Inference System of Body Sensors for Health and Internet of Things Networks

Abstract

Wearable devices have become popular and innovative and are converging with technologies such as big data, Cloud and Internet of Things (IoT). Traditional physiological sensors in fitness tracking and mHealth provide health data periodically or are captured manually when required. In future, physicians as well as IoT devices will benefit from this data to provide their services. These situations can cause rapid battery consumption, consume significant bandwidth, and raise privacy issues. There have been many attempts to extend battery life and improve communication methodologies; however, they have not been able to solve the resource constraints arising from physical hardware limits, such as the size of sensors. As an alternative, this paper presents a novel approach and solution to controlling body sensors to reduce both unnecessary data transmission and battery consumption. This can be done by implementing an inference system on sensors using sensed data to transfer it efficiently to other networks without burdening the workload from IoT onto sensor devices. In this paper, we experimented with reducing the bandwidth requirements for heart-rate sensors. Our results show savings in resource usage of between 66% and 99%. Such savings have the potential of making always-on mHealth devices a practical reality.