RE: [MIP-QOS] Re: MIP-QOS -- Scope of discussions

[marc.greis@nokia.com]

James,

> I don't think the radio access network should be a black box. 
> We are not doing
> requirements here for existing 3GPP/2 radio access networks, but
> for the Internet. We should be doing our requirements so that
> it may be possible to design an all-IP radio access network that connects
> up with the Internet directly, without all the layers of interfaces
> that the current 3G designs require.

Exactly. But in my eyes, that rather implies that we should try to create a
generic QoS mechanism which supports IP mobility without looking too much at
requirements coming from radio access networks.

> There was a requirement expressed at the NSS BOF to have a QoS mechanism
> that allows a better match between wire QoS and the air. 

This requirement seems to aim very much at the parameters which are provided
by a QoS mechanism, i.e. is the problem which you are trying to solve that
current IP QoS mechanisms (e.g. RSVP) may not contain enough information
which may be useful for setting up QoS for a wireless connection (e.g.
information related to error ratios, etc.)? Is that correct? I am just
trying to understand what this requirement implies.

> I agree
> with this requirement. I think radio networks have fundamental
> Intsrv like characteristics, I think any new QoS mechanism needs to
> provide the ability to do a smooth translation between wired QoS
> and wireless QoS. 

What exactly do you mean by that? Can you perhaps give some kind of example?

> I think this does not imply an end-to-end IntSrv
> model. This doesn't mean that the design needs to specify 
> wireless QoS, however, that is naturally up to the radio link protocol.

Exactly. And this radio link protocol may be quite different from one
wireless technology to another, which implies that it is potentially very
hard to develop a single mechanism which translates smoothly into all
potential wireless QoS mechanisms.

Marc