Is it possible to take the “only report deviant values” approach? If the timing is set, should be on the nose and deviations show issues why not log successful timings as good and use your backend to sub the numbers based on a solid start or stop time. When the deviation appears THEN you shoot out all of the particulars. You still have to measure local timing but I would think that a timestamp_start>Good would be smaller than all of the particulars which in essence mean the same thing? It may not apply here but thought I would throw it out there.
@fotoKrille, as both @bko and @LukeUSMC point out, you could simply do exception reporting or statistical reporting to reduce data bandwidth. However, I still don't quite understand your data collection. You mention:
and then:
These use wildly different capture methods and capture times! For the photosensor capture, the Electron can easily do the timing using interrupts and the millis() or microseconds() counters. The Electron could calculate a meaningful set of statistics at 1 second intervals (based on Time.now()) and queue the values into a fixed size payload (100KB for example) to upload at regular intervals.
As for the CANbus or Serial data, that would further discussions to understand the requirements.
1 Like
Yes, the photosensor if for creating a timestamp and the capture part is for what data did we have under that cycle.
{
deviceID: 'abc123',
timestamp: '2016-01-25 12:34:56:250',
data: [{
sensor: 'TEMP1',
value: 120
},{
sensor: 'TEMP2',
value: 38
},{
sensor: 'Cylinder_psi',
value: 10
}]
}