[ax25-layer2] More thoughts

[Bill Beech]

All,

A very lively discussion.  I want to apologize to Bob on my "broad 
stroke" on APRS and digipeating.  Bob took advantage of a capability to 
make the AX.25 work without having to develop a layer 3 protocol.  We, 
as a community, never did do that.

This is a TAPR based group, but I do want to note I still have the 
source for all my TheNet node software and the capability to recompile 
it.  I don't want to get into the discussion about whether it was stolen 
or not (I have good reason to believe it was).  That isn't the purpose here.

We have a clean slate.  X.25 came about in a standard in 1976.  I 
believe that standard generated an X.25 that DID NOT work.  The 1980 
standard was the first that worked correctly.  But it was designed for 
point-to-point wire links with good signal margins.  It was never 
designed to handle the hidden transmitter problem.  It had its own layer 
3 protocol, not IP.  Since then, we have done AX.25.  It addressed some 
of the problems, but in application, it fell short, as well.  And the 
application really drove the requirements - HF, VHF/UHF, APRS, 
TNOS/JNOS/KA9Q.  In the interim, we have a lot of wireless solutions 
that have come to be.  802.11a/b/g and 802.16, as well as the several 
cellphone technologies.  These are technologies that do deal with the 
half-duplex (802 stuff) and hidden transmitter problems we have in 
packet.  The 802.11b standards are available from the IEEE. 

The idea of open source is excellent.  We NEVER had access, as a 
community, to the code that ran on the TNCs.  It was held close by one 
person and that, in my opinion,  killed the development.  TheNet 
provided an alternate, but it was tainted by its history.  We must learn 
from that and keep the development open to anyone. 

I have a single board computer that uses an ARM7 processor and runs 
linux.  It cost $149 in units of one.  This could become a tnc quite 
easily.  The same code, with a different "hardware abstraction layer" 
(read drivers) could run on a notebook or PC. 

The idea of separating the layers gets hard, especially layer 1 and 2.  
This can all be done in software, but some of the interface chips 
provide the layer1 and part of the layer2 functionality.  Layer 3 and 
above are more easily isolated.  So we are really back to defining the 
requirement of the new protocol and applications and then splitting them 
into the layers of the model.  We need to think in term of applications 
as this will drive the protocol. 

I suggest we look at the popularity of PSK-31.  Runs with very low power 
on HF, has no provision to transfer files, and lets hams talk to each 
other.  Gee, wasn't that the goal of packet radio?  Did we get mired 
into to BBSs and junk mail, and miss the boat? 

73,
Bill, NJ7P