Indexing of spatiotemporal Data: A comparison between sweep and z-order space filling curves

Abstract

Storing constantly changing spatial and temporal (spatio-temporal) features requires multi-dimensional data support in data management applications. That necessitates allocation of large amount of data storage in computer systems. Commonly, relational database management systems (RDBMS) are used for managing those data. However, an algorithm should be implemented to map multi-dimensional data to one-dimensional data in RDBMS. We found that a group of researchers worked on this problem and solved with SPIT (Space-Partitioning with Indexes on Time) approach [Mallett 2004]. SPIT partitions space according to sweep-space filling curve and the researchers argue that using sweep-space filling curve puts lesser demands on I/O than z-order space filling curve. Accordingly, researchers suggest using the former method for storing and indexing of spatio-temporal data. In this paper, we present the results of a series of experiments where we partitioned space by z-order space filling curve and used a sliding window technique for tracking of moving objects. We suggest that in the cases where the tracked objects are close to each other, we can obtain better performance from z-order partitioning than sweep-space filling curve by using sliding window technique for tracking of moving objects.