Graph Processing For Spatial Network Queries

Spatial network queries often have performance bound by the structure and size of the underlying network. This paper discusses methods of improving the performance of these queries, specifically those using network expansion algorithms, by creating a graph representation of the network and removing unnecessary nodes and edges. In addition, two methods of storing and accessing the graph are compared for speed and usefulness in different applications. A main memory approach to graph storage using a shared library is compared to a database storage approach. Though fast, the main memory approach has some limitations in its usefulness.