An Efficient Scheme for Detecting Phenomena in Multiple Data Streams

A phenomenon appears in a sensor network when a group of sensors continuously produces similar readings over a period of time. This is ongoing research project in which we propose new efficient algorithms for detecting phenomena and the correlation between the phenomena. This paper presents a scheme for detecting the similar streams. The proposed method uses Discrete Fourier Transformation to reduce the dimensionality of the streams. Each stream is represented by a point in a multidimensional grid. We apply grid-based clustering to find the similar streams. Experiments on synthetic data streams showed that the proposed method is accurate and more efficient than other existing traditional clustering techniques.