residual, use the full 24-bit syndrome.
Previously this would mask off the low 7 bits, which isn't particularly
great - it doesn't really make sense in that it implies a somewhat
random IID value, and it would actually only correct 20 possible bit
errors out of the 51 possible errors, 16 of which were errors in
the top 16 bits of the CRC itself. It is also risky as it will accept
a larger number of possible garbage messages as the lower 7 bits may take
any value.
Instead, use the full syndrome, assuming that the damaged message
had IID=0 - which seems more likely than assuming a random IID, since
IID=0 is used for acquisition squitters and should be arriving regularly.
The reduced rate of corrections for DF11 messages shouldn't have
much of an impact as DF11s carry very little data themselves - they
are mainly used to acquire the aircraft address. Once one good message
for an aircraft turns up (which would require IID=0 anyway before we'd
accept it) it doesn't really matter if we discard more damaged messages,
as they're not contributing to anything useful in the aircraft state
beyond air/ground status, which is also carried in many other messages.
A "reliable" message is a DF17 or DF11 with good CRC.
A "reliable" aircraft is one that's probably real - we decide it's reliable when
we've seen enough reliable messages, or just enough messages in total, that it's unlikely
to be noise.
DF18 transponders don't necessarily do Mode S, so we shouldn't assume
that Mode S messages from that address are valid merely because we
heard a DF18. If it really is Mode S equipped, we should hear a
DF11 at some point.
The old matching process which tracked mode A values as pseudo-aircraft
got very, very expensive with a large number of mode A/C messages (and
with lots of single-bit errors, which seems common with a Beast doing
the reception)
Instead just count A/C messages directly into a 4096-entry array (which
is very fast) and periodically scan the mode S aircraft list to see if
we can match anything up (which is fixed overhead + cost proportional
to the number of mode S aircraft)
With forced inlining this is about as fast, and it is much less
errorprone than the twisty little maze of handcoded bitshifts that
it was before.
(notably, at least one error - in the ACAS RI field - has been fixed)
This DF is formatted with the first two bits as 1, and the next 3 bits
are part of the message rather than part of the DF code. So effectively
it spans 8 DF values if you are looking at the full 5 bits.
Oliver Jowett
Oliver Jowett
henry1952
Eric Tran
Raphael Geissert