Data stream consistency issue - Need information on how to get precise timestamps

edited December 2018 in General

Hello,

I'm currently using a MetaMotionR device with the Android Metabase application running on a OnePlus One smartphone under the current configuration:

  • MetaMotionR
    Model 5
    Firmware 1.3.7

  • Android phone
    One Plus One
    LineageOS 14.1 - Android 7.1

  • Metabase app version 3.3.3

I'm running a 8 min session(480 seconds) with
1. 50Hz - 2G Accelerometer
2. 50Hz - 500degree Gyroscope sensors enabled

I need to get precise timestamps for the data and I read the previous posts regarding data reconstruction:
https://mbientlab.com/community/discussion/1829/anyone-obtaining-reliable-data-when-streaming-from-metawear-cpro-crashes-and-timestamp-synch
https://mbientlab.com/community/discussion/1934/metahub-timestamps#latest

Here a graph displaying:
x-axis is the elapsed time since starting in milliseconds
1. the data sequence reconstructed with a 20ms interval between each samples,
2. the data sequence reconstructed with timestamp provided by metabase app

During the recording, we stay at 20meters maximum from the mobile phone
There is definitely a loss of 40 seconds in the data(see multiples missing data segments) that may be explained by com issues.

What would be the best way to handle with precision the timestamps (rely on BLE timestamp, reconstructed in some way) ?
Does the FIFO stores some data and send it afterwards ?
Is there a way to get proper internal timestamps in the stream ?

Comments

  • The best way to have accurate timestamps and avoid huge data gaps is to log the data.

  • edited December 2018

    Yes, but what would be the alternatives for long-term recording ?
    We are restricted to 2 hours storage in logging mode

  • edited December 2018
    • [Logging] Sample at lower frequencies
    • Use the accounter to mark the streamed data


      How long are you collecting data for?
This discussion has been closed.