[aprssig] Definitive digipeater specification?

AE5PL Lists HamLists at ametx.com
Mon Feb 27 18:42:27 EST 2006

> -----Original Message-----
> From: Robert Bruninga
> Posted At: Monday, February 27, 2006 4:47 PM
> Subject: Re: [aprssig] Definitive digipeater specification?
> And you know I am opposed to NSR routing unless it allows 
> DIGI1,DIGI2,DIGIn... routing.
> Remember, the 3 hop path above gets a packet from point A to 
> the desired Point B with only
> 3 copies  And is the most efficient on-freq method possible.  
> This is FAR LESS load on the network than even a WIDE2-2 
> packet which can generate
> 16 copies in some areas.  

Bob, the NSR algorithm specifies that only direct digipeater routing is
to be used unless determining best path (also well defined in the
digipeater specification).  You can see the simple algorithm at

> My position is that only the ORIGINATOR of a packet knows  
> where HE needs to communicate at any particular instance.  
> NSR is an attempt to give SYSOPS total control and force all 
> users into a one-size fits all definition of what APRS and 
> the users are (tracking for example)... And that just kills 
> the flexibility of the system to respond to what a USER may 
> need at any given instant.

The exact opposite is true as is known by the thousands of network
sysops out there.  While the originator of a packet wants to specify the
DESTINATION of a packet (and they should), the originator could care
less how it gets there unless they are trying to determine specific RF
propagation.  The APRS network (yes, it is a network) is not a
propagation tool (if that is of interest to you, use PropNet) but a
"tactical communications protocol" (your words).

AX.25 digipeaters are just that: digital repeaters.  The attempt to turn
them into layer 3 nodes by using the n-N algorithms is just another
example of amateurs trying to source route using a layer 2 protocol for
networking functions it does not have the bits to properly support.  And
this is shown in the mishmash of various n-N algorithms out there, all
of which have bugs and problems with routing.

As you also indicated, there is a way to implement APRS such that it has
very efficient use of the local RF networks: using layer 3 gateways to
carry the information over backbones and other networks.  We do this
today.  While it might be possible to flood a packet (or even direct
route a packet) out to Albuquerque from here through 7 digipeaters, it
makes a lot more sense, is more reliable, and less destructive to let
the local IGates do the "routing".  And to truly do this in the most
efficient manner, you must limit the "local" range of the networks to
prevent the gateways from improperly considering a distant station as
local and vice versa.

If people want to know more about using AX.25 as a layer 2 protocol
(what it is designed for), you can sign up to the AX.25 layer 2 SIG
hosted by TAPR.  That SIG is focused on moving AX.25 back to its
original design concept as a layer 2 protocol supporting small networks
while carrying higher layer (network, etc.) data to create large
networks.  With a "universal" underlying network architecture, the
ability to flexibly support multiple types of networks can become a


Pete Loveall AE5PL
mailto:pete at ae5pl.net 

More information about the aprssig mailing list