[aprssig] Tracker Decaying algorithm

Robert Bruninga bruninga at usna.edu
Thu Mar 23 09:45:14 EST 2006


>> That is how APRS was designed (APRSdos).  There
>> was no such thing as fixed rates.  Everything that
>> did not change, always doubled the period to the
>> next transmission.  Positions, Objects, Messages, etc.
>
>Ok, I missed this class.  This does sound like a very 
>good idea,  

Yes, and one of my biggest disappointements that most
follow-on clones completely ignored this fundamental
principle of APRS channel sharing.  New data is very
important and should get a high rate.  Old duplicate
data should decay down to the 30 minute rate to 
minimize channel load on fixed unchanging data.

Oh, and with the dacay algorithm there are NO USER
settings!  Thus, nothing to screw up the network with!  

You should see how rapidly two stations can QSO with
messages when using the decayed rate.   Example:
on a channel where your first new message line got
collided, the first retry from APRSdos or other software
with the decayed-rate system is 8 seconds later.
For, a system with the simplistic 1 minute fixed-rate 
system, it is 1 minute later.  THus an almost
8 to one speed advantage for APRSdos.

If the first two packets get lost, then it is 16 seconds
versus 2 minutes.  Still an 8 to one advantage...

And if in a 2 way QSO where APRSdos and some other
software (Xastir and APRSsce I think) use embedded
reply-acks, the probabiliy of acks improves by a huge
factor making messages fly at almost typing speed
compared to the simplistic fixe-rate, single-ack systems
which make messaging exceptionally slow.

So you see, APRSdos and other decayed-rate systems
are actually refreshing at that 8 second rate for NEW
data, but rapidly decaying down so as not to burden
the network.  After 6 minutes, both APRSdos and the
fixed rate system have sent 6 packets.  After 30,
APRSdos has only sent 9, compared to 30 for the
fixed rate system, though most fixed rate systems
just stop sending after a preset number.

Bob, Wb4APR





More information about the aprssig mailing list