Order Tray | Contact Us | Home | SIG Lists

[aprssig] The New N-N paradigm CAN WORK

Rick Green rtg at aapsc.com
Thu Oct 28 14:12:56 UTC 2004


On Thu, 28 Oct 2004, Robert Bruninga wrote:

> >>> rtg at aapsc.com 10/28/04 12:53:45 AM >>>
> >Well, yes, It could work, but it will take much coordination
> >to define the LANn-N areas...
>
> No, not really.  Each digi owner just decides what LANn-N
> his digi is part of and changes it immediately.  No coordination
> required.
  So who decides the basic definition of the LANs and their boundaries?
>
> > and travelling users will have to pay attention to where
> >they are and keep adjusting their path accordingly.
>
> Not at all.  RELAY,WIDE2-2 would work everywhere.
  ...except in sparsely populated areas, where digis and igates are rare.

>
> >People who live near LAN boundaries will have to choose
> >whether their packets will go one way or the other.
>
> No, since they are fixed stations, they can choose
> DIGI1, WIDE and go 2 hops in all directions.
  Actually, I was thinking of mobiles.  Position reporting of fixed
stations is IMHO just needless advertising.
In a dense
> area, there is no reason to go further as that is beyond
> his local area and he should not be going any further
> anyway...
  I agree.  But the definition of 'local' seems to be an ALOHA circle,
which contains enough stations to fully utilize the channel, yet not
over-saturate it.  This definition changes over time and geography.
Doesn't it make more sense to design a network where the need to analyze
the local network is limited to the digi operators, who are assumed to
have more local knowledge than the transient mobiles, who can then use a
'worst case' path that will reach the edges of the ALOHA circle and have
the best probability of reaching an igate even in the remote areas, yet
will not over-saturate the network in the dense areas?

> If he is in a sparse area, he can go as far
> as he wants using WIDEn-N since those digis will still
> allow it.
  I'm trying to define a scheme that would be optimum for the mobiles,
since that's the original, primary 'target user' of APRS.  I believe it
unrealistic to expect mobile operators to research every local LAN they
pass thru and adjust their path accordingly.

>
> >I see work, confusion, and frustration for all.
>
> I see:
>
> 1) a universal MOBILE path that will get back to the
> original intent of APRS to work everywhere
  My approach also allows a universal path to be used by mobiles, and
protects the network in dense areas from the long paths which are needed
by mobiles.  Think about a trucker crossing the NV-CA desert.  With
WIDE2-2, they are essentially invisible until that moment when they cross
the final pass into the LA basin.  With WIDE7-7, they have half a chance
of getting to an igate while crossing the desert, and when they cross that
pass, they immediately are seen as an abuser of the LAN.  While in motion,
and facing a long downhill in a heavy vehicle, is NOT the time to be
worried about changing settings on your APRS tracker.

>
> 2) An educated approach to fixed stations where the
> owner operator is required to at least "understand"
> the simplest issues about his local network and how
> it works and what his use of APRS is for in his area.
  Yes, let the fixed stations (digis) bear the burden of local knowledge
and LAN optimization.  Let the mobiles run long worst-case paths, and
truncate the packets at the edge of the ALOHA circle based on local
knowledge.

>
> >If instead, you give each digi operator the capability to
> >define their own service area, and encourage them to
> >simply drop any packets originating outside that area,
> >you make obnoxious DXing on APRS impossible..
>
> Two comments:
>
> 1) t is my WORST NIGHTMARE!  That would killl APRS
> because it makes operation ambiguous, arbitrary, locally
> dependent, and impossible to use for emergent requirements.
> That is the disaster we have now with some areas disabling
> RELAY, others disabling WIDE, etc.    The only filtering
> I will support is one-on-one bud-listing of consistent
> individual abusers...
  One-on-one budlisting is far more labor intensive and arbitrary than a
rule-based filtering based on ALOHA range and channel capacity.
>
> 2) It is easy for you to say to just "drop" bad packets,
 I'm not labelling either the packets or their operators as 'bad'.  Long
paths may be optimum in sparse areas, and are clearly far from optimum in
dense metropolitan areas.

> but 90% of all digis are KPC-3's and there is no way to
> do that.  Your statement is like saying the way to make
> money is to buy low and sell high.  Easy to say, but
> can't happen for a dozen years until all KPC-3's are
> made obsolete.
  I must admit I don't know the inner workings of a KPC3.  Is there not a
dedicated smart-digi rom for it, as there is for the TNC-2 clones, which
are the ubitiquous ones in my world.  If they're really that common, is
someone working on a UIDIGI port for that hardware?

  Dividing the country into multiple tiled LAN areas requires that mobile
users change their paths for optimum use as they travel.  Allowing each
digi operator to define their own ALOHA area gives a truly universal path
setting for the mobiles.  ALOHA areas aren't discretely tiled.  They morph
shape and size as you travel thru different digi coverage areas.  It's the
fixed digi owners, not the mobiles, that are in the best position to know
the characteristics of their ALOHA range and its population of users.  So
I believe it makes sense to give the power and responsibiliy for network
optimization to the digi operators, and be more forgiving of 'worst
case' path definitions on the mobiles.

-- 
Rick Green

"They that can give up essential liberty to obtain a little
 temporary safety, deserve neither liberty nor safety."
                                  -Benjamin Franklin





More information about the aprssig mailing list