[ax25-layer2] Re: DTN and LTP

[g0hww at mac.com]

Hi  again,

Maarten and Pete, thanks for replying.  Is this the right list or  
not? I'm not altogether sure.  Where's William of Occam when you need  
him?

If I may indulge myself briefly, having had one response here makes  
me think it may be a good starting point, and that I might risk  
saying some more.

For anyone wondering what this is all about, you could take a look at  
dtnrg.org, an IRTF research group's web-site.  They have defined a  
'bundling' layer, that operates as an overlay network providing  
incremental bundle advancement using store and forward nodes.  It  
runs over any 'suitable' convergence layer below and provides  
endpoint services based on a postal service model.  Nodes are  
typically deployed at the edges between homogenous networks, and at  
the 'tactical edge'.  They have a decent open source reference  
implementation that provides convergence layers for TCP, UDP,  
ethernet and bluetooth too, I think.

In related work, the Licklider Transport Protocol (LTP - http:// 
irg.cs.ohiou.edu/ltp/) has been developed for operation over point-to- 
point links offering delay tolerance suitable for interplanetary  
networking and supporting transfer of 'blocks' having either both red  
data, sent reliably, and/or green data, sent best effort.  Currently,  
I don't know of an open source convergence layer for running DTN over  
LTP, although I believe LTP has a Java reference implementation (I'm  
a C++ propeller-head and Python newbie, but not a Java hacker).

My reason for posting here is that I think development of a  
convergence layer directly supporting DTN in some way over AX.25  
would be pretty straightforward on Linux.  I'm not really interested  
in running DTN over TCP/IP or UDP/IP over AX.25, which I guess could  
be done right now if anyone fancied it.  I'm interested mostly in the  
possibilities of running DTN over point to point  LTP, over point to  
point AX.25, with each node performing store and forward at the DTN/ 
LTP layer.  In this way, red part data could be sent point-to-point  
in connected mode, incrementally advancing when links are available,  
and green part data could be sent in disconnected mode. Link layer  
cuing (i.e. provision of link up/down notifications) is an important  
feature.

I think the disruption/delay tolerant  concept could be useful in  
many situations, such as in emergency operations, or where ham  
licencing restrictions preclude transmission by unattended stations  
which are perfectly capable of receiving incoming traffic.

I also imagine (I do too much of that) the potential for deploying  
DTN/LTP using other modes/links available to hams, with the potential  
for a huge net operating over a multitude of heterogeneous sub- 
networks (including the internet, if desirable) and ad-hoc links.

I have wondered if APRS should be the first convergence layer to  
tackle for ham DTN purposes, having managed to fire some QRP packets  
through the ISS recently, or even one of the hf digital modes, such  
as Olivia.

So, thanks for reading this far.  Interested?  If you have an opinion  
either way, or an interest in taking this elsewhere, please post,  
even if it is only to tell me/us to go away and come back with a  
firmer idea of what we want to do with AX.25.  Maybe I could see if  
the dtnrg would host an amateur list, but got no replies to a similar  
query there, or kick off yet another yahoo group.  I've got a feeling  
that we won't stray too far, though.

Maarten,  I wonder what direction you would like to take.  I've got  
some computing hardware ideas to discuss with you too.  Feel free to  
reply off-line if you prefer.

Apologies for the length of the post.  I haven't been here long  
enough to know if people read this over the air...

Cheers,

Darren, G0HWW