[aprssig] The WORST digi setting
bruninga at usna.edu
Fri Dec 10 07:49:31 CST 2004
The WORST digi setting is UIDWAIT ON.
If you are seeing lots of dupes, then yours or other digis
around you probably have UIDWAIT *wrongly* set to ON.
Please make sure it is off (the default) in your digi.
Like some other TNC parameters, a casual reading may
make this paramater sound beneficial when in fact, it is
BAD for APRS. Here is why.
APRS was designed around the fundamental principle of
radiating information outward as efficiently as possible. That
is, the next outward tier of digis are all supposed to digi
the same packet at the same time so that only one time-slot
is consumed per tier, not per digi.
Turning on UIDWAIT eliminates this efficiency and leads
to a factor of 2, 3 or 4 fold increase in the channel load
and an increase in dupes received by users. Don't do it.
Example: lets say all digis can hear 3 others and digi A
launches a WIDE,WIDE packet, What should happen
in 2 seconds is that the 3 surrounding B digis all digipeat
it at the same time, and then the next tier of C digis all
do it at once as well for a total ot two time slots (2 sec)
for 2 hops: And all digis get only one copy and users
inbetween digis only see at most 2 copies.
1st sec 2nd sec
But with UIDWAIT ON you get an explosion of time
slots drastically cutting throughput and in this case
users inbetween digis may hear 3 to 6 copies each
And if the original path was for 3 hops, many of these
preserved copies have the potential to now propogate
further outward. A REAL MESS. Dont do it.
Set UIDWAIT OFF (the default).
Of course there are lots of other 2nd order effects
and considerations that can significantly affect these
two simple models, but they are secondary to the
original APRS design principle that packets are
supposed to radiate outward in tiers and all tiers
are supposed to transmit at the same time to minimize
Please, set UIDWAIT OFF (the default).
The next worst command is ID instead of NOID
in the UIFLOOD parameter, and the 3rd is the HID
which should be OFF. etc, etc
de WB4APR, Bob
More information about the aprssig