[aprssig] APRS routing strategies

[Henk de Groot]

Robert Bruninga schreef:
> LOCAL=12km
> REGION = 50km
> 
> Not a good idea.  Makes no sense in the RF world.
> Your 12km radius can be hit WITHOUT a digi from a home
> station, but need a digi with an HT.  How does digi decide?

Same question, how does the user decide? Please notice that this was the 
*first* idea, we have had the same discussion about it back then in 2001 
and nothing has changed in the meantime.

LOCAL is just one hop. With a HT conditions change too fast, but if you at 
least hit the digi you will be heard in the digi's local coverage area.

For HOME stations this new scenario's may be nice but I think they are not 
the group I had in mind for this change. HOME stations know where they are 
and which digis surround them. It's mobile users and visitors who will 
profit from this. No need to configure your tracker while you are half-way 
on your journey.

For example:

PE1DNN>APRS,CALIFO

Is in my opinion much better than

PE1DNN>APRS,CA4-4

How does the *user* detemine it must be 4 hops? Why not 3 or 5? Better 
yet, does the user care? I think not, his intent is to spread his packet 
over California.

Now having the digi expand CALIFO into a suitable path will make it easier 
for the user. In area's without CAn-N routing the DIGI can just send the 
packet to named digis, in area's with CAn-N the DIGI can use that path, or 
a mix.

For the end user is stays crystal clear, where ever I am in Calicornia, if 
I address my packet to CALIFO the network will do a best-effort attempt to 
get my packet all over the state.

Of course when it is just not physically possible (because of the local 
traffic conditions for example), then it is not possible. There are no 
guarantees, but a manual setting would not be able to do a better job 
either (or even worse cause intollerable QRM in an attempt to force its 
way through with an excessive path).

In this scenarion the user never has to change his setting because for 
example in one corner of California he would have to use WIDEn-N, in 
another area CA4-4 and in the middle of the state CA2-2. Why should the 
user be bothered with that, all he want is to cover the state and with 
this proposal he can just do that without constantly having to change his 
path.

> REGION makes no sense.  Alomst every digi on just flat

I invite you to come up with some better aliases. The alias should refelct 
the intent od the user. Where does *he* want his packets to go to. Think 
about the end user, *what* does he want. He does *not* want to know 
hopcounts, he wants his packets *to go somewhere*.

> Already EVERY packet goes to world.

No, sometimes one hop is enough to reach an IGATE and sometimes more hops 
are needed. When the user uses IGATE the network should make a best effort 
attempt to reach an IGATE (or IGATES, to have some redundancy). Sometimes 
it means 0 hops (when the DIGI itself is also and IGATE) and sometimes it 
means a few hops. Nice thing is that the DIGI may know there is a LNKn-N 
path and take advantage of that to reach the location of the IGATE, which 
saves flooding with WIDEn-N. Or maybe the DIGI even knows the exact 
callsigns of the intermediate DIGI's. A traveling user cannot compete with 
that and will always use more bandwith than the optimal path defined by 
the owner of the DIGI.

> This idea makes no sense in any sense of the word, since
> what limits APRS is NOT RANGE, but BANDWIDTH and
> LOAD.  In Kansas it is safe to send ones packets out
> 300 miles in all directions.  In LA, 30 miles is about right.
> Thats a 100 to 1 difference in area.

But in LA you may want to use LNKn-N paths to go beond that.

> APRS is limited not by mileage but by the ALOHA range 
> (a 100% channel) and it can be ANY PHYSICAL SIZE.

So how is the user to *know* what path is permitted. Especially a 
traveler. But if the user would use:

PE1DNN>APRS,WORLD

Where I defined WORLD as: "This is a route with the maximum number of hops 
that can reasonably tollerated in the area."

So in Kansas it will expand to a path that reaches 300 miles in all 
directions while the very same packet will reach 30 miles in LA (300:30 by 
the way is 10 to 1, not 100 to 1).

So you can travel arround the globe with PE1DNN>APRS,WORLD an never use an 
excessive amount of bandwith anywhere but also always have the maximum 
possible coverage. The setting that works in Kansas also works also in LA 
without any angry faces. Try that with your manual setting, you can't! You 
would have to know the local conditions everywhere and constantly adjust 
the path. It's just not realistic.

PE1DNN>APRS,WORLD will always keep you within ALOHA limits since the DIGI 
is at a fixed point and the local traffic density is known by the 
opertator of the DIGI. Therfore the DIGI substitues a shorter path if the 
density is higher.

But I know why you are opposed to this idea, and I can't change that:
1) It's not your idea
2) It doesn't work on a Kantronics box

So I will step down from my soap box and watch the "new paradigm" stopgap 
"solution" unfold. Up to now I only saw the confusion grow, kudos to 
anybody that still knows how the configure their digi according the most 
recent proposal (where "most recent" means just that, a moving target). It 
is your APRS protocol, if you insist to keep Kantronics boxes forever then 
so be it.

Kind regards,

Henk.