[aprssig] Tools for Reliability Improvements

[Robert Bruninga]

> What is the groups thoughts/ideas on imbedded 
> intelligent algorithms in digipeaters and iGates 
> to reduce the QRM?  

IT was the rise of draconian approaches at each different digi,
to filter, trap and dump packets that gave rise to the New-N
paradigm.  The new-N paradigm came about to prevent the
degradation of the APRS network into individual kingdoms by
often-times absentee digi-lords giving inconsisisten performance
everwyhere that would destroy the integrity of the APRS network.

> We already have "path correction" to downsize 
> the number of hops automatically.  A step in 
> the right direction.  

In the new-N paradigm, I think we use the term "trapping" to
prevent the spread of abusive paths.  But we allow that poor
path to at least be digipeated once, so that the user gets
feedback that his system is working, but that it is not a
preferred path for that area. 

Further, the digi that traps the packet has clearly announced
recommended N value in its beacon, so the user is fully informed
of the expectations of the network at that point.  Each digi
contains "WN, SSn-N" to inform its users of the accepted value
of N for that area.   And this trapping is the ONLY method used,
so that users can learn APRS easily without having to learn
different rules for every digi.

I am strongly opposed to any other manipulation of "correction"
of paths.  Such draconian measures undermine the consistent
performance of the network.  We cannot all be considered dumb
users.  We have to focus our efforts on user education,
otherwise the use of arbitrary smart-digi techniques simply
destroys the ability of some users to accomplish their immediate
communications goal.  Only the sender of a packet knows what he
wants the network to do with that packet, not some arbitrary
un-changeable rule installed by an absenteed digi-lord.

> Ignore packets passing from user selected 
> iGates or wide coverage digi's assuming 
> that the packed is already successfully 
> in APRS_IS? 

Although some people are focused on getting packets to the
APRS-IS, others depend on the RF network for their local
communications planning and needs.   The RF network should
server the local area on RF alone.

> The digi operator could preset a list of 
> other iGates or digi's that can be passed 
> to APRS-IS, but not digipeated, or ignored 
> altogether to avoid placing redundant 
> packets back on-the-air if they have already 
> been dealt with by others.   

Such a draconian measure could prevent someone's RF packet from
doing what the sender intended.  Thus completely destroying the
integrity and consistent expectations of the APRS network.

> A fill-in digi with "polite" channel access 
> dumps packets that are repeated by a wide-area 
> digi?  Some conditions about 
> the wide area coverage must be applied.    

Inherent in the WIDE1-1 design is the concept of fratricide.
The wide area digi and the WIDE1-1 digi will hear the packet at
the same time, and both are supposed to respond at the same time
(UIDWAIT is supposed to be OFF).  Thus there should be no dupes
to digipeat.  Further, the WIDE1-1 will not repeat the wide
area's dupe, and the wide area digi will not repeat the
WIDE1-1's dupe due to the use of WIDE1-1 as the first hop.
Again, there is no problem here that needs to be fixed?

> Since position is usually sent by APRS mobiles, the smart 
> digi or iGate could remain fully engaged with close in 
> mobiles similar to how "# filter m/15" works?  It can keep a 
> stack of mobiles known to be in the area and exclude others.  

This kind of filtering frightens me.  There are many instances
of incorrect position reporting, and more often than not,
someone with a crisis or emergency issue that is trying to send
a packet, may in fact have an old or outdated position!  Thus
the network would deny him access exactly when he needs it most!
Please no...

> A more complex approach would involve 
> dynamically identifying abusive paths...
> A method of self healing to protect the 
> channel loading? ...
> the intelligent digi or iGate would start 
> ignoring it in a ramped rate.  
> smart-beacon traffic control implemented 
> at the digi.  

But this approach ignores three fundamental rules of human
nature.

1) The user response to poor digipeating is to INCREASE his
rate, so throttling him at the network will only cause more QRM
on the input channel not less, and it is congestion on the input
channel that is the problem.

2) Draconian rules are too often set-and-forgotten by absentee
digi-lords who are not around when you need an immediate change
to respond to an emergent communications need.

3) The problem is USER-INCORRECT-SETTINGS and this can only be
fixed at the users transmitter, no where else, or he continues
to  abuse the network forever.

These three human nature principles led to the New-N paradigm.
And the New-N paradigm is based on the cornerstone of these
approaches:

A) FOCUS on USER EDUCATION by:
B) Eliminating all legacy paths.
C) Standardizing on only one path: WIDEn-N
D) Educating on the proper setting for N.
E) DIGIS are allowed to trap large values of N in their area.
F) DIGIS identify their rules and settings to users.

Fully implementing the New-N paradigm is all about USER
Education and getting HUMANS to talking and communicating with
HUMANS again.  Any particular problems can be and should be
fixed at the user end at the local level.  Then let the network
serve all according to their needs.  Even one-way dumb trackers
are encouraged to include their VOICE listening frquency in
their packets so they are two-way human participants in the net.

HAM radio is not CB.  Ham radio is not cellular-for-dummies.
HAM radio operators are supposed to learn and understand the
details of their communications systems in order to be flexible
and responsive in time of need.  

Kludging up the network to protect it from ignorance only kills
the golden goose, throws the baby out with the bathwater, and
prevents legitimate users of the network from getting consistent
expectations.  Integrity is lost.

> This sort of autonomous decision making 
> could reduce the "pollution" on a simplex 
> channel.

One man's "pollution" is often another man's urgent need to
communicate.  It is not our job to try to limit and control how
people use their radios, since only the sender knows his
immediate need to communicate.  But we can put our efforts into
education.  That is what Ham radio is all about.

> Alas, I'm not a programmer, 
> just a very capable RF guy.  I'd be happier with a better 
> channel access method to avoid collisions in the first place. 

Its called share-and-share-alike like all other ham radio RF
bands.  I agree entirely with you desire to clean up the
network, but the best method, as instigated by the new-N
paradigm, is user education.  Go visit ham clubs, give talks,
educate the users.  And by all means, monitor the network for
"problems" and help those users nicely to learn how to use the
network more equitably.

Hope that helps...

Bob, WB4APR