A bio-inspired field estimation scheme for wireless sensor networks

Abstract

This paper proposes and analyzes a bio-inspired field estimation scheme using wireless sensor networks. The proposed scheme exploits the temporal pattern of the sensed process to reduce the number of samples sent back to the sink by a sensor node and, as consequence, decrease the energy consumption in data transmission. The proposed scheme is orthogonal to the techniques that reduce the spatial density of collected samples deactivating nodes with similar measurements. Thus, the proposed scheme can be used along with these techniques. We present two variations of this scheme: a sample-bounded and an error-bounded. The sample-bounded limits the maximum number of samples sent back to the sink, while the error-bounded guarantees the observation of every event of interest. Results show that for very regular processes the scheme can reduce up to 90% the total amount of samples sent in the network and even for less regular processes the proposed scheme can reduce the total amount of samples sent from approximately 10 up to 20%, with small reconstruction errors.