A data imputation model in sensor databases

Abstract

Data missing is a common problem in database query processing, which can cause bias or lead to inefficient analyses, and this problem happens more often in sensor databases. The reasons include power outage at the sensor node, sensors time synchronization, occurrences of local interferences, unstable wireless network communication, etc. Therefore, in sensor database applications, there is a need for data imputation, especially for those applications in which the query response time is tight, and the accuracy of the query results is important. In this paper, we present a data imputation application based on association rule mining of closed frequent itemsets. They are subsets of all frequent patterns but provide complete and condensed information since they do not include redundant patterns. Experimental results compared with the existing techniques using real-life sensor data show that our proposed technique effectively imputes missing sensor data as well as achieves time and space efficiency. © Springer-Verlag Berlin Heidelberg 2007.