Radar Signal Processing for Elderly Fall Detection

Radar is considered an important technology for health monitoring and fall detection in elderly assisted living due to a number of attributes not shared by other sensing modalities. In this paper, we describe the signal processing algorithms and techniques involved in elderly fall detection using radar. Radar signal returns from humans differ in their Doppler characteristics depending on the nature of the human gross-motor activities. These signals are nonstationary in nature, inviting time-frequency analysis in both its linear and bilinear aspects, to play a fundamental role in motion identification, including fall features determination and classification. The paper employs real fall data to demonstrate the success of existing detection algorithms as well as to report on some of the challenges facing technology developments for fall detection.