Fall detection and reducing detection error using FMCW radar

Elderly people are at high risk of serious injuries from falls and need help quickly. The use of conventional contact sensors or cameras can cause inconvenience to users, so research on fall detection using non-contact sensors is attraction attention. In this paper, we propose an algorithm to classify falls by using Frequency Modulation Continuous Wave (FMCW) radar, and feature to minimize false positives. The proposed feature's performance is evaluated with supervised machine learning, such as Support Vector Machine (SVM) and Linear Discriminant Analysis (LDA).