Sensor data surveillance

Abstract

A growing volume of geodata requires for appropriate data management systems, which ensure data acquisition and memory-preserving storage as well as continuous surveillance of this unbounded amount of georeferenced data. Trend cluster discovery, as a spatiotemporal aggregate operator, may play a crucial role in the surveillance process of the sensor data. We describe a computation-preserving algorithm, which employs an incremental learning strategy to continuously maintain sliding window trend clusters across a sensor network. The analysis of trend clusters, which are discovered at the consecutive sliding windows, is useful to look for possible changes in the data, as well as to produce forecasts of the future.