Signal Processing and Speech Communication Laboratory
homeresults of the month › Variational Message Passing-Based Respiratory Motion Estimation and Detection Using Radar Signals

Variational Message Passing-Based Respiratory Motion Estimation and Detection Using Radar Signals

Published
Thu, Dec 01, 2022
Tags
rotm
Contact

We apply an variational message passing scheme in order to detect the presence of children im parked cars using multistatic UWB radar.

To detect a person in a car, we use a structured mean field approach an apply variational message passing to maximize the ELBO, a lower bound on the model evidence. The ELBO is then used to calculate the odds-ratio of the two cases (the car being either empty or occupied). During the inference process, the radar channel and the respiratory chest motion of the target ar estimated, in order to coherently add up all of the energy from the target present in the received signal. Therefore, we make not only use of the direct interaction of the target with the transmitted signal (line-of-sight, LoS), but also of the multipath components (MPCs) that bounce around in the car before interacting with the target, which increases the SNR. Since the respiratory chest motion is estimated in the process, this message passing scheme can also be used for contact-free vital sign monitoring e.g. in a hospital setting.

This work was submitted to the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing. If you are interested have a look at the arxiv preprint.

Browse the Results of the Month archive.