Multilayered Database Model and Approximate Query Processing Model for Mobile Computing

In environments such as wireless computing, it is neither practical nor necessary to adhere to provide correct (i.e. sound and complete) answers for database queries. Users may not obtain perfect answers to their queries within an acceptable time and therefore, within known limits of correctness and precision, an approximate answer may suffice for some mobile users. Due to the limited bandwidth, direct answering of queries from a mobile device will likely jam the communication and thus will be very slow. Therefore, it is better to feed the information progressively as the client refines the query. For examples, when a traveler drives to a place, she uses her PDA to find about interesting local attractions. To answer her query, the server can first send a list of attractions with brief description in text, and then send thumbnail images and directions for some of the attractions, and eventually full size images, maps, or video clips for the interested attractions. Such practices of information filtering have been done in an ad hoc way by many Web sites.

In this research, we proposed to create a generalized multilayered systems where each layer can be generated from the layer below using some generalization operators such as Concept hierarchies (described later), sampling, transformation, grouping etc. We have explored multi-layered database and its potential and effectiveness in intelligent query answering in mobile environments. A multi-layered database (MLDB) is a database composed of several layers of information, with the lowest layer corresponding to the primitive information stored in a conventional database, and with higher layers storing more general information extracted from lower layers.

Resercher

Dr. Sanjay Madria

Dr. Y. Fu, Nanyang Technological University, Singapore