Characterization of Physical Movement Signatures from WiFi-based Passive Forward Scattering Radar

Abstract

Assistive living technologies has been continuously developed and improved. The use of related devices promotes independence, safety, and improved quality of life. These are relevant for elderly and individuals with special needs who live in private, nursing, and elderly homes. Various type of sensors has since been developed and tested. This study aims to assess the capability of passive forward scattering radar to characterize physical movement signatures. To reduce the cost of bistatic topology, the source of signal for transmitter uses wireless fidelity technology. A total of 144 samples of radar signatures are acquired for walking, bending, sitting, and kneeling. Initially, the pulse-based signal is filtered and undergoes enveloping procedure. Subsequently, the signature of different movements is characterized. The time-series signal is then transformed to Fourier spectrum and power spectral density. The characteristics in frequency domain are analysed, and spectral centroid features are extracted. The distribution of features is visualized through box plot, which indicates good separability between the four movements.