Order Tray | Contact Us | Home | SIG Lists

[aprssig] APRS W,W and W,W,W example

Robert Bruninga bruninga at usna.edu
Fri Feb 11 18:17:44 UTC 2005

>>The lessons are loud and clear.  Its time to abandon 
>>W,W (and R,W) type paths because their generation of 
>>3 to 5 times the number of dupes compared to WIDEn-N is  

>Could you explain?
>I understand about the RELAY, but WIDE,WIDE and 
>WIDE2-2 seem to me to have the same number of repeats 
>for a given configuration of digis.

This is the "sleeper" in the evolution of APRS which we
have kinda known about all along, but because of recently
doing detail analysis of off the air data, we had not
realized just how bad R,W and W,W were at generating
dupes.  Now we know...     Here is the difference.

Callsign Subsittuton was introduced by Paccomm in 1996 and
went a big way towards reducing dupes for W,W packets, 
but it only keeps a digi from repeating the exact same packet 
it already has callsign substituted.  It does not prevent that 
same digi from digipeating all the other copies from all the 
other digis that also heard it...

Thus, in a typical area where each digi may her 4 other 
digis, it may digipeat this packet 4 full times itself.  Once
when it heard it, then once for each of the copies that
each of the other digis heard.    This really blows up fast.
In our area, most W,W packets may be seen 3 to 5 or
more times each..

Whereas the WIDEn-N algorithm does not depend on
callsign substitution ,but does dupe elimination by
exact comparison to the full packet (with a CRC check).
Thus, once it has digipeated the packet once, no matter
how many more fresh copies hit this digi, it will not digipeat
it again.  (this dupe elimination is typiclaly 30 seconds deep).

THus WIDEn-N has perfect dupe elimination, (each digi
only digipeats it once) but also assures almost perfect 
propogation even in the presence of collisions because
if it misses the packet, it will  still see copies from its
neighbors and have a second and a 3rd chance to
get this packet.

The bottom line,  just looking at W,W and W,W,W dupes
compared to W2-2 and W3-3 packets is at least 3 to 5
times as many wasted QRM dupes.  So when we started
this New n-N Paradigm crusade, the initial target was
Big values of N.  But in looking at the data, the even
bigger cause of problems was all this W,W and W,W,W
dupes banging about.   

A three to 5 times improvement in efficiency by simply
switching to WIDEn-N instead of W,W, ro W,W,W
is a phenominal paradigm shift and is so easy to accomplish.

I admit, I have been part of the problem because in our
area, until recently fully HALF of the digis in the mid-
atlantic states did not do WIDEn-N, and so we could not
reliably use WIDEn-N in this area because those old WIDE
only digis would not go away...

BUT when we started getting into this new n-N paradigm
and realized that the old TRACE-only and WIDE-only digis
could be "fooled" into digipeating a WIDE2-2 packet by
simply adding that as an alias to these old digis, all of a
sudden, we were no longer hampered by being not able
to support at least WIDE2-2.  Now we could encourage
at least WIDE2-2 everywhere...

Thats when we began seeing that it wasnt big N's that
was the problem in this area, it was the 3 to 5 times
dupes of all W,W and W,W,W packets that was really
killing us.  Plus, looking at the map, we are just as
dense as the LA basin, but we were compounding our
density by all these unnecessary dupes...

This is really exciting to now see how easy it is to
gain an instant factor of 3 to 5 improvement in out
APRS by simply moving to WIDEn-N and discourraging
W,W etc...  Of course R,W is just as bad if not worse
because you then also have the home stations
throwing in even more dupes..

But what about home station digis?

1) If we reduce the number of packets on the air by
a factor of 3 to 5, then I bet a lot more mobiles are
going to be heard in fringe areas that could not be 
heard before.    Thus we dont need as many

2)  If a fill-in RELAY digi is needed, then make it
a WIDEn-N digi!  The advanteage of WIDEn-N
is that adding one more digi does not add any
significant QRM.  The perfect dupe-elimination
of the WIDEn-N system prevents this new digi
from adding any additional dupes.

And now, any *required* home station even 
without a KPC-3 can be a WIDEn-N digi.
But we MUST use this carefully , because too much
of a good thing is BAD...

Hope that helps...


More information about the aprssig mailing list