Efficient Route and Area Matching Query in Dynamic Road Networks

Nowadays, ride-sharing is developing rapidly because of its economic and environmental advantages. Recent studies investigate the benefits of introducing meeting points during task assignments, allowing riders to be picked up or dropped off near their requested locations. In this paper, we study the route and area matching (ROAM) problem in dynamic road networks. ROAM query aims to find a detour path from source to destination visiting an area, meanwhile satisfying a detour budget. Existing method excludes unmatched queries in a compacted sketch graph, but the Dijkstra-based routing process is still time-consuming. Moreover, maintaining the sketch graph in frequently changing road networks is challenging because it requires computing from scratch. To overcome the limitations, we propose a simple yet effective framework named meeting point search (MPS). A novel index structure named GS-Tree is constructed to integrate spatial information for selecting meeting points and graph shortcuts for routing queries. GSTree has structural stability and can be efficiently maintained in dynamic networks. Theoretical analysis and experimental studies demonstrate the superiority of our methods. The optimal MPS with GS-Tree averagely outperforms the existing methods by two orders of magnitude.