[aprssig] Smart beaconing and the tragedy of the commons

[Joel Maslak]

On May 23, 2007, at 9:10 PM, Robert Bruninga wrote:

> "What kind of algorithm can we place on smart beaconing
> *settings* that can protect the network from improper settings.
> If we can come up with a consensus on how to implement it, then
> I think we have a great thing to go forward with and hope that
> Kenwood will implement it in the D710...

You can see an implementation of what I felt was a reasonable  
SmartBeacon algorithm a few years ago.  Note that this is the  
algorithm, not the defaults:

http://smartpalm.cvs.sourceforge.net/smartpalm/main/aprs.c?annotate=1.20

Scroll to the bottom, in the procedure called "smartBeacon".  Anyone  
who wants to steal my code, feel free.

A few things:

1) If you turn a corner shortly after a normal beacon, it won't  
beacon unless it's been at least the number of seconds specified in  
the turn threshold.  So you'll miss some corners.

2) It throws away the first couple "good" GPS readings since they  
turn out to often be bogus, at least with the GPS I was using when I  
developed the code.

3) There are some slight modifications to the SmartBeacon idea in the  
code.

4) We beacon less often on corners when going slow.  Basically, the  
faster you go the more reliable the GPS direction is, so I trust it  
more to indicate a corner if you are going fast.  I don't trust it  
nearly as much if you are crawling along.  I don't corner peg *AT  
ALL* when you're speed is below the slow speed.

Bob, I definitely want to know where the problems with the algorithm  
as written are - not the thresholds themselves, but the actual  
algorithm.