Opportunistic Sensing and Learning in Sensor Networks

Romain Tavenard, Onkar Ambekar, Eric J. Pauwels, Max Waaijers

Abstract: In this paper we will argue that sensor networks in which multimodal sensors are connected to each other and computational devices capable of data mining, offer the possibility of serendipitous and opportunistic sensing in which unanticipated associations are detected and used to produce more robust event recognition. To illustrate this point of view, we outline two simple scenarios and conduct some preliminary experiments to show that this type of on-the-fly data analysis can indeed be useful.