Robustness Assessment of Urban Road Networks in Densely Populated Cities.

Abstract

This paper presents a robustness assessment in terms of inducing damage to the functioning of real-world urban road networks via a comparative analysis of the efficacy of various network perturbation strategies. Specifically, we assess the network robustness through an iterative node removal process considering five targeted (deterministic) and two random (stochastic) strategies. The targeted node removal strategies are based on different centrality measures. We study the robustness of 10 road networks of densely populated cities using three different metrics: the size of the largest connected component, global efficiency, and local efficiency. Our findings suggest that targeted disruptions utilizing centrality measures are more effective in disrupting the network than random ones. However, some centrality measures have a strong correlation with each other and thus, requiring combinations of different removal orders to gain more comprehensive insights into the ability of the network to withstand perturbations. We find centrality measures considering shortest paths are more effective in degrading the robustness of the network as a whole while centrality measures that only consider directly connected neighbours are better in disrupting the local effectiveness of the network. Interestingly, we also find that removing nodes can counter-intuitively increase the local efficiency of the network.