Detecting Critical Streets in Road Networks Based on Topological Representation

Abstract

We provide a novel problem of analyzing geographical road networks from a perspective of identifying critical streets for vehicular evacuation. In vehicular evacuation behaviors during disasters and emergency situations, the shortest distance routes are not necessarily the best. Instead, routes that are easier to traverse can be more crucial, even if they involve detours. Furthermore, evacuation destinations need not be limited to conventional facilities or sites; wider and better maintained streets can also be suitable. Therefore, we focus on streets as basic units of road networks, and address the problem of finding critical streets in a geographical road network, considering a scenario in which people efficiently move from specified starting intersections located around their residences to designated goal streets, following the routes of easiest traversal. In this paper, we first model a road network as a vertex-weighted graph obtained from its topological representation, where vertices and edges represent streets and intersections between them, respectively. The weight of each vertex reflects its ease of traversal. Next, we extend a recently introduced edge-centrality measure, salience, for our problem, and propose a method of detecting critical streets based on the vertex-weighted graph of topological representation by incorporating the notion of damping factor into it. Using real-world road network obtained from OpenStreetMap, we experimentally reveal the characteristics of the proposed method by comparing it with several baselines.