Re: [manet] New DLEP extension draft for WG

On Wed, Jan 27, 2021 at 4:53 PM Rick Taylor
<rick@tropicalstormsoftware.com> wrote:
> > I finally got to the point where I had time to publish the first
> > parts of my DLEP extensions I would like the Manet WG to adopt. I
> > talked with Ronald and we agreed that I keep the drafts quite
> > "narrow".
>
> "Narrow" is good, but in my opinion if you have <6 data items that are
> all very much in the same context, e.g. Radio Signal Info, it mike make
> sense to roll them all up into 1 extension.  I suppose the test would
> be that if an implementation supports 1 data item, then it probably
> could support the others, but that is a little subjective.

I will post a the two "lists" of PHY TLVs tomorrow so you can see the
split I plan to do.

> The disadvantage is that you then need to specify which data items MUST
> be used by an implementation supporting the extension, and which are
> optional.

> > https://datatracker.ietf.org/doc/draft-rogge-manet-dlep-radio-band/
> >
> > At the moment I plan to do two more drafts about PHY aspects of the
> > radio (signal strength related stuff and channel utilization) before
> > moving to MAC related TLV extensions.
>
> I think these radio data items sound like they are logically all part
> of the same extension.
>
> I do have a question: As DLEP is primarily about informing layer-3
> devices about dynamic layer-2 capabilities, are radio signal metrics
> really of use to a router?  To me they sound a lot like layer-1
> information, if you get my meaning, and I wonder why a router would
> even care?

First option I see is a routing metric that takes the EM spectrum cost
into account...
twice as fast is good... but twice as fast with 10 times the bandwidth
might not be the best choice if you want to maximize the total
capacity of the network.

> Can you explain why you think these data items are important to a
> layer-3 device?  I think I'm looking for a "rationale" section in the
> draft.

There is also the option to (re-)use these options together with
"Request Link Characteristics", which could be interesting for
Software Defined Radio.

Last (e.g. for military networks) it might be interesting to determine
if a "jammer warning" in a certain frequency band is relevant to your
radio so you can either use a different link or reconfigure.

I hope this makes sense...

Henning

> I understand that you have done some work before around Airtime Cost
> Metrics derived from signal information, so I wonder whether the modem
> should be exposing a derived data item (post-processing) rather than
> the raw radio signal data items (pre-processing).  My assumption is
> that the DLEP protocol is significantly slower than the computation.
>
> I suppose the counter-argument is that there is really no such thing as
> a "useless" metric, and I can imagine some routers in the near future
> using AI to predict link behaviour - I expect someone is already doing
> a PhD on that right now!
>
> Cheers,
>
> Rick
> _______________________________________________
> manet mailing list
> manet@ietf.org
> https://www.ietf.org/mailman/listinfo/manet