Hallway Gait Monitoring Using Novel Radar Signal Processing and Unsupervised Learning

Hallway Gait Monitoring System Using an In-Package Integrated Dielectric Lens Paired with a mm-Wave Radar

“…We have been using a commercially available mm-wave FMCW radar system [ 11 , 20 , 21 , 22 , 23 , 24 , 26 ] (AWR1443Boost) operating at the frequency range of 77–81 GHz. One challenge we encountered was the wide beam of the radar antenna covering a wide range of the surrounding environment creating multipath effects [ 23 ].…”

“…The use of a radar system to monitor gait is appealing due to its reliable functionality during different lighting conditions, the avoidance of having to wear, carry, or charge a device, and protection of privacy [ 7 , 8 , 9 , 10 , 11 ]. CW (Continuous Wave) radars have been explored through various studies to extract gait characteristics [ 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 ], however, an extra device (e.g., a treadmill) is required to obtain both spatial and temporal parameters because the position of the subject could not be obtained by a CW radar.…”

“…To eliminate the multipath reflections, we propose two viable solutions: 1. novel signal processing and 2. radar antenna modification. In our previous work [ 11 ], we put forward a novel radar signal processing and unsupervised learning algorithm to track a person walking in the hallway and distinguish the direct signals from multipath reflections. The results provided in [ 11 ] showed that the proposed association and tracking method could reliably track the walking person and extract spatiotemporal parameters accurately.…”

“…In our previous work [ 11 ], we put forward a novel radar signal processing and unsupervised learning algorithm to track a person walking in the hallway and distinguish the direct signals from multipath reflections. The results provided in [ 11 ] showed that the proposed association and tracking method could reliably track the walking person and extract spatiotemporal parameters accurately. However, multiple transmitters and receivers are required to obtain a range-azimuth heatmap of the environment for single-person monitoring.…”

“…The purpose of this work is to showcase the effectiveness of radar antenna modification in mitigating multipath reflections in hallway gait monitoring without the need for complex signal processing methods. We have been using commercially available low-cost radars that typically have wide beamwidth and suffer from low gain and short-range coverage [ 11 , 20 , 21 , 22 , 23 , 24 ]. In [ 23 ], we proposed a hyperbolic 3D printed dielectric lens to sharpen the radiation beam of an FMCW radar antenna to mitigate multipath reflections.…”