Seminar: AI-Driven Approaches for Network Resilience: From Attack-Defence Strategies to Cascading Failure Mitigation

Student: Jordan Lanctot

Supervisor: Dr. Sean Cornelius

Abstract

Networked systems are integral to modern infrastructure but are particularly susceptible to cascading failures, where small disturbances can trigger widespread disruptions. This research examines network robustness by applying the Abelian Sandpile Model (ASM) to temporal networks, a model where nodes topple and redistribute stress when exceeding critical thresholds, leading to further topples and resulting in avalanches. This novel application of the ASM to temporal networks allows us to explore local overloads propagate through a system, leading to large-scale failures, and how changing network topologies change exacerbate or mitigate resulting widespread failures. Here, we simulate the dynamics of cascading failures and investigate mitigation strategies, including targeted sand dropping and link rewiring, to reduce the frequency and severity of large avalanches. Our findings contribute valuable insights into failure propagation, offering potential solutions for enhancing the resilience of complex networks such as power grids, communication systems, and transportation networks. By refining our understanding of these failure mechanisms, we aim to support the development of more robust and adaptive networked systems.