Radar-based fall detection based on Doppler time–frequency signatures for assisted living

Falls are a major public health concern and main causes of accidental death in the senior U.S. population. Timely and accurate detection permit immediate assistance after a fall and, thereby, reduces complications of fall risk. Radar technology provides an effective means for this purpose because it is non-invasive, insensitive to lighting conditions as well as obstructions, and has less privacy concerns. In this study, the authors develop an effective fall detection scheme for the application in continuous-wave radar systems. The proposed scheme exploits time-frequency characteristics of the radar Doppler signatures, and the motion events are classified using the joint statistics of three different features, including the extreme frequency, extreme frequency ratio, and the length of event period. Sparse Bayesian classifier based on the relevance vector machine is used to perform the classification. Laboratory experiments are performed to collect radar data corresponding to different motion patterns to verify the effectiveness of the proposed algorithm.