Mining association rules between sets of items in large databases
SIGMOD '93 Proceedings of the 1993 ACM SIGMOD international conference on Management of data
Algorithm 145: Adaptive numerical integration by Simpson's rule
Communications of the ACM
Aurora: a data stream management system
Proceedings of the 2003 ACM SIGMOD international conference on Management of data
TinyDB: an acquisitional query processing system for sensor networks
ACM Transactions on Database Systems (TODS) - Special Issue: SIGMOD/PODS 2003
Streaming pattern discovery in multiple time-series
VLDB '05 Proceedings of the 31st international conference on Very large data bases
Using Data Mining to Estimate Missing Sensor Data
ICDMW '07 Proceedings of the Seventh IEEE International Conference on Data Mining Workshops
DEMS: a data mining based technique to handle missing data in mobile sensor network applications
Proceedings of the Seventh International Workshop on Data Management for Sensor Networks
Research issues in outlier detection for data streams
ACM SIGKDD Explorations Newsletter
| Hi-index | 0.00 |
A sensor network is a valuable new form of collective computational instrumentation by virtue of its ability to sense physical quantities of interest and to transmit such readings via sub-networks of nodes/computers for processing. Such computing environment typically generates massive amounts of data rapidly in real-time. These infinite volumes of online time-series are formally characterized as data streams. The wealth of fast incoming data streams presents both overhead and logistic challenges for sensor network applications. In this research, we introduce an online data mining framework to serve as an overhead-bounded knowledge discovery tool for sensornet applications. Our framework extends the notions of traditional association rules to multivariate continuous data and uses spatio-temporal correlations to make intelligent inferences about the monitored variables. Our mining framework is additionally pertinent to data estimation, which is an important capability given the inevitability of data loss/corruption with the current sensornet technology. Experimentation shows efficiency of our approach both in terms of overhead cost and quality of missing data estimates.