A mechanism for data quality estimation of on-body cardiac sensor networks

In this paper, we present a mechanism for estimating data quality of BANs composed of cardiac sensors. Currently available cardiac monitoring sensors suffer from high level of noise generated from loose physical contact of the sensor node due to the highly mobile and pervasive environment of the BAN (e.g., at-home remote health care applications). Therefore, there is a need to estimate the data quality of individual cardiac sensors as well as the data quality of the overall BAN while particularly considering the resource scarceness of BAN-scale mobile systems. The proposed method successfully estimates the data quality of a BAN without employing computationally expensive machine learning techniques. It also provides a number of resource management options that enable efficient data quality estimation. We present experimental results of four participants with three off-the-shelf cardiac sensors to form a BAN. We also present simulation results to examine if the proposed mechanism can successfully detect health hazardous events such as heart arrhythmia.