Internet of Things (IoT) Privacy–Protected, Fall-Detection System for the Elderly Using the Radar Sensors and Deep Learning

摘要

The increase in the number world population of elderly citizens, as well as those who live in solitude, needs an immediate solution with an intelligent monitoring system at home. In this work, we present an intelligent fall-detection system based on IoT to monitor the elderly with your privacy-protected. Currently, fall detection has attracted significant research attention and deep learning has shown promising performance in this task using conventional cameras. However, these traditional methods pose a risk of the leakage of personal privacy. This work proposes a novel fall-detection system that uses a continuous-wave Doppler radar sensor to acquisition the elderly movements and sends this information thought the internet to a server with deep learning using a convolutional neural network (CNN) that identifies the fall. The radar sensor is inexpensive, completely camera-free, and collects no personally identifiable information, thereby allaying privacy concerns. Additionally, unlike traditional cameras, it has environmental robustness and dark/light-independence. The proposed system obtained 99.9% accuracy in detecting falls by using the GoogleNet convolutional neural network. The proposed system is also capable of detecting other types of movements in addition to those tested, including the detection of diseases such as COVID-19 through the cough movement.