Scalability in spatio-temporal data management systems
Abstract
One major challenge in building spatio-temporal data management systems is to enhance their scalability. This dissertation presents novel techniques for building scalable spatio-temporal data management systems. The techniques presented span various components and levels of a data management system, including moving object indexing, continuous query execution, and distributed server networking. For moving object indexing, this dissertation explores memo-based techniques to enable scalable object updates in existing spatial indexes. Specifically, the RUM-tree and the LUGrid indexes are introduced for moving object indexing while taking into consideration the frequent updates incurred by the object movements. For continuous query execution, techniques for incremental evaluation and shared execution are explored, in particular, a scalable algorithm for evaluating a large set of k-nearest-neighbor queries. For distributed server networking, the scalability problem is addressed when one single server cannot support large numbers of objects and queries in wide or dense areas. PLACE*, a distributed spatio-temporal data stream management network for moving objects, is introduced. PLACE* supports continuous spatio-temporal queries that are answered by a network of data stream management systems.
Degree
Ph.D.
Advisors
Aref, Purdue University.
Subject Area
Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.