Affiliations: [a] Department of Biomedical Informatics, Faculty of Medical Sciences, Tarbiat Modares University, Tehran, Iran | [b] Department of Software Engineering, Faculty of Engineering, Islamic Azad University (Tehran South-Branch), Tehran, Iran | [c] Nima Institute, Mahmoodabad, Mazandaran, Iran
Correspondence:
[*]
Corresponding author: Ali Asghar Safaei, Department of Biomedical Informatics, Faculty of Medical Sciences, Tarbiat Modares University, Tehran, Iran. E-mail:aa.safaei@modares.ac.ir
Abstract: In a Data Stream Management System (DSMS), data is
received with a high speed and it has a high volume. Normally, data stream
processing has real-time requirements and needs one-pass algorithms. On the
other hand, DSMSs have limited memory capacity. Thus, approximate answers
might be used in most of data stream applications. The need for aggregate
functions is a common feature of both the traditional databases and the data
streams, that are also used as a solution for summarizing the massive data
received from data streams. In this paper, a method for answering Ad-hoc
Continuous Aggregate Queries over data streams is employed that is uses a
Dynamic Prefix Aggregate Tree. As each tuple from data stream arrives, the required data is stored as a
tree structure and used when applying the aggregate functions. Also, it is
important that all phases of tree management including tree construction,
maintenance and application perform in an online manner. To achieve this,
all of the results have to be stored. But in order to have online tree
construction and maintenance, we just need to store some of the results. So,
the proposed method aims to push the selected answers closer to the
registered queries answers, as much as possible. It has been found out both
empirically and analytically that the proposed method is more cost-effective
than using the conventional Prefix Aggregate Trees, in terms of time and
memory capacity.
