[nos-bbs] Remote AX25

[Barry Siegfried]

[wa7nwp at jnos.org wrote]:

[snip]

> > > Anybody thinking about dynamic AXUDP setups ? ...
> >
> > I don't know what you're getting at.  Please explain.
>
> With conventional AXUPD, both sides have to manually set up
> the circuit.  Suppose it was automatic on the side of the
> server.  Then anybody could log in and establish a circuit.
> The return path would have to be set up automatically to the
> station logging in.  (Like a TCP connection...)
>
> Today it's a one to one relationship.  If I set up an AXUPD
> connection to a server and that server went away, I'm out of
> luck until I contact somebody to set up a new link.  It would
> be good if we could generalize that.  Imagine a page of AXUPD
> ampr connections similar to the port 14501 of APRS servers
> resulting from a similar login scheme.

The original code to implement the AXIP specification in RFC1226
(which AXUDP follows) was designed so that each side of an
AXIP(UDP) link would have prior knowledge about who its AXIP(UDP)
partners are.  The idea was that a machine should have prior
knowledge of the IP address of any peer that is sending IP(UDP)-
encapsulated AX.25 frames to it before it will accept these frames.
I'm not sure *why* that restriction was put in, but nevertheless
it was.

I am not personally familiar with the APRS login sequence on TCP
port 14501 (since I don't personally use it), but there is nothing
that can't be done if somebody wants to design an algorithm for it
and both ends of a proposed link agree on the code to implement the
algorithm.  Rather than a "login" per se, such an algorithm could
be modeled after the dynamic gateways route advertising code where
a port (and possible password) would be defined to send a "broadcast"
of an IP address which would then cause the installation a new
AXIP(UDP) virtual interface on the machine that received the
advertisement which points back to the advertiser.  As part of the
process, there could also be a timer installed on the machine that
received the advertisement to cause the interface to automagically
"detach" if the advertising end of the link goes away (much the way
it works with IP routes on "hosting" gateways to dynamic gateways).

This is just an idea.  I have no actual intentions of coding such
an algorithm but Maiko would probably be able to answer whether or
not this would be simpler or more complicated than an TCP-style
APRS login sequence?

73, de Barry, K2MF >>
           o
          <|>      Barry Siegfried
+---------/-\---------------------------+
| Internet | bgs at mfnos.net              |
| HomePage | http://www.mfnos.net/~bgs  |
+----------+----------------------------+
| Amprnet  | k2mf at nnj.k2mf.ampr.org     |
| PBBS     | k2mf at k2ge.#cnj.nj.usa.noam |
+----------+----------------------------+