Modeling and Predicting Malware Propagation in Double-Layer Computer Networks

Traditional malware propagation models exhibit inherent limitations in addressing topological heterogeneity, computational complexity, and quantifying perturbation responses. To overcome these challenges, this paper proposes a novel malware propagation dynamics model based on a double-layer network architecture. By decomposing large-scale networks into core and peripheral subnetworks, we develop a heterogeneous recovery strategy framework: the core subnetwork adopts a state-dependent feedback mechanism, while the peripheral subnetwork employs a fixed recovery rate. To reduce computational complexity, a degree-weighted dimensionality reduction method is integrated into the framework, transforming high-dimensional differential equations into a low-dimensional system. Propagation behavior is analyzed via the basic reproduction number ($R_{0}$) to identify dominant spreading factors. Validation experiments on Erdős-Rényi (ER) and Barabási-Albert (BA) synthetic networks, as well as real-world Facebook and Email-Eu-core datasets, demonstrate the model' s effectiveness in predicting the impacts of structural and dynamic parameter perturbations on malware propagation. Further analysis reveals how these perturbations quantitatively influence propagation dynamics. This research establishes a predictive framework for malware propagation in double-layer networks and provides a theoretical foundation for proactive defense strategies in critical infrastructure systems such as power grids and communication networks.