The output from a digitizer is the following set of numbers:
Is there a way to tell if the true value is 2343, with a noise spike giving 2342 or a true value of 2342.8?
From just this set of data, there is no way to tell the difference between the two scenarios. Taking data over a prolonged period of time might reveal
- that the spike to 2342 happens only every few hundred samples, at which point it's probably noise
- that the spike to 2342 happens periodically, in sync with the power line or some external noise source, and so is deterministic error that can be ignored
- shows up at random points in the series of measurements but, on average, 1 time out of 5, in which case the true value is 2342.8
Until we know more about noise behavior of data converters, there's no way to definitively answer this question!