A real IoT device deployment for e-Health applications under lightweight communication protocols, activity classifier and edge data filtering

Abstract Recently the topic of health monitoring is growing in interest and several applications and devices were designed and put on the markets. This interest is promoted by an increasing enthusiasm about physical activities to improve health and wellness. In this work, we present a wearable device able to collect data from different body sensors to send them towards a cloud platform. The Message Queuing Telemetry Transport MQTT communication protocol is used to send data by customizing the messages payload and building up a customized protocol between devices and cloud. These data, are used to recognize user activity and health status sending alerts when an anomaly is detected. These results can be used to monitor performances or to investigate anomalies discovered during activities. In addition, we propose a fuzzy-based Human Activity Recognition (HAR) classifier which continuously acquire data from body sensors exploiting an Internet of Things (IoT) architecture. In order to improve HAR, different classes of data filters have been analyzed to reduce the amount of data sent to the cloud. These filters introduce smart functionalities at the edge of the network reducing the data-flow without compromising the classifier goodness. A smart device based on embedded technologies is designed to realize the wearable device.