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

• Main Author: Loe, Daniel Kit Leong
• Other Authors: Luo Jun
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence; Engineering::Computer science and engineering::Computing methodologies::Document and text processing
• Online Access: https://hdl.handle.net/10356/156590
• Institution: Nanyang Technological University

In the field of healthcare, a human’s respiration heart rate and respiration rate can play a crucial part in detecting certain cardiopulmonary diseases or interpreting in real time human physical status such as walking, running, sleeping and etc. In current times, this measure of a person’s respiration rate is recorded using specialised contact-based equipment such as Respiration Monitor Belt and Electrocardiogram monitoring systems (ECG). As crucial as these equipment functions are, they too have their own set of drawbacks. Certain scenarios exists where patients/users are not able to wear the physical equipment for long periods of time due to either discomfort from the clunky equipment or in an extreme case, the user is not physically able to put the equipment on such as a burn victim. An intuitive approach would then be needed to solve such a problem. Studies from past research including wearable technology does help to minimize the discomfort however the underlying problem still remains that the user must have the device on for it to function as intended. The main goal of this project is to come up with a model to process, measure and transform the data collected from a radar sensor into human readable respiration rates of users. To build the model, data samples were collected from 4 Nanyang Technological University (NTU) students using both the radar sensor and a respiration sensor simultaneously which was used as the ground truth. Machine learning techniques were then introduced to create the network which accepts the one-dimensional RF data and outputs the human readable respiration rates. Based on the data collected and processed, results suggests that there is a feasible chance of contactless based sensing replacing contact-based sensing equipment in the near future as the model allowed accurate transformation of unknown RF data into human readable respiration rates.