Complex Networks Antifragility under Sustained Edge Attack-Repair Mechanisms

Abstract

Resilience is an important property of real-world complex networks with many applications in technological, biological, and social systems. While many natural systems are particularly resilient, some are antifragile, namely, they become stronger when being subjected to attacks, volatility, or errors. In this paper, we consider an edge-attack and local edge-repair response mechanism over several synthetic and real-world datasets, on which we quantify both antifragility (as the dynamics of the largest connected component) and the cost incurred by edge repairs. Our findings show that (1) random repairs generate a stronger antifragile response, thus confirming that antifragility manifests itself in the context of random, rather than deterministic events; and (2) antifragile behavior is fostered by strongly clustered topologies (e.g., real-world networks and the synthetic Watts–Strogatz model with degree distribution). Our results represent a first step towards designing highly resilient networks and developing new methods for thwarting the antifragile response of harmful and hostile systems.