Fast Computation of Betweenness Centrality to enable Real-time Resilience Assessment and Improvement of Complex Transport Networks

With the growth of the population concentrated in urban areas of large agglomerations, the need for e cient and resilient multi-modal transportation systems is paramount. To model, analyze and improve transportation dynamics at large scale, complex networks represent an extremely versatile toolkit: multi-modal mobility networks can be modelled as a multi-layered weighted graph. In the last decade, several works [1, 2, 3] have shown that complex network approaches based on computation of centrality metrics can be extremely useful to model and analyze the resilience properties of complex networks. In such representation, each layer of the graph can be associated to a transportation mode (e.g., road, metro, buses, etc); each node of the network is an intersection between roads, a parking spot or a bus/metro stop/station; and the edges are links between the nodes, possibly belonging to di↵erent layers of the transportation network (e.g., links connecting bus with metro stations or parking areas in the proximity of a road intersection with bus stops, etc.). In this graph-based model, it is also possible to associate weights to the network edges, e.g., (average) travel time or, alternatively, speed/flow measurements retrieved via sensors (e.g., floating car data, loop detectors, cameras, etc.) and matched to the links of the network. We could also consider the possibility that weights dynamically evolve in time, as it could be the case in a futuregeneration smart city, equipped with an online monitoring system capable of continually collecting tra c speed observations at cityor region-scales (e.g., via navigation systems, mobile phone apps, etc.).