Entropy-Based Human Activity Measure Using FMCW Radar

TL;DR

This paper introduces a new radar-based method to measure human activity more accurately and privately than existing tools.

Contribution

The novel approach uses both distance and velocity data from FMCW radar and applies Shannon entropy for improved activity estimation.

Findings

The proposed method achieves lower RMSE and higher correlation with ground truth data compared to conventional methods.

Using both distance and velocity information enhances measurement accuracy for activity estimation.

FMCW radar shows potential for non-contact and unrestricted human activity monitoring.

Abstract

Existing activity measurement methods, such as gas analyzers, activity trackers, and camera-based systems, have limitations in accuracy, convenience, and privacy. To address these issues, this study proposes an improved activity estimation algorithm using a 60 GHz Frequency-Modulated Continuous-Wave (FMCW) radar. Unlike conventional methods that rely solely on distance variations, the proposed method incorporates both distance and velocity information, enhancing measurement accuracy. The algorithm quantifies activity levels using Shannon entropy to reflect the spatial–temporal variation in range signatures. The proposed method was validated through experiments comparing estimated activity levels with motion sensor-based ground truth data. The results demonstrate that the proposed approach significantly improves accuracy, achieving a lower Root Mean Square Error (RMSE) and higher correlation with ground truth values than conventional methods. This study highlights the potential of FMCW radar for non-contact, unrestricted activity monitoring and suggests future research directions using multi-channel radar systems for enhanced motion analysis.

Figures