Realization of radar-based fall detection using spectrograms

Radar has emerged as a leading technology supporting large sectors of commerce, defense and security. Enabled by the advent of small, low-cost solid-state and software-defined radar technologies, new radar applications involving cognitive radar, medical and biometric radar, passive radar, and automotive radar have been made possible. In this paper, we examine redundancy in human motion signatures along the data and short-time Fourier transform (STFT) parameters. With an "eye" on a final product, we evaluate the effect of reduced sampling along slow-time on classification performance. The goal is to determine the degree of data down-sampling that can be tolerated without compromising feature extraction or significantly impeding motion classifications. We search for the optimum STFT parameters that provide the best classification performance for the given radar measurements and gain an understanding of their respective nominal range values.

[1]  Branka Jokanovic,et al.  Effect of data representations on deep learning in fall detection , 2016, 2016 IEEE Sensor Array and Multichannel Signal Processing Workshop (SAM).

[2]  Haiping Lu,et al.  MPCA: Multilinear Principal Component Analysis of Tensor Objects , 2008, IEEE Transactions on Neural Networks.

[3]  Moeness Amin,et al.  Radar for Indoor Monitoring: Detection, Classification, and Assessment , 2017 .

[4]  Jieping Ye,et al.  GPCA: an efficient dimension reduction scheme for image compression and retrieval , 2004, KDD.

[5]  Boualem Boashash,et al.  Time-Frequency Signal Analysis and Processing: A Comprehensive Reference , 2015 .

[6]  Haiping Lu,et al.  Multilinear Subspace Learning: Dimensionality Reduction of Multidimensional Data , 2013 .

[7]  Lillemor Lundin-Olsson,et al.  Circumstances surrounding falls in patients with dementia in a psychogeriatric ward. , 2009, Archives of gerontology and geriatrics.

[8]  Abdesselam Bouzerdoum,et al.  Continuous-wave doppler radar for human gait classification , 2017 .

[9]  Branka Jokanovic,et al.  Multi-window time–frequency signature reconstruction from undersampled continuous-wave radar measurements for fall detection , 2015 .

[10]  N. Altman.  An Introduction to Kernel and Nearest-Neighbor Nonparametric Regression , 1992 .

[11]  Adam Thrasher,et al.  Gait changes with walking devices in persons with Parkinson’s disease , 2012, Disability and rehabilitation. Assistive technology.

[12]  H. Wechsler,et al.  Micro-Doppler effect in radar: phenomenon, model, and simulation study , 2006, IEEE Transactions on Aerospace and Electronic Systems.

[13]  J. Stevens,et al.  Circumstances and outcomes of falls among high risk community-dwelling older adults , 2014, Injury epidemiology.

[14]  Jin Jiang,et al.  Time-frequency feature representation using energy concentration: An overview of recent advances , 2009, Digit. Signal Process..