Application of Micro-Doppler Signatures for Estimation of Total Energy Expenditure in Humans for Walking/Running Activities

• Published in: IEEE access Vol. 4; pp. 1560 - 1569
• Main Authors: Youngwook Kim, Choudhury, Songita, Hyoun-Joong Kong
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2016; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Correlation analysis; Doppler effect; Doppler radar; Energy efficiency; Estimation analyzer; Gas analyzers; Human activity monitoring; Human factors; Human motion; Human performance; indirect calorimetry; Legged locomotion; micro-Doppler; Radar signatures; Regression analysis; regression model; Regression models; Running; total energy expenditure; Treadmills; Volume measurement; Walking; micro-Doppler; regression model; total energy expenditure; Human activity monitoring; indirect calorimetry
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2016.2547948

We investigate the feasibility of estimating the total energy expenditure (TEE) of a human for walking/running activities with micro-Doppler signatures. Doppler radar can capture micro-Doppler signatures produced from limb motions when a human moves. As the micro-Doppler signatures contain information regarding limb movement, TEE can be estimated by analyzing the Doppler spectrogram. To understand the relationship between the TEE and micro-Doppler signatures, basic arm and leg motions are measured by the Doppler radar, whereas a respiratory gas analyzer measures the volume of exchanged respiratory gas (O 2 and CO 2 ) to obtain a reference TEE. The area of micro-Doppler signatures in a spectrogram has been suggested to serve as key information to estimate TEE. For the verification of the suggested approach, TEE was measured for seven human subjects, who performed walking and running activities on a treadmill at five different speeds using both the Doppler radar and the respiratory gas analyzer. We confirm that strong correlations exist between the micro-Doppler area and the TEE. Finally, a regression model for walking and running activities is developed for a person. Then, the model calculates the TEE under two scenarios, and we find that the estimation errors are 13.2% and 12.3%.