Radar-Based Soft Fall Detection Using Pattern Contour Vector

The Internet of Things (IoT) technologies reserves a large latent capacity in dealing with the emerging fall detection problem of elder people. The radar-based IoT methods are considered one of the optimum solutions to indoor fall detection problems. In this article, a millimeter-wave frequency modulated continuous wave (FMCW) radar-based fall detection method using the pattern contour vector (PCV) is proposed. The soft fall motions, which were not considered in most previous literature, are studied and analyzed. The motion attributes of velocity, intensity, and trajectory can distinguish sudden and soft fall motions from nonfall ones. PCVs of Doppler time (DT) map (DT-PCV), regional Power Burst Curve (rPBC), and PCVs of range time (RT) map (RT-PCV), interpreting the aforementioned attributes, respectively, are used as the inputs of the two convolutional neural networks (CNNs). The experimental results show that the proposed method can detect sudden and soft fall motions with high accuracy, sensitivity, and specificity.

Automated summary

This article proposes a millimeter-wave FMCW radar-based fall detection method using the PCV. It uses DT-PCV, rPBC, and RT-PCV as inputs to two CNNs, and can detect sudden and soft fall motions with high accuracy, sensitivity, and specificity.