Race Cars and Databases

Recently much research has been done in applying database technology to tracking moving objects.  Most techniques assume a predictable linear motion and slowly changing velocity.  This seminar presents a new application domain for moving object databases, characterized by repetitive, unpredictable motion with rapidly changing velocity and very high data rates.  In particular, the domain of athletic and auto races is presented.  Existing moving object methodologies do not adequately address this application domain.

A data model will be presented for storing the data efficiently.  Since existing spatiotemporal indexing methodologies were found to be inadequate, a new spatiotemporal index was developed as part of this research.  The data model, along with the new spatiotemporal index, allows fast retrieval of data.  A database management engine was implemented, incorporating the data model, storage method, and new index.  A performance study will be presented showing that this implementation has far better performance than those based on relational DBMSs.

This talk is based on a paper which will be presented at the DEXA 2002 conference in September.