Near-optimal routing for contour detection in wireless sensor networks

Wireless sensor networks are often deployed to monitor scalar fields. Commonly, these networks employ a query-based architecture, where specific data are requested by the sink node, and the network responds. Contours or isolines are compact representations of the state of a scalar field, and are useful abstractions of scalar field behavior. Previously researchers have examined algorithms and policies for detecting contours in query-based sensor networks, but have only addressed individual components of the process. In this paper we present a set of algorithms for contour detection in WSNs, and associated oracle functions to provide an unbiased optimality comparison. We show that our algorithms, and in particular our response and aggregation policy provide near-optimal performance in simulated networks measuring physically realistic scalar fields.