Characterising the robustness of coupled power-law networks

Abstract

Many networks exhibit a power-law configuration, where the number of connections each node has follows a power-law distribution, including the Internet, terrorist cells, species relationships and infrastructure. Given the prevalence of power-law networks, studying the effects of disruptions on their performance is of interest. Previous work has investigated the influence of network topology on the effects of random node failures for independent networks. Many networks depend on others to function and thus, exploring the influence of network topology on the effects of failures in interdependent networks is of interest. The present paper extends the previous work to coupled power-law network systems. For a set of randomly generated coupled systems, each containing two networks, we investigate the significant topological factors for different dependency types. Failures in the coupled networks are simulated and the effects on the system performance are analysed by performing a beta regression. The results are consistent across the dependency types, with the most influential topological factors being mean nodal degree and factors relating to the dependency type. The results are also compared with those of the independent networks and their potential relevance to the design of interdependent networks is indicated, for example, their use within an infrastructure setting.