Re: [802.21] Meeting minutes of today's ad-hoc teleconference
Hi Yoshihiro,
Thanks for your taking minutes.
I'd like to correct a part of the minutes which is about
what I said in the telecon last night.
As for Ia interface,
I meant that "it would be (n:n), not end-to-end
as multiple UEs also can communicates with one NISP.
But, if Ia is an interface between one UE and the network like Subir said,
(1:n) will be fine.
By the way,
when will it be the case of (1:n)?
My understanding is that one UE can communicate with only one NISP at a moment.
The reason for having (1:n) would be the case that the UE moves so needs to connect
with a different NISP from the previous one. Am I right?
Regards,
Eunah
----- Original Message -----
From: "Yoshihiro Ohba" <yohba@TARI.TOSHIBA.COM>
To: <STDS-802-21@listserv.ieee.org>
Sent: Wednesday, August 24, 2005 1:40 AM
Subject: [802.21] Meeting minutes of today's ad-hoc teleconference
>
> Meeting minutes of ad-hoc teleconference about
> IS reference model and use-cases
>
> August 23rd, 9am-11:10am EST
>
> Discussed document: 21-05-0336-00-0000-IS_Reference_Model_and_Use_Cases.ppt
> (posted on the reflector.)
>
> Participants
> ------------
>
> Subir Das
> Farooq Bari
> Vivek Gupta
> Ulisis Olvera-Hernandez
> Qiaobing Xie
> Ajoy Singh
> Ajay Rajkumar
> Peretz Feder
> Soohong Daniel
> Stefano Faccin
> Cheng Hong
> Eunah Kim
> Kalyan Koora
> Yoshihiro Ohba (minutes taker)
> Mattias Pettersson
> Prasad Govindarajan
> Srinivas Sreemanthula
> Reijo Salminen
> Eleanor Hepworth
>
> What was agreed
> ---------------
>
> - Ia: joint-scope of IEEE/IETF.
>
> - Ia': joint-scope of IEEE/IETF if it is different from Ia. If Ia'
> and Ia are the same, just replace Ia' with Ia.
>
> - Ix: out of the scope of IEEE.
>
>
> Action item
> -----------
>
> - Upload the current slide.
> - Update the slides based on the agreement.
> - Before the next teleconf, create a strawman of actual IETF requirements
>
> Detailed discussion
> -------------------
>
> [Subir] First slide is to capture rererence model. First figure
> describes single hop model. 2nd figure multi-hop model. Single-hop
> basically means IP link (in terms of IETF req.). Network IS provider
> is where we can get the information. In Information Database,
> information is stored. Interface Ix is not 802.21 scope. This is Ix
> is for capturing the scenarios. Ia' could be Ia, we don't see much
> differences.
>
> [Ulisis] Why Ia' could be Ia could be different?
> [Subir] Ia' some information may be added by 802.21 ISF in the network.
> [Ulisis] Are we going to define an intermediate function?
> [Ulisis] Each NISP tends to have its own information database. How it
> is covered.
> [Subir] It is captured in the first case.
>
> [Ajoy] A protocol is defined between MN and AR?
> [Subir] Placement of protoocl entity can be separately discussed.
> [Subir] Two NISP communicating each other is covered in the reference model
> [Ajay] This discussion is only for IS. Two IS functions in the
> network talking to each other is separate discussion.
>
> [Subir] If we define interfaces, why can't we use the same interface for
> communications between IS Functions in network.
>
> [Farooq] What is the diffrence in requirements? If two diferent autonomous systems
> communicate, then the interface is Ia'.
>
> [Ajoy] communications between IS Functions in network can be a
> peer-to-peer model. Ia is client-server model.
>
> [Subir] It is our job to put requirements if Ia' has some difference from Ia.
>
> [Ajay] Definition of IEs can be done in parallel to the discussion of
> reference model and use-cases discussions.
>
> [Ulisis] Upper-layer proxy is talking to information database via Ix?
>
> [Subir] Yes, but the two end of Ix are not an IS function. Ix can be
> anything. We can mention it it is outsope.
>
> [Qiaobing]: There is another model we may need to capture. For
> example, IS function in UE is communicating with a proxy that is also
> inside the UE and then it is using the Ix interface to communicate
> with the information database.
>
> [Subir] Yes, that is a valid model but it will out of scope since Ix
> is out of scope. But we will capture this model and Xiaobing will
> send the model diagram.
