Abstract
Spatio-temporal data streams that are generated from
mobile stream sources (e.g., mobile sensors) experience
similar environmental conditions that result in distinct phenomena.
Several research efforts are dedicated to detect
and track various phenomena inside a data stream management
system (DSMS). In this paper, we use the detected
phenomena to reduce the demand on the DSMS resources.
The main idea is to let the query processor observe the input
data streams at the phenomena level. Then, each incoming
continuous query is directed only to those phenomena that
participate in the query answer. Two levels of indexing are
employed, a phenomenon index and a query index. The phenomenon
index provides a fine resolution view of the input
streams that participate in a particular phenomenon. The
query index utilizes the phenomenon index to maintain a
query deployment map in which each input stream is aware
of the set of continuous queries that the stream contributes
to their answers. Both indices are updated dynamically in
response to the evolving nature of phenomena and to the
mobility of the stream sources. Experimental results show
the efficiency of this approach with respect to the accuracy
of the query result and the resource utilization of the DSMS