Description

This project addresses indexing of spatial, temporal, and spatiotemporal data. More specifically, the project focusses on applications where data is continuously evolving. In conventional databases, data is assumed to be constant unless it is explicitly modified, whereas in new application areas, there is often a need to support continouosly evolving data - data that changes continuously as a function of time, even without being explicitly updated. Indexing of two broad classes of continuously evolving data are investigated in the frame of this project.

In spatiotemporal applications, there is often a need to record continuously changing spatial information about objects that move and/or change their shape. Examples of such objects include aircraft, taxis, boats, mobile computers, communication equipment, military equipment, people, forest fires, hurricanes, to name but a few. In application areas such as vehicle navigation or mobile data management, there is a need for indices that would efficiently index the history of the evolution of spatial objects, supporting interpolation between explicit updates of the database.

Another interesting problem is indexing of the current state of continuously evolving spatial objects, with the ability to extrapolate their positions (and shapes) into the future. In such a scenario, the "current state" would be computed each time the database is queried. The computation would be based on the latest information stored in the database about positions (and shapes) of objects and about their intents, e.g., as indicated by their velocity vectors.

In temporal and spatiotemporal databases, continuously evolving data occurs naturally when we want to support data related to the continuously progressing current time, termed now-relative data. Consider an example where we want to record information about the residence place of a customer in the database. We know when the customer starts living at a specific address, but we probably do not know the date when the customer will move out; instead, we want to record that the customer is living there until now, which changes with each "tick of the clock."

In addition, there is often a need to support bitemporal data. In a bitemporal database, both valid and transaction times of facts are recorded, and both of them can be now-relative. The combined valid and transaction times of a fact can be treated as a region in two-dimensional space. Furthermore, if transaction time or both valid and transaction times are now-relative, the region grows continuously as time passes. Thus, indexing of now-relative temporal data is related to indexing of continuously evolving spatial data. The idea of this Ph.D. project is to conduct research in parallel on indexing of both kinds of data, so that synergy occurs between the two research tracks. 

Further readings:

	S. Saltenis, C. S. Jensen, S. T. Leutenegger, and M. A. Lopez, Indexing the Positions of Continuously Moving Objects In the Proceedings of the SIGMOD'2000 Conference [.ps.gz]
	S. Saltenis and C. S. Jensen, R-Tree Based Indexing of General Spatio-Temporal Data [.ps.gz]
	R. Bliujute, C. S. Jensen, S. Saltenis, and G. Slivinskas, R-tree Based Indexing of Now-Relative Bitemporal Data In the Proceedings of the VLDB'98 Conference [.ps.gz]
	R. Bliujute, C. S. Jensen, S. Saltenis, and G. Slivinskas, Light-Weight Indexing of General Bitemporal Data [.ps.gz]
	R. Bliujute, G. Slivinskas, S. Saltenis, and C. S. Jensen, Developing a DataBlade for a New Index In the Proceedings of the ICDE'99 [.ps.gz]