Tracking Uncertain Shapes with Probabilistic Bounds in Sensor Networks.

We address the problem of balancing trade-off between the (im)precision of the answer to evolving spatial queries and efficiency of their processing in Wireless Sensor Networks (WSN). Specifically, we are interested in the boundaries of a shape in which all the sensors’ readings satisfy a certain criteria. Given the evolution of the underlying sensed phenomenon, the boundaries of the shape(s) will also evolve over time. To avoid constantly updating the individual sensor-readings to a dedicated sink, we propose a distributed methodology where the accuracy of the answer is guaranteed within probabilistic bounds. We present linguistic constructs for the user to express the desired probabilistic guarantees in the query’s syntax, along with the corresponding implementations. Our experiments demonstrate that the proposed methodology provides over 25 % savings in energy spent on communication in the WSN.