Spatiotemporal databases is a domain that in recent years has gained a lot of attention. Spatiotemporal databases combines two types of databases: spatial databases and temporal databases. Spatial databases are designed for storing and processing spatial information efficiently.Temporal databases have a built-in time concept incorporated to handle queries related to temporal aspects. A spatiotemporal database designed for managing geometric objects that move and change through time is called a moving object database. This report will focus on moving object databases and the different data models that exist to enable them. The moving objects in this study will be represented by flies, moving around randomly in a simulation.
This study evaluates two different data models for storing spatiotemporal data: the Snapshot Model and Sliced Representation. The Snapshot Model is very simplistic. An entry in the database is a snapshot of an object at a specific time. Sliced Representation is a bit more complicated. Temporal development of a value is decomposed into fragments called slices. Within the slice the temporal development of a value can be described by a mathematical function. To evaluate the Snapshot Model, PostgreSQL with an extension called PostGIS which supports spatial data will be used. The temporal aspect will consist of timestamps which will be added to every entry in the database. To evaluate the slicing model, an application called Secondo, developed at the Hagen University in Germany, will be used. There is a set of conditions to determine which method is the most appropriate for the given problem: the creation of the flies and their movement should not exceed 10 seconds. Retrieval of flies that occur in a specified space during a specified moment should not exceed 10 seconds.
The conclusion of this report indicates that the favored data model for the objective is Sliced Representation. The application which uses Sliced Representation, Secondo, could manage 178 moving objects while satisfying the conditions. When the amount of moving objects was increased to a breaking point the application became unstable and started crashing. The breaking point lies between 200 and 500 flies. The application which uses the Snapshot Model, PostGIS will still function at this point but up until this point it performs worse than Secondo. With the conditions in mind, the implementation of the Snapshot Model can handle a maximum of 92 flies and the implementation of the slicing model can handle a maximum of 178 flies correspondingly. Thus, the slicing model is best suited for the objective.