Distance Closures: Unifying Search- and Lookup-based Shortest Path Speedup Techniques

Daniel Bahrdt, Stefan Funke, Sokol Makolli, Claudius Proissl

Abstract: Most popular speed-up techniques for shortest path queries in road networks are based either on pruned graph search or clever lookup schemes and allow for answering of shortest path distance queries within continent-sized road networks in less than a milli-(search-based) or even microsecond (lookup-based) compared to several seconds of a normal Dijkstra run. While both paradigms previously have been considered mostly separately, we present a framework that unifies these seemingly different views on shortest path computations. Apart from the conceptual novelty, this allows for new and (practically very attractive) space-time tradeoffs for shortest-path computation. To our knowledge we are also the first to report on computational results for the largest connected component of the Open-StreetMap planet road network with more than half a billion nodes.