[ax25-layer2] Tower of Babel

[pete at ae5pl.net]

It has been quiet here so I thought I might stir some thought.  There
has been significant discussion here and elsewhere about how this
protocol or that protocol is better, how packet radio is antiquated even
though it has never been truly used solely as a link layer, how this
group or that group wants to use a specific protocol in their area for
specific purposes.  It is interesting that these discussions do not
occur with other communications types that are in use in amateur radio,
just packet radio.  Why is that the case?

Back in the late 1970s and early 80s, amateur radio was given permission
to begin experimenting with digital communications (beyond RTTY).  Our
first stab at it was developing the AX.25 packet.  A well-defined
specification (although not necessarily complete, certainly complete
enough to lay a foundation for amateur digital networks) was released
and then updated in 1984.  Manufacturers and different groups began
making TNCs to support this "new" protocol.  Since there was nothing
available at the time focus on implementing disparate layer 3 protocols
over AX.25, we decided we could do layer 3 ourselves by attempting to
expand AX.25 into layer 3.  We had this tempting feature in AX.25 that
made everyone believe that it was already designed to do this, the
repeater address fields.  Unfortunately for amateur radio, the repeater
address fields are just that: address fields to denote _repeaters_, not
nodes.  There are no other features of the AX.25 header which even hint
at being able to support layer 3 operations within the AX.25 protocol.
We also found out the hard way that using digital repeaters for more
than two hops on the same frequency caused massive network hemorrhaging
and hence network failures.

We also created another "monster" by trying to use the repeater
addresses for layer 3 networking: unlike _every_ other network
implemented by amateurs up to that point, the radio amateur had to know
the network topology and how to use that network every place they went.
AX.25 became a Tower of Babel scenario where everyone implemented their
own style of protocol and anyone who didn't have a solid understanding
of how the local network operated couldn't use it.  This continues
today.

Amateur radio has always been based on getting anyone who is licensed
and has the necessary equipment communicating.  Recently, many areas in
the US have had long, hard fought battles on whether to use subaudible
tones on their repeaters.  Why?  Because it was feared by some that
putting a tone on a repeater somehow limits access.  Because virtually
all radios manufactured of the past 15 years have CTCSS capability and
there are daughter boards available for those which don't have the
capability, proper notification to the repeater community overcomes that
specific argument.  Yes, there are other arguments, but technology and
common sense notification has made the "limited access" argument mute.

So why is it that we cling to the concept that all AX.25 networks are,
by their very definition, limited access and limited use?  I have heard
it said that "only the packet originator knows where his packet is
destined for" and therefore "must be able to define how it gets there".
The first part is true.  The second part is not nor has it ever been
with amateur networks _other_ than AX.25.  This is simply an excuse by
some to try and force limited use of that specific AX.25 network.

What has been proposed on this SIG is to do away (over time) with the
disparate AX.25-based protocols and developed a standardized, generic
protocol specification _and_ implementation which can be used by any ham
that has the proper equipment without regard to how each individual area
has implemented their specific network.  APRS and Winlink Paclink on the
same frequency?  Why not!  Yes, some areas have bandwidth limitations,
but that doesn't mean they should be mutually exclusive, especially
during an emergency.  Can we throw in IP over AX.25?  Absolutely!  The
key here is that individual users should _only_ be concerned with
bandwidth availability, not how to configure their software for every
individual network that might be in their area at the moment.  To
accomplish this, we need to have a basic network architecture that is
focused on supporting diverse layer 3 protocols and providing a common
layer 2 _link_ platform.

Why not just use WiFi?  Or WiMax?  Or whatever somebody else comes up
with?  We can and should where it makes sense (part of the WiFi 2.4 GHz
operation is in the amateur band in the US).  But packet radio provides
something that these newer technologies don't: use of frequencies and
communications speeds which can effectively cover larger areas from both
fixed and mobile stations.  True, you are not going to pump a video feed
down a 1200 bps channel (along with the other uses that share that
frequency), but messaging (_communication_) can certainly occur.

Hopefully this has spurred some thought.  I have seen posts on other
SIGs about "this AX.25 network" and "implementing that AX.25 network"
which show that we continue to see a need, but we also continue to fall
short because each of these networks require that the users of the
network know how to operate in each individual location and with each
individual protocol.  AX.25 is old enough to be used as a communications
medium.  Let's see if we can migrate there.

73,

Pete Loveall AE5PL
mailto:pete at ae5pl.net 

PS: Please do not quote this entire post in your responses.  If you wish
to quote, please restrict your quote to the specific areas you are
responding to out of respect for the people receiving digest versions of
this SIG.  Thanks.