Order Tray | Contact Us | Home | SIG Lists

[aprssig] ideas for a simple digi

Chris Kantarjiev cak at dimebank.com
Tue Mar 1 16:50:14 UTC 2005


> 
> Hopefully any digi that terminates long paths will have an
> IGate, so the packets are not lost.  The two should  be
> considered a synergistic pair.   Bob

I understand this intent, but I think it's not appropriate
in the scenario I'm painting - which, at its simplest,
is a star network with a HIGH digi in the middle and these
simplified digis at the edges.

There's no point for the edge digis to repeat a packet that's
already been repeated in such a way that the HIGH digi
might choose to repeat it again. You might counter that the
HIGH digi should be "smart", but maybe it isn't. (It seems
that lots aren't.) I'm trying to cut the QRM.

I came up with a couple of concerns with this scheme. One seems
like pure ALOHA, with no solution; the other seems to be mitigated
by a small change.

1. If I imagine these laid out along a road such that they can
hear each other (and there's not much point if they can't hear
each other in this case - trying to reach a HIGH digi that
will get a distant packet into the APRS-IS), then a packet
that is heard by a middle digi will be sent to its neighbors,
who will repeat it. So the guy in the middle will hear its
own packet, possibly from two digis, and they may step on 
each other. This seems like a variant of hidden terminal.

2. If they don't step on each other(!), then the middle digi will
be tempted to repeat the packet. First simple rule that wasn't
expressed: don't digi your own beacon/packet, tempting as it might
be.

But if, instead, the packet was sent by a station nearby, you
need another rule. I think that if the digi keeps track of the originating
call of the last packet it sent, and refuses to repeat another packet
from that station without an intervening packet, a lot of the 
ping-pong will be cut off (in fact, that should handle the
previous case as well).

73 de chris KG6VYD




More information about the aprssig mailing list