Sleep Anomaly Events Detection Based on Adaptive Double Threshold CFAR Detector Using UWB Radar

Detection of sleep anomaly events, such as get up, sit up and turn over, is one of the most important tasks in human sleep monitoring. In this paper, we propose a novel sleep anomaly events detection algorithm based on an adaptive double threshold constant-false-alarm-rate (ADT-CFAR) detector using ultra-wideband (UWB) radar. The proposed algorithm firstly removes the static clutter, detects the human and performs the time-frequency transform for the received signals. By extracting the power burst curve (PBC) from the obtained time-frequency maps, we can formulate the sleep anomaly events detection by a binary hypothesis test and then develop an ADT-CFAR detector, which is based on log-normal distribution. A series of experiments are conducted and the results show the effectiveness of the proposed detector.