Svarin, #848
Do you mean subsystems inside one given computer/unit, inside one PRIM or
inside one ADR for example ?
I'm thinking about something that has one function. An adr is probably a
good example, as it's function is simply to measure air data parameters
and send the results elsewhere. It's functionality is quite limited and
the only thing the internal software must do is produce accurate results
when the inputs are within range and an error message when it finds that
the inputs exceed the limits of the measurement hardware. That may be a
little oversimplified, but it's basically a box that converts barometric
pressures to electrical signals that other parts of the system can
understand. Because of the limited functionality, it's not too difficult
to devise tests that provide a high degree of confidence that the thing
is working to spec.
The complexity problem arise when many of these individually reliable
boxes are connected together, as you then have time as an added variable
thrown into the mix. Depending on conditions, the various boxes send
their messages with slight variation in timing, so that, for example a
data or error message may arrive at one box before, or after, a message
from a third, or fourth. The testing and validation problem is one of
how to model or prove the system correct when timing variation, together
with all possible system operating and failure modes, probably gives
millions of possible combinations / sequences to test against.
You might compare an adr to a fiddler in the local pub, while a complex
interconnected system is more akin to a symphony orchestra. If the
orchestra is playing Bach, then you're probably ok, so long as everyone
is in tune and in sync. Jazz, and you might be in big trouble :-)...