[aprssig] APRS Meshing

[Robert Bruninga]

We used the messh concept back in 1995 or 6 I think
for the SNOW-POLE system  PACcomm built for the
Army for use in Czech Republic.  It is when PacComm
introduced the original callsign substituting TNC's
so that once a snow-pole had digipeated a packet
it would not digipeat that packet again.  (That
was our RELAY system).

Problem was it would repeate all the other copies though
of its own packet digipeated by others.  That is the
RELAY dupe problem that we have finall gotten around
to forcing obsolete in 2004 to avoid all those dupes.

I proposed the WIDEn-N system back in 1994 but it
wsa not until 1998 that Kantronics put it in their TNC's.
It is a perfect MESH system.  Every station digipeats
it once and never again.

Bob, WB4APR

>>> clementrw at hotmail.com 01/05/06 12:09 AM >>>
I am testing the viability of using APRS in a non-fixed station environment
- eg. no established digipeaters and no base stations.  The intention is to
use a number of lightweight, portable radios such as the TH-D7a to create an
ad hoc mesh that is survivable even if some radios fail.  I have worked with
trying to create a mesh network - in one case with nine radios using WIDE7-7
to ensure that as many radios copied the signal as possible.  My reasoning
for this was that if each radio was almost in the fringe area of its
neighbor radios and the radios were in a line, for N radios it would take
N-2 hops for an end radio's data to traverse the network to the other side.
Obviously, this creates many unneeded duplicate packets in less extreme
cases, and I am wrestling with ideas on how to solve that issue.  One method
I have come up with would be rather more software intensive:
 
1. Remember your immediate neighbors.
2. For a given packet, choose a random wait time.  If you haven't heard
every neighbor digipeat the packet after a given wait time, digipeat.
3. Maintain packet in memory for a fixed, user settable amount of time to
ensure that you do not digipeat it again.
 
I still see problems with the above protocol, although it would cut down
some on packets.
 
I have tried researching the spec to see if the meshing concept is covered,
but all I have seen so far assumes that there are fixed stations and
digipeaters.  Is there a part of the spec I have missed that addresses this?
Has anyone else worked on a similar experiment?  Thanks for your help.
 
73,
Richard Clement (AE6QE)