WiFi-based respiration monitoring system based on deep learning and Internet of Things

• Main Author: Li, Jingyuan
• Other Authors: Xie Lihua
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering; WiFI; Channel state information; Signal processing; Deep learning; Random forest
• Online Access: https://hdl.handle.net/10356/177019
• Institution: Nanyang Technological University

In this project, I explore the use of Wi-Fi signal’s Channel State Information to investigate the disrup- tions caused by human motion on the signal propagation path. This analysis enables the estimation of human respiratory rates using Wi-Fi signals and make it possible to develop a non-intrusive, accurate, and user-friendly solution for monitoring respiratory rates using CSI derived from Wi-Fi signals. The project includes the development and discussion of various signal processing algorithms aimed at noise reduction and filtering to minimize error rates in the CSI data analysis. Additionally, a shallow Convolutional Neural Network is trained to di↵erentiate between the presence and absence of people based on CSI data. Moreover, the project includes the construction of an HTML based front-end for CSI data visualization, interaction and the results of the algorithmic processing. The system demonstrates the practical application of Wi-Fi signals beyond traditional communication purposes, showcasing potential in health monitoring and ambient sensing domains.