Good luck with that Uwe.
For 2 char messages with 16 bit CRC, the max useful list
length is probably around 1k but I don't yet know where
the optimum is. Beyond a certain length (corresponding
to nearly 3dB gain from the outer code) further list
length increases the false hit probability as much as
genuine decode probability so there is no further gain
to be had (without invoking operator cognition as a
3rd layer to the coding).
A nice new computer arrived last week to help with
that project. I tried hard to find a way to calculate
performance, I even worked out a formula but it had
354 terms and took a day to derive the first term, a
week for the 2nd, and I wasn't brave enough to try for
any more.
So lining up silicon for a brute force evaluation.
Also bought a kite but that's for another thread.
--
Paul Nicholson
--
|
