In-network Processing by the Example of Maxima Estimation in Spatial Fields

Wireless sensor networks are often used to observe the spatial distribution of a physical quantity. By summarizing the observations of hundred or more nodes in a small set of parameters, the amount of data that has to be transmitted from the network to an external observer can be reduced. In this article we show by the example of an algorithm for estimation of local field maxima, that it is also possible to reduce the data exchange within the network by distributed in-network processing. Each sensor node requires information only from its 15 closest neighbors, thus reducing communication by more than 50% compared with central data processing by the gateway between network and external observer. The algorithm is well suited to run in parallel on all sensor nodes. The accuracy of the estimation was evaluated for different measurement noise powers, numbers of sensors, and sensor placement schemes.