Questions about time synchronisation between devices and sensors

[merschel]

Platform your question concerns:

• Android
• iOS
• Other
• [x ] Platform is not relevant for this question

SDK version:

Customer app url
Link to app store / play store / your website

Device:

• Grit X2 Pro
• Vantage V3
• Vantage M3
• Polar OH1
• Polar Verity Sense
• Polar H10
• Polar H9
• Polar 360
• Other
• Device is not relevant for this question

Description:

Hi,

related to #580, I have a similar problem. I would like to synchronize the timestamps between an H10 ECG signal and the Verity Sense PPG signal. To achieve this, I came up with the idea of using the ACC signal from both devices.

I set the clocks of both devices using the same smartphone, one after the other, via setLocalTime().

Then I start a measurement and hold both devices in one hand. I hit this hand (with both devices on it) against the other hand, which results in a sharp and clear peak in the ACC signals of both devices. I repeat this procedure again after 24 hours. Using these peaks in the ACC signals, I try to calculate the time drift, and this works quite well.

In one experiment, I observed an initial time drift of 622 ms. After 24 hours, the time drift was approximately 2.4 seconds. This means the relative time drift between the two devices is about 1.8 seconds per day.

Based on this information, I am able to synchronize the timestamps of the ACC signals.

Now, I would like to use the time drift information from the ACC signals to also synchronize the timestamps of the PPG and ECG signals. However, this requires that the ACC and PPG/ECG sensors use the same internal clock within each device. Is this the case? Or do they have separate clocks with different time drifts?

My second assumption is that the time drift of each device remains relatively constant over 24 hours. In the discussion #580, I learned that this might not be true, as the drift depends on temperature and battery level.
Would you say that my assumption of linear drift over 24 hours is sufficiently accurate, especially for the Verity Sense, since it depletes its battery completely within that time? Or would you consider the assumption too inaccurate to rely on linear drift for synchronization?

Thanks for your help.