Data processing using edge computing : a case study for the remote care environment

Abstract

Falling is one of the most common concerns among caregivers. For people with dementia and the elderly in remote care and hospitals, immediately informing caregivers of abnormal behaviour such as a fall can improve their quality of life. Latency occurs when processing massive amounts of continuous data from the Internet of Things devices in the cloud. Network latency impacts latency-sensitive critical real-time applications, such as those used in the healthcare sector. This study seeks to reduce latency and network bandwidth when sending continuous data from wearable sensors in a remote care environment in order to meet the latency requirements of health applications. To reduce latency and network bandwidth, a framework is proposed that deploys edge computing using a geo-distributed intermediate layer of intelligence in the middle of the sensor and cloud layers. It includes raw collected data processing, early sensor fusion, missing data, data reduction and conversion and data storage. The case study is a remote care environment focused on fall detection. The research focuses on fall detection and analysis of sensor data for human fall detection using various activity recognition techniques, threshold-based and Machine Learning algorithms. As a result, a fall activity recorded from the wearable device to the edge server could be processed, predicted, and reported to the caregiver in 294 milliseconds.