Malware propagation in clustered computer networks

Abstract

Previous studies have been dedicated to strategies for combating malware propagation in data networks. One method to combat these deleterious actions is to establish preventive policies for system operations using network macroscopic models based on epidemiological studies, which is an efficient alternative compared to other methods. Among several studies based on the susceptible–infected–removed (SIR) epidemiological model applied in the context of computer networks, the introduction of antidotal populations in models has provided adequate descriptions of the real behavior of networks under attack. Currently, the susceptible–infected–removed–antidotal (SIRA) model is considered realistic for representing robust and complex networks. In this study, an approach related to the connection of different networks is presented as malware combat strategy design, deriving the effect of a network reliability loss on its neighborhood. Two clustered SIRA networks connected with whole dynamics were analytically and numerically studied, which allowed us to obtain the fundamental parameters for describing the malware dissemination between connected networks. Disease-free and endemic conditions, and several possible qualitative behaviors associated with the parameter space were identified, permitting the calculation of the basal reproduction number. This calculation allowed the design of control strategies to be implemented during the planning of essential safety measures in data networks.