-tree for Temporal Spatial OLAP in Traffic Data Warehouse

In order to execute analytical queries in data warehouses, many researchers proposed various system architectures. We also develop spatial information systems for real time location services using PHS and GPS, we need OLAP (On-Line Analytical Processing) technologies in our traffic data warehouse. In order to deal with huge amount of data, many kinds of attributes, and a lot of dynamic queries, we store temporal spatial data using ∑ -tree, which is the data structure to calculate traffic parameters fast. Our proposed ∑ -tree can decrease computing costs by making nodes with spatial constraints and using prefix-sum techniques. We also discuss how to calculate traffic parameters in ∑ -tree. We compare ∑ -tree with 3DR-tree by using simulation data based on the traffic model of Hanshin expressway. We evaluate “the number of operations in a query region” by counting the numbers of nodes in a query space and objects in leaf nodes. As the result, we make clear that ∑ -tree is superior to 3DR-tree.