In-network data reduction approach based on smart sensing

Abstract

The rapid advances in wireless communication and sensor technologies facilitate the development of viable mobile-Health applications that boost opportunity for ubiquitous real- time healthcare monitoring without constraining patients' activities. However, remote healthcare monitoring requires continuous sensing for different analog signals which results in generating large volumes of data that needs to be processed, recorded, and transmitted. Thus, developing efficient in-network data reduction techniques is substantial in such applications. In this paper, we propose an in-network approach for data reduction, which is based on fuzzy formal concept analysis. The goal is to reduce the amount of data that is transmitted, by keeping the minimal-representative data for each class of patients. Using such an approach, the sender can effectively reconfigure its transmission settings by varying the target precision level while maintaining the required application classification accuracy. Our results show the excellent performance of the proposed scheme in terms of data reduction gain and classification accuracy, and the advantages that it exhibits with respect to state-of-the-art techniques.