>
> [Eunah] What is the meaning of "(1:n)"?
>
> [Subir] It means UE can communicate with multiple MIHFs.
>
> [Yuna] Ia could be end-to-end.
>
> [Farooq] Mapping is UE to the network.
>
> [Ajoy] What is end-to-end? How UE can choose one MIHF.
>
> [Ajay] Ia is possible IS function and information database are
> combined.
>
> [Farooq] Ia and Ia' may have different security characteristics.
>
> [Kalyan] Can two UEs communiating each other via Ia? Another UE may
> have obtained the information from the network.
>
> [Subir] It might be a multi-hop model. It could be described in Case
> 1 or 2 as a note.
>
> [Kalyan] The intermediate network may be an ad hoc network. The term
> NISP in that case should be ISP
>
> [Farooq] Is it not then the network from an UE?
>
> [Ajay] We are going to some philosophical discussion here.
>
> [Ajoy] How many interfaces we are discussing in 802.21.
>
> [Subir] Ia and Ia'.
>
> [] Not all elements should have information database.
>
> [Farooq] It should be captured in the spec.
>
> [] All IS Function in network has an access to the information
> database. Then why Ia' is needed?
>
> [Subir] Multihop model is trying cover AAA-proxy like scnearios.
>
> [Ajay] I agree with Subir.
>
> [Ulisis] From the UE's perspective, Ia' or Ix does not matter.
>
> [Subir] True. From the UE perspective it is a single model, similar
> to AAA model.
>
> Slide 5:
>
> [Peretz] Is this scoping discussed in Paris IETF meeting?
>
> [Subir] Yes. If there is a scenario that is missing, please bring it.
>
> [Kalyan] Ix is now IETF scope? If there is Ix and it is IETF scope,
> why do we want to show this interface?
>
> [Subir] We are not saying that Ix is IETF scope, we are questioning
> about this.
>
> [Kalyan] Is Ix implementation specific?
>
> [Ajay] Yes.
>
> [Farooq] But we need to say something about Ix to explain the scenarios.
>
> [Ajoy] What is the meaning of joint-scope?
>
> [Subir] IEEE relies the interface to be standardized in the IETF and
> requirements are sent to IETF.
>
> [Peretz] Are you saying that communication betweeen IS functions
> between different NISPs are in the scope of 802.21?
>
> [Subir] Yes.
>
> [Ajay] Case 1 is ientifying multi-hop case. In the slide we may need to
> explicilty mention about this.
>
> [Subir] We will separate Case 1 into multiple cases. One case with
> only one Ia and the other with two Ia-s. And some other.
>
> [Ajoy] Ia' should be in scope. Ia could be used instead of Ia' when
> one NISP is acting as an independent client for the other NISP (Ia''?)
>
> [Farooq] What Ia'' is diffrent from proxy?
>
> [Subir] We need to identity in various scanrios raised during this
> discussion. the interface is Ia, Ia' or some other. (proxy, relay or
> server)
>
> [Ohba] UE-to-UE communication in Case 1 might have an issue about MIH
> discoverying. There will be a big problem if every UEs are trying to
> anser discovery query.
>
> [Subir] Issues should be discussed when trying to create another use
> cases for UE-to-UE communication information service and we can try to
> describe requirements about it.
>
> [Kalyan] Broadcast-based information service is another use case.
>
> [Subir] Ia' is in scope only when it is for proxy and server.
> If server to server then it is Ia''
>
> [Farooq] Ia' or Ia'' is just a client-server interface.
>
> [Vivek] That is mostly out of scope of 802.21?
>
> [Subir] If we don't see different Ia and Ia', let's just rename Ia' to Ia.
>
> [] Why server-to-server communication is out of scope?
>
> [Subir] To make it in scope valid scenarios is needed.
>
> [] Can Ix go across NISPs?
>
> [Ajay] No. If every NISP has its own information database, why Case
> 2c and Case 3 are needed? It is unreastic to consider a NISP that
> does not have its information database.
>
> [Subir] If we are considering models where NISP always has information
> database, then Case 2c and Case 3 are not needed.
>
> [Ajay] Once agreed, the contents of the slides can be included in the
> spec during the september meeting.
>
> [Vivek] Before the next teleconf, we really need to create a strawman
> of actual requirements.
>
> [Subir] I agree.