An improved network clustering method for I/O-efficient query processing

Abstract

Efficient network query processing is extremely important in Geographical Information Systems (GIS) and Intelligent Transportation Systems (ITS) which include various applications of transportation, utility and communication networks, etc. In order to reduce the I/O cost in network query processing a given network should be stored with high disk-space utilization and a low edge-cut ratio. To do so the nodes in the network should be clustered in such a way that each cluster fits in a disk page with as small number of edge-cuts as possible. This is known as the network clustering problem which is NP-complete. In this paper we present a network clustering method, called Network-Traversal Clustering (NTC) method, which assigns the nodes of a given network to disk pages with optimal disk-space utilization and a low edge-cut ratio. NTC is suitable for the bulk loading operation into a network indexing structure. In the experimental results NTC outperformed the Connectivity-Clustered Access Method, the Spatial Partition Clustering (SPC) and SPC-R which is a modification of SPC on various networks.