Hi JD-EE. I think you have overcomplicated the situation a bit. There are smoothing functions after the initial reading of the sensors. The data from the 3 airspeed channels are compared for consistency and must be within X knots of each other to be considered acceptable. The 3 channels are then averaged in some manner and that number is used as the airspeed. If any channel is out of the acceptability window of x knots for y seconds, then it is rejected and the remaining two channels are used as the basis for airspeed. If those remaining two channels then fail a consistency check, that is when all heck breaks loose.
The system is going to hang together for a little while before data inputs begin to get rejected (at least y seconds).
I'll have to dig out the reference for airspeed data processing again and post it.