Respiraton Module Field Test.

Discussion in 'Respiration Sensor' started by Peter Prodinger, Jul 21, 2015.

  1. Peter Prodinger

    Peter Prodinger New Member

    Hallo everybody.

    Today we tried the XeThru Module X2M1000 on a walker. The goal was to track the person's respiration while pushing the walker. We were aware that we cannot track the person's respiration all the time, because of the movement of the person and the sensor, receiving respiration data from time to time would be sufficient for us. The problem was, that we received data, but most of the time we did not receive accurate respiration data. We got values from 15 to 65 RPM, although the person was breathing steadily with approximately 15 RPM.

    What do you think is the problem?
    Do you think, it is even possible to use the sensor in a setting like we did?
    Where do you think, these very high RPM measurements are coming from?

    I also tried the sensor afterwards placed on my desk which brought more convenient results, but there are still some outliers in the recorded breathing pattern, I cannot explain.

    You can find the recorded .csv files of the “walker-experiment” and the “desk-experiment” attached to this message.

    I would appreciate some input and constructive hints how we could enhance the performance.

    Some more details:
    We used the XeThru Module X2M1000 model and the XeThru Explorer to record the data.
    Distance: 0.5 - 1.20 m
    The firmware is X2M1000-fw-1.0.11-beta

    Best Regards,

    Attached Files:

    Last edited: Sep 30, 2015
  2. Olav Liseth

    Olav Liseth Administrator Staff Member


    The respiration sensor requires a static environment and the walking human is not a valid sensor setup.
    The sensor has different criteria to distinguish breathing from other movement patterns. The walking human results in breathing like patterns and reports false positives.

    I've plotted your "desk-experiment" and have a couple comments:
    1. Seems like the "Include breathing pattern" was unchecked. This results in logging only every second and the breathing pattern is undersampled.
    2. The reported distance seems to vary, and is close to the start of the detection zone at 0.5 meters. Ensure that the breathing person is within the detection zone.

    I repeated your setup, here is the result:
    I walk in at ~0.2 minutes and get a breathing lock at ~0.65 minutes. I increase my respiration rate until a walk out again. I've added the python plotting script to this post.

    Hope this will get you going.

    Best Regards,


    Attached Files: