> Since the
> radios use XON/XOFF flow control, those values are obviously out.

ekk! You'd have thought that they would have worked around that.

> D-RATS does compression and error detection.  At 900 baud, FEC starts
> to eat up your channel pretty quick.  For most connections, framing
> and checksums provide a good balance of speed and overhead.

Yes, in relation to the size of the actual data sent the FEC will be about the 
same size (of extra bytes) for short messages.

However there are other 'costs' in having resend corrupt data.
1) The message coming back to say retransmit the data or part of it (as D-Rats 
does).
2) Waiting for the channel
3) 'Warm up'/TX wait time on the transceiver
4) The header (48 bytes), plus the 72:24 bit(or byte?) ratio of voice to data 
and any sync frames (1st and every 21st)
5) End sequence
6) Transmit tail

With all that it might be worth it. Taking the SMS example (which is up to 160 
characters/140 bytes) would require (approx.) 68 bytes of FEC.

Of course not every transmission/block is corrupt, and the 'typical' corruption 
might actually prevent FEC from working at all (although it seems to work for 
the Audio). The effects of interference would (could?) be lessened if the data 
and FEC where spacially spreed in the transmitted sequence.

The problems will obviously be more apparent as more people start using the 
D-Star channels/repeaters and as they become 'busy', but also as people's 
expectation of range increases (and thus meaning a higher percentage of failed 
transmissions/blocks).

Anyway, it's an interesting topic to discuss...
Simon.


Reply via email to