Distributed event detection in multidimensional data streams in wireless sensor networks

Modern ecological and geophysical monitoring requires continuous real time data collection for many param-
eters (temperature, humidity, etc.) on large surface (square kilometers) with high resolution in space and time
(meters, seconds). The example may be monitoring of active volcanoes, earthquake precursors, oil-dynamics of
glaciers and snow cover, when experiment conditions do not allow to develop wireline networks for energy supply
and communication. Wireless sensor network (WSN) technology can combine sensors, microprocessors for data
collection and first processing, power supply systems and radio link in a small enclosed device and join many such
units into a wireless network of data acquisition capable to work in the field for years. The fundamental problem
here is to develop event detectors and real time data processing algorithms at network nodes which would allow
to reduce the data traffic in the wireless network thus extending it into hundreds and thousands of nodes. We
will present algorithms for event detection at wireless network nodes and for synchronization of the detection
results over the net to monitor extended areas by a large number of sensors; active prototype is also introduced.