Radar sensors for smart human activity detection and monitoring

Due to the non-contact and non-invasive features of radar sensor, researches on using radar sensors for human activity detection and monitoring have been carried out for decades. Different types of radar were used on varies applications. Study results showed the capability of using radar sensors for different applications from security applications, smart home, to senior care. On the other hand, radar micro-Doppler and range-Doppler signatures are two popular tools in human activity classifications. Both of the signatures are widely employed in activity recognition studies. In this dissertation, radar sensors for three different human activity detection and monitoring applications have been studied. The first one is using multi-radar system for directional identification. A method for identifying the direction of a certain activity using radar micro-Doppler signature was proposed. The results showed the capability of using multi-radar system for obtaining directional information. The second research work is using radar sensor for respiration monitoring. In this topic, experiments on monitoring infants’ respiration were carried out. A dummy baby and a respiration simulation system were used to simulate the respiration under different scenarios. An automatic DC correction algorithm was proposed to realize movement-immune respiration monitoring. The third study is on using radar sensor for detecting a potential active shooter with concealed rifle/shotgun. In this study, both micro-Doppler and range-Doppler signatures were used. To test the radar’s detection capability, seven other activities similar to a person walking with a concealed rifle were studied. The classification results demonstrated the feasibility of using radar sensor to help identifying a potential active shooter with a concealed rifle/shotgun.
Further more, in order to achieve low-power, low-cost and small size features of radar sensors, a K-band on-chip radar sensor design is introduced in this dissertation. The radar sensor chip was designed in Global Foundry (GF) 180 nm process and adopted direct conversion radar architecture. DC coupled structure was employed and baseband DC tuning circuits was designed to tune the DC level to a proper range. Experiments were carried out using the fabricated radar sensor to sense a periodic signature generated by an actuator.