[Packet] Decode of incomplete packets?
Dr. Gerald N. Johnson
geraldj at netins.net
Sun Dec 23 01:25:40 EST 2012
Monitor mode in my hardware TNCs does that. Its really difficult because
the data is a synchronous string the only way to align characters all
the way through is to detect the leading sync characters at the
beginning. Then the tnc firmware has to keep receiving bytes in the
right order until the end of the packet which has a check sum. Sometimes
a few crashed bits make the recognition of the sync bytes fail, then
there are no start and stop bits in the data string to synchronize
characters and every byte is wrong. If the leading sync character(s) is
destroyed, there's no hope for the rest of the packet. The format of the
system allows for a connected station to request packets be resent.
That's the error correction mechanism. Sometimes there are many
collisions destroying second and third frames when the first frame was
OK. Some early TNCs in radios can only handle MAXFRAME of 1. They don't
have enough buffer for several frames and will blat a response packet in
frame 2. Early software TNCs in slow PCs did that too.
The only way that I can think of receiving data with bad sync characters
would be to capture all the bytes, string them end to end and shift all
the bits (like about 700 for an 80 character transmission) and shift
until there's detectable text in the result. That's hard to do realtime
and has to be in the firmware. I don't know of any that does that
because even though the text string might be the language spoken by the
users, without sync characters at the beginning the check sum isn't
going to compute correctly to allow the data to be accepted.
73, Jerry, K0CQ
On 12/22/2012 10:35 PM, KD7JYK DM09 wrote:
> Is there something out there that can decode, as best as it can hear,
> incomplete or corrupt packets?
>
> Sometimes I get bits and pieces of a packet. I may lose part for whatever
> reason, but, is there a way to get what little data it contains, even if
> only a few good characters?
>
> Kurt
>
More information about the Packet
mailing list