[MNAPRS] Why we do long path truncation in APRS

[Doug Reed]

This message thread is lifted from the APRSISCE group on Yahoo.com.
It talks about why we truncate long paths at an APRS digi and how it is 
done in the New-N Paradigm.

If you are thinking of setting up an APRS computer and digi at home and 
want to do an I-gate as well, the two most common options are UIview and 
APRSISCE. UIview has been dormant for a number of years but APRSISCE is 
current and under continuous development. APRSISCE is EXTREMELY complex 
and can handle some remarkably complex situations. But for most people 
the default setup may be all they need. The best thing about it is that 
it can download maps from the Internet. My main complaint is that it was 
designed for Win2000 and newer.

73, Doug Reed, N0NAS.

________________________________________________________________________
3a. Re: Help with digi
     Posted by: "Bob Bruninga" bruninga at usna.edu wb4apr
     Date: Wed Oct 26, 2011 11:00 am ((PDT))

 >> If I were over there and wanted to be gracious, I would handle all of
 >> the possibilities, but decrement them substantially like:
 >>
 >> WIDE7-7=WIDE7-2
 >> ...
 > Ignoring the long paths is probably the best option...

The Primary driver of the New-N paradigm back in 2004 was to address this
problem of PATH MODIFICATION.  APRS was being quite fragmented and broken as
a consistent national network by independent DIGI sysops taking their own
draconian hacks at modifying peoples path to suit their own definition of
APRS in their area.  It was destroying the integrity of APRS.  Users had no
idea what was happening to their packets and no way to control it.

Hence the New-N paradigm was developed and it had these primary goals:

1) Consistent performance and ecxpectations everywhere (no local hacks or
mods)
2) User education to WIDE2-2 as the nominal path (or 1-1,2-1 for mobiles)
everywhere
3) TRUNCATION to 1-hop in congested areas where needed.

Truncation meaning that any local packet that violated the general desired
number of hops at any digi, would be digipeated once and then marked as
used-up.  Thus, everyone gets heard via the local digi, but abusive paths
would go no further.  This lets the sender see his mistake, see that the
digi terminated its further propagation, and it let others in the area see
the issue and help correct it.  Also, it kept the issue local where it could
be fixed locally.  Notice this is not really modifying someone's path, but
simply digipeating it once and then marking it as used up.

So a WIDE7-7 packet would ideally be digipeated once as a WIDE7-6* packet
and it would go no farther.  Of course, since most digis were KPC3's that
lacked this smart system, the way it is done there is that we put "WIDE7-7"
as a literal alias, so that it was digipeated once, and the digi's CALLLSIGN
SUBSTITUTION would in effect, stop further propagation.  So a WIDE7-7
arriving at a KPC3 or UIDIGI digipeater would then be substituted as DIGIX*,
the callsign of the digi, and it would go no farther.

So given that APRSISCE can be a smart digi, we would prefer to see WIDEn-N
where it exceeds the locally determined limits to be simply decremented
once, but then marked as fully used up.  So WIDE7-7 would become WIDE7-6*

Bob, Wb4APR


________________________________________________________________________
3b. Re: Help with digi
     Posted by: "daveocn" daveocn at talktalk.net g1ocn
     Date: Wed Oct 26, 2011 11:09 am ((PDT))

Hi Bob,

Would you agree that your explanation pertains mostly to the U.S. ?
Abusive paths et al.

You didn't mention that some of the problems appeared due to badly
configured Kenwood TMD700 digis too.

Dave G1OCN.


________________________________________________________________________
3c. Re: Help with digi
     Posted by: "Lynn W Deffenbaugh (Mr)" kj4erj at arrl.net ldeffenb
     Date: Wed Oct 26, 2011 11:13 am ((PDT))

I've read in numerous places that ignoring offending stations just
encourages them to do more to "get it to work".  That means that WIDE2
might grow eventually to WIDE7 and (more often) low or medium power gets
switched to high power.

 > 2) User education to WIDE2-2 as the nominal path (or 1-1,2-1 for mobiles)

Education may work well here in the contiguous US, but when there's
WIDE7-7 packets coming from operators in another country, it can be a
bit more difficult.  Not saying that's where Colin's WIDE7's are
originating, but I've noticed lots of packet lines crossing the channel.

 > 3) TRUNCATION to 1-hop in congested areas where needed.
 >
 > Truncation meaning that any local packet that violated the general 
desired
 > number of hops at any digi, would be digipeated once and then marked as
 > used-up.  ...  Notice this is not really modifying someone's path, but
 > simply digipeating it once and then marking it as used up.

Hence my recommendation against something like WIDE3-3 becoming
WIDE3*,WIDE2-1.  That'd class as modification, not truncation.

