Real-time sampling rate adaptation based on continuous risk level evaluation in wireless body sensor networks

Abstract

Wireless Body Sensor Networks (WBSNs) are a low-cost solution allowing remote patient monitoring and continuous health assessment, thus reducing healthcare expenditure. In such networks, sensor nodes periodically collect vital signs and send them to the coordinator for fusion. However, sensor nodes have limited energy and processing resources and transmission is the most power-hungry task. In this paper, we target data reduction and energy consumption. We propose to locally adapt, in real-time, the sampling rate of a sensor node according to the variations in the vital sign being monitored and its risk. We propose to dynamically evaluate, in real-time, the risk of any vital sign given the information about the severity level of the patient's health condition and the severity level of the vital sign itself. We have tested our proposed approach on real health datasets in order to evaluate it. The results show that the percentage of detected critical events and the mean-square error (MSE) are both acceptable. In addition, the percentage of data reduction is around 50% implying a reduction of the energy consumption. Adjusting the risk of a vital sign, over time, ensures the adaptation of the sampling rate according to the overall health condition of the patient as well as the severity level of the collected measurements.