Online Timing Correlation of Streaming Data with Uncertain Timestamps

Abstract

We introduce the interval timing correlation, which can establish timing correlation conditions to handle interval timing timestamps. Interval timestamps are adopted to handle the temporal uncertainties in the timestamps of stream data. A probabilistic querying approach is taken in order to support timing predicates such as deadline, delay, and within over interval timestamps. A timing correlation condition entails a desired confidence threshold (minimum satisfaction probability). We define the interval timing correlation and discuss how to implement the algorithm. We present an analysis result which can effectively identify only tuples that need to be considered in determining the correlation. The performance of the proposed algorithm is shown.