Multi-Scale System Reliability Analysis of Multi-Layer Network Infrastructures

This study presents a novel framework that integrates dynamic multi-layer graph theory with cascading failure dynamics to analyze the resilience and reliability of multi-layer network infrastructures. These infrastructures include critical systems such as power grids, which are inherently susceptible to cascading failures due to their complex, interconnected structures. Our research introduces a sophisticated model that dynamically captures the multi-dimensional interactions of the systems across various scales. Each layer of the network, from physical components to control systems, is continually updated with real-time operational data and historical incident records, enhancing the model’s predictive accuracy and responsiveness to emerging threats. The core of our multi-scale system reliability analysis is the evaluation of Node-Centric and Edge-Centric Subgraph Reliability Metrics (N-SRM and E-SRM), examining both intra-layer and inter-layer dynamics. Through extensive simulations under two critical scenarios-cyberphysical attacks and extreme weather conditions-we quantitatively assessed the network's resilience. For cyber-physical attacks, rapid propagation of failures was observed, significantly impacting the network's operational integrity. The N-SRM values across the layers exhibited steep declines within a short duration, demonstrating pronounced susceptibility to these attacks. In contrast, the impact of extreme weather was characterized by a gradual decline in reliability, suggesting a robustness that allows for effective adaptation strategies over an extended period. In addition, the inter-layer connections, particularly in Scenario B, demonstrated higher resilience with slower rates of reliability decline, underscoring the structural integrity and operational capability of the network.