ZWave Data Quality


I don’t think that there’s anything in the standards about there’s anything about the quality of the data provided by any ZWave sensor, but is there any work that’s been done to assess the quality of device types?

I’m now reaching the point where I’m starting to collect meaningful timeseries from collections of sensors in houses and I’m not super-impressed:

  • motion and contact sensors seem to send out both false positives and false negatives (it could be that some messages are missing, too).
  • thermometers offer lots of precision, but not much accuracy (>0.5deg C difference in reported temperatures.
  • there’s no obvious way to identify obvious types of failure from the protocol (eg randomised sequence numbers to spot missing messages, legitimate and illegitimate state and event combinations, heartbeat abstractions).

Is anyone aware of activity in the area of data quality - even if it’s only measurement of sensor accuracy based on real samples, it would be step in the right direction.