Field trial results using a novel integration of unique millimeterwave Doppler radar for high performance non-obtrusive life sign (breathing and heart beating) monitoring of high suicide risk prisonner in observation cell

Abstract

This paper presents the fields results in non-obtrusive life sign monitoring performed using a low emission, compact high frequency Doppler radar, which can detect body movements associated with breathing and the heart beating. Such a system is particularly useful where attaching sensors to the body is undesirable; for example, monitoring detainees in a prison who are at risk of suicide, self-harm, and medical complications due to drugs or alcohol. This method of monitoring can aid greatly in situations where it can be difficult to ascertain a person's status, such as a person who is sleeping under a blanket. While a breathing and cardiac signal is visible in the Doppler radar signal, it can be corrupted by noise and interference by so-called nuisance signals (e.g., movements associated with fans, water flowing from faucets and toilets, door micro-motion, light fixture ballast vibration and other body movements). This work is part of a three-year project that involves four organizations; CSC (end-user), KG Spectrum (radar-based perimeter intrusion detection systems), Carleton-University (bio-medical) and Ottawa-University (signal processing). The final goal is to develop a robust system for detection of attempted suicide events in prisons in time to allow for lifesaving interventions. This has been done in a novel way by installing one high range (75cm) spacial resolution, scanning antenna and high frequency (24.125GHz) radars in prison cells and by processing signals extracted from the radars in real-time. This unique radar architecture allows the usage of novel signal processing and pattern recognition algorithms to locate the subject and removing interference and reliably estimate breathing and heartbeat signals, even when the subject is non-stationary and then to produce an alarm when these signals cannot be observed or significant changes, in breathing pattern or heart rate pattern, representing abnormal behavior have been detected within the observation cell. This research project is performing time, frequency and pattern domain analysis on the radar data and explores signal processing approaches for the robust and accurate detection and estimation of the respiratory with heartbeat rate. Outcome of this research will be useful in mitigating the risks associated with detainees by providing a life sign monitoring approach that can help enable timely responses. Such a system can also find application in smart health homes for monitoring people at risk, such as the elderly or infants, as well as in psychological institution. In addition to presenting the results of filed trial, this paper covers the innovative engineering aspects of using high frequency, high range resolution and high sampling rate as a mean to geo-localize the source of movement within the prison cells, thereby increasing the efficiency of the signal processing eliminating the need to process on overall volume wise integrated Doppler signal. The use of geo-localization provides natural discrimination of source of nuisance, allowing the signal processor focusing on Doppler signal solely produced by the inmate heart beat and breathing responses.