Contact-Free Human Posture Estimation: an Analysis of the Influence of mmWave Array Polarization

In this paper, the millimeter-wave (mmWave) multiple-input multiple-output (MIMO) array aperture is studied for the capability in estimating human postures in a contact-free manner. A mono-static array with Mills cross topology is utilized for transmitting and receiving the scattered fields from a human body. The 3D image of the human body in the radio “vision” of the array is thus synthesized using the back projection algorithm. In order to conduct a quantitative evaluation of the estimated target geometry associated to postures, the 3D image moments are calculated as the performance metrics. Furthermore, the influence of circular and linear polarization of antenna arrays on generated images are demonstrated. It is shown that linear polarized antennas can capture details of body in simple postures as in standing, however circular polarized antennas can perform satisfactorily in more complex postures as in the sitting. The investigation in this paper is instructive and transferable to specific hardware use cases in practice.