Order Tray | Contact Us | Home | SIG Lists

[aprssig] Fillin digi experiment

Robert Bruninga bruninga at usna.edu
Tue Jan 19 18:16:28 UTC 2010


> Yes, you are just setting up your digipeaters 
> in a wrong way.... Simultaneous transmissions 
> should be avoided by all costs. I don't know
> where this idea came from, but it does not work 
> in the real world, and never will.

It is fundamental to the APRS network from day one.

APRS is based on getting informaion outward from a user on a
very collision limited shared network.  A packet is desired to
propogate outward radially a few hops in all directions.  

In a typical high density APRS network, each digipeater can hear
from 3 to 5 or more other digis directlly.  It is imppssible to
manage this come-as-you are network by allowing each digipeater
to have a dedicated slot time for repeating a packet.  That is
why a two-hop packet is designed take up only TWO time slots.
One slot at each radial tier of digipeaters.  Each digi has
DWAIT set to ZERO so that it keys up immediately when presented
with a packet and does not hold off waiting for the channel to
clear (in mosts cases it never clears!)...

It is very bad in an APRS network to deviate from this practice.
Having each digi hold-off until the other digis all finish
digipeating each packet exponentially decreases the network
capacity.  Sure it can be done in small networks where there are
very few users, very few digipeaters and there will be an
improvement in local packet success.  But it restricts the
network to probably less than 10% capacity...  Most APRS
networks are far beyond being able to provide this level o f
service.

Lets say you let the Digis all be gentlemenly and hold off...
In our area where each digi can hear say 4 other digis strongly
and say 10 other ones at least opening squelch, then a 2-hop
packet will occupy as much as 16 slot times longer to propogate.
Compared to the desired 2 slot times, this loss in channel
capactiy of 8 to one cannot be tolerated in most networks.

> Bob, do us a favour and stop yakking of this concept. 
> It may suit SoCal and some other extreme examples of 
> crowded traffic, but it introduces
> very bad network behaviour everywhere else.

NO matter where you are, in most cases the APRS network would be
severly hampered in capacity if each digipeater had its own turn
to transmit.  And remember, each USER on the ground can usually
only hear ONE digi at least 13 dB stronger than any other so HE
is NOT impacted if the digis all transmit at the same time.  He
still gets it and the netowrk handles the traffic 8 times
faster!

> Many of our systems are computer-based 
> so I can bring up multiple monitor
> windows on the screen. Most entertaining, 
> dear Watson :)

Entertainig yes, but deceptively uselsess.  Again, it is
impopssible to assess the health of an APRS network by
monitoring decoded packets.  For example  an empty network and a
totally jammed network look the same.  Nothing getting through.
Yet there is a huge difference..

> PS. We have been experimenting with these 
> "vicious digipeaters" in secrecy for a while, 
> and they work well. You just have to set up the
> digis to avoid transmitting over one another 
> in your whole network :)

Yes, I agree.  It is possible.  But it drastically reduces
overall capacity.  So please do not propose these viscious QRM
generating hold-off-listen fill-in digipeaters in other areas
where they are not a good idea.  We need to be clear when we
make general statements about how good a system "works" when we
do not at the same time specifyu the conditions or local
situation..

Similarly, I will try to make sure my condemnation of them does
not include sparse areas of low activity where their added QRM
and network loading can be tolerated.

Bob, WB4APR





More information about the aprssig mailing list