Re: [Autoconf] aspects of multi-hop wireless communication

Hello Alex,

I have to object!!

Alexandru Petrescu wrote:
> Observation should report the name of the link layer, otherwise a peer
> may think the author logically extends what s/he sees on one link layer
> to another (instead of experiencing same phenomena on each link layer
> individually).

Not really.  We've observed that various link technologies exhibit
some or all of the communication problems under consideration.  I do
not want to write a document specific to a particular link technology.
>
> The scientist orders the frog to jump, who so does.  Then cuts its legs,
> orders it again to jump; the frog obviously can't jump.  Scientist
> deduces the frog can no longer hear.  However, if the order were issued
> to the legs (instead of ears) in the first place then maybe the false
> conclusion could have been avoided.

Interesting story, but not characteristic of the abilities of the people
involved here.  I think we're a bit beyond such knee-jerk reactions
or frog-jumping contests.

>
>> The draft is not saying "there is a need for a solution". The draft 
>> is saying "this is what is often observed". That's it. I think we can
>>  agree on this ;)
>
> I agree one may have observed this behaviour.  I don't agree it is an
> often observed behaviour - I didn't.  I often observed wifi works ok in
> the types of deployments described in the draft (R1,R2,A).

First, you have to be kidding about nontransitivity.  That's obvious.
It even has a famous name -- "hidden terminal" problem.  You can
read about that in a google of places.

Second, take a look at the MIT Roofnet project papers.  That project
was a very strong re-verification of the well-known properties discussed
in the draft, along with numerous implications drawn out for IP routing
protocol behaviors.

I'd be very curious if you could find anyone who has seriously studied
the matter (e.g., refereed publications) to support your position.

Regards,
Charlie P.

_______________________________________________
Autoconf mailing list
Autoconf@ietf.org
https://www.ietf.org/mailman/listinfo/autoconf