ABSTRACT
This chapter focuses on the former, that is, narrowband radar, which is often also referred to as Doppler radar. It provides the radar signal model and introduces several time–frequency analysis techniques that can be used for human fall detection. The chapter presents the radar Doppler signatures from various kinds of human fall and non-fall motions. It describes a machine learning approach for fall detection that involves prescreening, feature extraction, and fall versus non-fall classification. The chapter reviews experimental results and elaborates on the benefit of fusing radar and infrared sensor measurements to improve fall detection. Fall detection performance evaluated using datasets collected from experiments and real-world operations was presented. In addition, motion sensor network using low-cost passive infrared (PIR) sensors may supplement radar sensors to enhance fall detection and reduce false alarms. Such sensors have been placed at Tiger Place Apartment, Columbia and Missouri, to report the absence or presence of the resident at a certain location in the home.
