Structural Vulnerability Assessment in Urban Transport Networks: A Network-Wide Geometric Approach Using Gromov-Wasserstein

TL;DR

This paper introduces a novel geometric approach using Gromov-Wasserstein distance to assess the global structural vulnerability of urban transport networks, effectively identifying critical links overlooked by traditional metrics.

Contribution

It proposes a new method to quantify network disruptions via Gromov-Wasserstein distance, providing a global perspective on network vulnerability and critical element identification.

Findings

Gromov-Wasserstein distance is uncorrelated with local measures like betweenness.

It strongly correlates with transportation service level degradation.

Method effectively ranks critical edges for infrastructure resilience.

Abstract

Urban transportation networks are inherently vulnerable to disruptions that affect connectivity and passenger mobility. Traditional graph_theoretic metrics, such as betweenness and degree centrality, offer insights into local network structure but often fail to capture global structural distortions resulting from link failures. On the other hand, global indices, such as those based on spectral analysis of the networks graph, fail in identifying critical elements. This study proposes to quantify the structural modifications implied by the disruption of single elements in a transportation network through the Gromov-Wasserstein distance. Specifically, we iteratively remove one single edge from the original network to simulate a disruptive event and then compute the Gromov-Wasserstein distance between the original network and the disrupted one. Finally, edges are ranked depending on the observed Gromov-Wasserstein distance: the higher the value of the distance, the more critical the edge is in terms. Two transportation networks from Berlin are considered in the experiments, namely Berlin Friedrichshain Center (BFC) and Berlin Tiergarten (BT). Results reveal that Gromov-Wasserstein is largely uncorrelated with edge betweenness (rho<0.1), proving its ability to capture vulnerability aspects overlooked by local network measures. Moreover, Gromov-Wasserstein exhibits an almost perfect correlation (rho=0.9999) against a proxy measure of the transportation service level, that is, the increase in the maximum shortest path. As a result, the Gromov-Wasserstein distance can be used to rank edges depending on their criticality with respect to their individual impact on the overall infrastructure and level, allowing for prioritizing maintenance, emergency planning, and enhancing the resilience of the urban transport network.