On Fri, Dec 9, 2011 at 10:10 AM, Remi Chateauneu
<[email protected]>wrote:

> On Fri, Dec 9, 2011 at 4:56 PM, Brian Lloyd <[email protected]>
> wrote:
> > And the Bell 103 modem is a really, really, really, really, really,
> STUIPID,
> > STUPID choice for HF. This is where it would make a huge amount of sense
> to
> > use a different modem with IP, TCP, and UDP layered atop. Just look at
> all
> > the much-better modems exist in fldigi. Why keep using crap like the bell
> > 103? If we had a way of running IP over HF, with or without link-layer
> ARQ,
> > using decent modems with FEC, it would be SOOO much better and also quite
> > interesting!
>
> Does this go to the right direction ?
>
> "FX.25 is a protocol extension to the AX.25 Link Layer Protocol. FX.25
> provides a Forward Error Correction (FEC) capability while maintaining
> legacy compatibility with non-FEC equipment. FX.25 was created by the
> Stensat Group in 2005, and was presented as a technical paper at the
> 2006 TAPR Digital Communications Conference in Tucson, AZ."
>
> http://en.wikipedia.org/wiki/FX.25_Forward_Error_Correction


Well, FEC is the right idea but building it into the network protocol as
opposed to making it part of the modem, doesn't make a lot of sense. (AX.25
is actually an adaptation of a link-layer protocol to be a network protocol
for a datagram-based network, something that annoyed the heck out of people
when Phil Karn pointed this out in the late '80s.) If we were dealing with
a pure AWGN-dominated channel, e.g. space communications, you can get away
with putting it there. It is less so if you are using it on NLOS
terrestrial VHF/UHF/SHF links and even less so on HF links.

So, no, I don't think that is the proper approach. The FEC should be
matched to the characteristics of the MODEM in the medium where the MODEM
is being used. Olivia, MFSK16, and Thor have the right idea. Now we need to
layer a proper ARQ on top of that. And please don't assume it should
necessarily be an antiquated stop-and-wait ARQ with fixed timing like we
have used in AMPR, PACTOR, and AMTOR. Adaptive ARQ is now common AND we
have the option of changing what we think the unit of data transfer should
be.

AX.25, even when used as a point-to-point reliable link, fails badly for
all the reasons that Phil and I outlined in our paper for the ARRL 7th
Annual Computer Networking Conference in 1988. We applied our findings to
running TCP/IP over packet radio (this is where the KISS TNC came from) and
proved just how much more robust packet radio could be.

It was kind of funny. For that effort I was accuse of trying to ruin packet
radio for everyone else. :-) From that I went on to develop PPP.

-- 
Brian Lloyd, WB6RQN/J79BPL
3191 Western Dr.
Cameron Park, CA 95682
[email protected]
+1.767.617.1365 (Dominica)
+1.916.877.5067 (USA)
_______________________________________________
fldigi-alpha mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/fldigi-alpha

Reply via email to