 > So a WIDE7-7 packet would ideally be digipeated once as a WIDE7-6* packet
 > and it would go no farther.

WOW!  How could I have missed THAT one!  It communicates what was done,
but still prevents further motion.  Of course, it doesn't work if you
want to grant the 7-7 one or two more hops and maybe one more hop for
the 7-6, but WIDE7-7=WIDE7-6* is a totally acceptable <DigiXform> in
APRSISCE/32 today.

 > Of course, since most digis were KPC3's that lacked this smart system,

That probably explains how I missed it.  I've observed more packet paths
than I care to admit and don't recall ever having seen a WIDE7-6* or any
other n-N with a used bit.  They're always down to the n*, so I thought
(didn't think, actually) that that was the norm and expected truncation.

 > So given that APRSISCE can be a smart digi, we would prefer to see 
WIDEn-N
 > where it exceeds the locally determined limits to be simply decremented
 > once, but then marked as fully used up.  So WIDE7-7 would become WIDE7-6*

Done.  Supported in both the current development and even the generally
released versions.  If you want/need to truncate a path considered
abusive in your area, you can do so with:

<DigiXform>WIDE7-7=WIDE7-6*</DigiXform>  (Changing the n-N as necessary
or desired).

Lynn (D) - KJ4ERJ - Author of APRSISCE for Windows Mobile and Win32

PS.  Anyone got time to boil this down for an update to
http://aprsisce.wikidot.com/doc:digipeating

________________________________________________________________________
3d. Re: Help with digi
     Posted by: "Colin 2E0XSD" colin2e0xsd at sky.com colin2e0xsd
     Date: Wed Oct 26, 2011 3:26 pm ((PDT))

Thanks for your responses and confirming I was thinking along the right
lines (second time around).
(I hope that is true, I'm reading this quickly and it is past my
bedtime, so maybe not)

I have only come across 1 digi over here that used 7-7 & I didn't even
bother to trace it since I was only running Rx only at the time. I can't
say I have noticed it in recent times, so hopefully they have been educated.

My point was that since we now have a new Digi that should give us
international coverage (from Scotland into the other nations that make
up the United Kingdom at least). I don't want to encourage the use of
high Wide N-n without totally ignoring the existence of them.


73 for now,
Colin.


________________________________________________________________________
3e. Re: Help with digi
     Posted by: "Bob Bruninga" bruninga at usna.edu wb4apr
     Date: Wed Oct 26, 2011 3:31 pm ((PDT))

 > Would you agree that your explanation pertains mostly
 > to the U.S. ? Abusive paths et al.

I do know that things were quite different across the Atlantic, but I
thought eventually most areas adopted the New-N Paradigm.

But then I understand that it is very difficult to have a legal digi in the
UK, and so you guys are more dependent on home stations doing the relaying?
Maybe you can give me an update on the last few years on any general
changes?

 > You didn't mention that some of the problems appeared
 > due to badly configured Kenwood TMD700 digis too.

Can you fill me in on that one?  Usually it is not the digipeater settings
that cause abuse of a channel with excessive hops  but the users themselves
causing too many hops.  SO fill me in on what you are seeing?

Thanks
Bob, Wb4APR

Dave G1OCN.


________________________________________________________________________
3g. Re: Help with digi
     Posted by: "James Ewen" ve6srv at gmail.com ve6srv
     Date: Wed Oct 26, 2011 9:37 pm ((PDT))

On Wed, Oct 26, 2011 at 12:13 PM, Lynn W Deffenbaugh (Mr)
<kj4erj at arrl.net> wrote:

 >> So a WIDE7-7 packet would ideally be digipeated once as a WIDE7-6* 
packet
 >> and it would go no farther.
 > WOW!  How could I have missed THAT one!  It communicates what was done,
 > but still prevents further motion.

I thought Bob had fallen and cracked his coconut again when I read
that on my Blackberry this afternoon, but then I looked again and
noticed the has-been-digipeated bit, and that spark of genius that Bob
has shone through...

 >> Of course, since most digis were KPC3's that
 > That probably explains how I missed it.  I've observed more packet paths
 > than I care to admit and don't recall ever having seen a WIDE7-6* or any
 > other n-N with a used bit.  They're always down to the n*, so I thought
 > (didn't think, actually) that that was the norm and expected truncation.

We've been using UIDIGI alias substitution to trap long paths for many
many years, I'd never seen that idea before either, but then none of
the hardware I had used had the capability.

I love these little nuggets of information that pop up. Now I'm
thinking of how that could be implemented in my favourite digipeater,
the OT2... the wheels are churning.

-- James VE6SRV Messages in this topic (22)

________________________________________________________________________