Disagreements
When i look at the published diagrams of the system, the signal from a pitot tube is measured/digitized by an ADM unit and the resulting data sent on a data-bus to the corresponing ADIRU unit. That unit also obtain digital data from one of the static ports and its ADM-unit. Based on these data the ADIRU computes the actual speed. This system is in triplicate.
Resulting three air-speed values from the three ADIRU:s are then sent (in syncronized or random fashion?) to each of the three computers where they are compared before use.
If the data refer to three measurement made at slightly different times, they may differ if the airflow is shifting around the aircraft.
The logic is said to be that each computer tests the differences between these speed values. If they agree within a given limit their median is used. If the difference is larger, the average of the two closest is used instead and the value with the largest deviation is assumed to be false (resulting in an error msg). In case there are three AoA units their data are treated in the same way. When only two AoA:s are in use, their average is said to always be used.
In the special case that the remaining two values also differ by more than a given amount the msg for three invalid air-speeds is generated.
In case that one or more of the ADIRU:s cause the speed error, this message has perhaps nothing to do with the pitot tube, its connection to the ADM, or the ADM unit. The ADIRU:s have selftest programs and should therefore report any internal inconsistency - however this selftesting has not worked as designed in a few cases. The possible cause is still under investigation.
Thus the Airbus system differ from the one used by Boeing, where three ADIRU:s are combined into one unit and the speed comparison seems to be done inside this module.
It is a pity that we don't have accsess to the source codes used nor program flow-sheets.
I hope I have avoided most mistakes in this description.
Regards