Radio-frequency (RF) sensing for deep awareness of human physical status

• Main Author: Ang, Adrian Jun Hao
• Other Authors: Luo Jun
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Wi-Fi sensing; Machine learning
• Online Access: https://hdl.handle.net/10356/181110
• Institution: Nanyang Technological University

This work has been motivated by a significant adoption of WiFi Sensing technology. It utilizes standard WiFi signals for capturing subtle changes in the environment, such as movements, which is ideal for application in health and localization. Specifically, Channel State Information (CSI), a critical parameter in WiFi systems, can capture signal variation due to environmental interactions. Potentially enabling non-invasive real-time health monitoring and indoor positioning. This motivates the project on breathing rate estimation and deep learning to determine localization. This project further investigates advanced signal processing mechanisms like phase correction, subcarrier selection, and noise filtering to extract respiration signals from CSI data. Thereafter, pre-processing will be performed, and it will be used to train and classify into locations using a Convolutional Neural Network (CNN). Despite some challenges, the project highlights the potential of integrating wireless sensing with deep learning for health monitoring and indoor localization, contributing to the broader exploration of wireless sensing technologies. Future improvements in CSI data processing and deep learning algorithms will render better system accuracy as well as robustness.