameters denoted by J and ¿, which models the linear response and nonlinear effects respectively. They can reduce the infectivity of the malware as a function of the infected neighbors. We study the models in the mean-field approximation and by numerical simulations for the three kinds of networks. The results show that for homogeneous and random networks, there is always a value of perception that stops the malwares. But in the ¿worst case¿ scenario of a scale-free network with diverging connectivity, a linear perception cannot stop the malwares. With the nonlinear increase of the perception risk, however, the malware tends to be extinct. This transition is not continuous and is presumably induced by fluctuations in center nodes such as hubs or switches. An understanding of the risk perception in modeling malware propagation in networks is very important for designing effective detection and prevention strategies for such networks.