MARATHON.txt 6.3          APRS at the Marine Corps Marathon 1993 (and 4)

APRS LESSONS LEARNED @ MC Marathon, 1994!
Its over!  14,000 runners, LOTS of hams, and our second year with APRS!  Last
year we had GPS on the LEAD, LEAD Handicapped, & TAIL chase cars. It was great,
but predictible.  This year we let APRS dead-reckon the predictible movements
of the chase cars and built 11 Trackers for the ambulances.  Lessons learned:
 * Completely "sealed" GPS/TNC/Radio boxes should have drain holes in bottom!
 * Maritime GPS units withstand immersion in water. TNC's don't
 * You can't duct-tape GPS trackers to vehicles in the pouring rain
 * New Marine Corps Tents (made by lowest bidder) leaked everywhere!
 * 14,000 runners, vehicles, etc + RAIN = MUD
 * Mag-mount GPS trackers wont stick to aluminum HUM-VEES.
 * Watching ambulances, which are parked for 99% of the event is BORING!
 * Ambulances with emergencies are under such close control by the ambulance
   direction net control, that he knows EXACTLY where they are anyway.
Of 11 units, 2 never quite got finished, one just couldn't be attached in the
rain, one leaked, flooded and died, the tinyest (running on AA cells) lasted
6 hours.  It rained from 5 AM until 1400.  Now for the good news:
 + We got double milage out of most APRS mobiles.  They put their GPS's in
   stand alone trackers for the ambulances, but kept their LAPTOPS and used
   the INPUT-MY command to manually report their position.
 + The 6.0 alt-IGNORE command let the entire event operate on 145.79 while
   ignoring ALL other non-participating stations.  THis keeps all APRS pages
   free of non-participants.  But the 6.0 transitioning in an out of
   this SPCL mode needs some more work.  Many stand-alone trackers are XTAL
   controlled, so you MUST plan on using the normal APRS freq for spcl events.

CONCLUSIONS:  Next year, we will probably go back to tracking the high-profile
chase vehicles and HAM mobiles that are always moving, rather than ambulances.
These 11 trackers now are permanent HAM radio assets in our community in
addition to the other 6 HAM mobiles that are permanently configured for APRS.


SUMMARY OF 1993 MARATHON!

Whew! Its over!  14,000 runners Plus family and spectators at the Marine Corps
Marathon!  Here are the immediate lessons learned today:  1) MC chase vehicles
(HumVees) were not identified nor available until 45 minutes before start, to
install 3 GPS-packet tracking devices!  A) Mag mounts dont stick to ragtops
or Aluminum, B) HumVees run on 28 Volts, C) battery cables are sealed in tuna
fish size pot of grease under passenger seat, C) clearance between Battery
posts and seat is less than 1/4 inch!  Last one was finished 1.5 minutes
before the starting Howitzer fired! (30 feet away!)

    All three vehicles were tracking and showing up beautifully on the PC
screen in the COMM tent.  The rest of the event went beautifully, with all
three vehicles (Lead handicapped, Lead runner, and Tail-end-Charlie)
transmitting their GPS position once a minute.

Here are the lessons learned for the APRS software (AND SINCE CORRECTED!):

   1) Automatic Dead Reckoning was a pain.  A 1 minute error in the PC clock
resulted in a 1000 yard error in dead reckoning.  FIX:  DR is now an ON/OFF
toggle.  Also the alt-SETUP-GPS-TIMEsync command will sync each PC to any GPS
packet on-the-air.

   2) Downed runners and medical reports filled the screen.  There was no
mechanism for removing old positions and objects!  FIX:  APRS now keeps a
separate timer on EVERY OBJECT packet and there is now a KILL command to 
remove old objects from ALL screens.

   3) APRS does its job well, but connected links are better in classical
applications, such as passing patient lists to hospitals.  We had a separate
medical packet link which performed that function admirably.  A single APRS
net could not possibly "do everything" at an event of 14,000 runners (at 1200
baud anyway).  Separate APRS nets on separate frequencies for separate
functions could be built into an impressive "TACTICAL" network system.

   4) In the MAIN COMM tent with 4 two meter nets (plus other bands) there was
very little QRM.  All APRS packet stations at all checkpoints were mandated
to operate with only 1 watt.  A central digipeater on a building more than a
Mile away from all other stations ran 15 watts and all packets went via this
WIDE area digipeater. 

   5) The APRS message mode was slow with ACK times often minutes or more. 
FIX: Each message line is now TIMED individually (instead of as a whole group)

   6) The event operated on the local 145.79 APRS frequency with dozens of
other APRS stations, the 9 Marathon stations and ALSO the Point-to-point
packet link!  The conventional packet link soon moved to another frequency
for their own sake, and our packets got through better.  FIX:  In version 6.0,
APRS now has an alt-IGNORE command so that all special event stations ignore
all APRS and other packets EXCEPT those addressed to SPCL.

   7) With APRS we could easily track the lead, pack and tail runners, which
was great for PR, but after the Lead and PACK crossed the finish line, only
then did all of the 80 ambulances begin to move to pick up downed and dead
runners.  We did not have enough GPS devices.  FIX: In subsequent events we
have noticed that it is a WASTE to put GPS on the LEAD, PACK and TAIL vehicles
since the position, course and speed of these vehicles is TOTALLY PREDICTABLE!
With the DEAD-RECKONING of APRS, these vehicles can be placed on the APRS net
as objects and only occassionally updated based on voice reports.  APRS still
moves them by dead-reckoning and all other APRS stations cant tell the
difference!  In 1994, we hope to have 8 or more GPS devices on the critical
life support ambulances.  (DO consider a GPS for the TAIL vehicle, because it
is far less predictible than the others)  See first section now, for 1994
results.

   8) The complete event can be re-played from MARATHON.hst to see how it
went.  To make sense out of it all, play back only one mobile at a time, and
turn Callsigns off.  WB4APR-9 was the lead Handicapped vehicle (started 15
minutes or so before all runners), W3ADO-9 was the lead runner, and MOBILE-9
was Tail-end-Charlie.  Statistically, we did very well.  W3ADO-9 was turned on
at 0827 but did not move until 0902.  It was removed from the vehicle at about
1127.  Transmitting at once a minute, there should have been 145 posits trans-
mitted.  We counted about 115 in the file.  The missing packets could have been
either colisions, or bad GPS fixes (masked by buildings) so that the same posit
was transmitted more than once (and therefore filtered out as a dupe by APRS).
The result computes to almost an 80% success rate!

