Purpose
=======
802.21 Higher layer requirements for IETF
Date
====
July 26, 9am-11am EST.
Participants
=========
Alistair Buttar, Subir Das, Stefano Faccin, Peretz Feder,
Andrea Francini, Prasad Govindarajan, Eleanor Hepworth,
Benjamin Koh, Kalyan Koora, Hong-Yon Lach, Xiaoyu Liu, Andrew
McDonald, Yoshiro Ohba, Ajay Rajkumar, Reijo Salminen, Ajoy
Singh, Srinivas Sreemanthula, Qiaobing Xie
(I apologize in advance if I missed somebody, as I'm sure I
did; also, i apologize for any mispelling)
Discussion
========
* Ajay summarized discussion that took place last week at
IEEE meeting regarding 802.21 and IETF
* the current result of the discuss with Gabriel
Montenegro (chair of MIPSHOP WG) is that the MIPSHOP is
willing to take up IS-related work through re-chartering.
Requirements would have to come from 802.21 WG. The MIPSHOP
WG chair made clear that ES and CS most probably do not fit
the MIPSHOP framework
* Ajoy brought up CARD applicability. It was agreed that
the L3 requirements are being worked out and the protocol
selection is out of scope at this time
* Stefano presented the high-level kickoff slides
(previously distributed)
* With respect to next IETF: Stefano indicates he will
give up the slot currently allocated to the Faccin/Daley ID
to present the requirements coming from 802.21. Also, a MIHEP
Bar BOF will take place to complement the 20min slot in
MIPSHOP at IETF meeting.
* Ajoy commented that ES and CS need not be on L3. No
real discussion took place, since it was agreed that present
focus (due urgency to provide requirements for IS to IEEE.
* The question of what is "L3 transport" came up. The
term may be misunderstood by IETF (e.g. Gabriel had indeed
misunderstood it), and there does not seem to be complete
consensus in 802.21 yet. Comments were raised that if by "L3
transport" for 802.21 we actually consider just transport
aspects, in theory 802.21 could define the protocol by itself
and then specify TCP or UDP transport, and ask IANA for
allocation of port numbers.
* During the discussion it was indicated that by "L3
transport" we mean also architectural aspects such as
discovery of MIHF functions/capabilities and security (i.e.
aspects that are more protocol oriented)
* Discussion led to identifying three scenarios: (1)
802.21 defines only IEs, IETF defines the transport aspects,
and no protocol definition takes place; (2) 802.21 defines
both the IEs and the protocol, and IETF defines the transport
aspects; (3) 802.21 defines the IEs, IETF defines the
transport aspects, and 802.21 and IETF collaborate in
defining the protocol. Security aspects are definitely
defined in IETF (out of scope for 802.21). discovery aspects
are defined by 802.21 and specified in IETF. Ajay also
indicated that the target at present is (2) or (3)
* Some parties commented that (3) is more in line with
the way IETF works
* As for discovery aspects, some parties indicated that
it can be part of work already on-going in other WGs, as an
extension of current discovery solutions or as part of host
configuration solutions
* Ajoy asked if we should first define the protocol
802.21, then bring it to IETF. Stefano indicated that timing
is very important and that we should not miss the current
opportunity we have with MIPSHOP willing to re-charter to
include 802.21 aspects. Stefano reminded that the
re-chartering must close soon (Gabriel indicated he needs to
provide the new charter to the ADs just after the next IETF,
but Gabriel mentioned he can stay a bit vague to allow for
adjustments)
* Ajoy asked if #1 can be more suitable for the success
of 802.21, i.e. 802.21 would not need to have the work in
IETF completed before saying it has completed its duties. WG
think #3 would be better for the success. Ajay reminded that
the success of 802.21 does not depend on completion of work in IETF
* Discussion about basic and extended information
service. Kalyan asked if the "L3 transport" is only for
extended-set? No, it is applied to all of IS, since in some
scenarios it is relevant only for extended IS, in some other
also for basic IS
* Ajoy raised a question if two MIH servers can talk to
each other. It is not clear if two MIH functions in network
can talk to each other. Yoshi mentioned there is no need for
such communication. Kalyan asked how e.g. is the neighbor
graph exchanged? Yoshi mentioned that transferring neighbor
graph is out of scope of 802.21. Peretz indicated that one
scenario is where MIH is proxied, e.g. MIHF in UE talks to an
MIHF in the network it is attached to, and the MIHF in the
network proxies MIH information to another MIHF e.g. in the
home network. It was mentioned this could be decided later,
but since it affects the L3 requirements, Stefano suggested
to assume that there "may" be communication between two MIH
functions and discuss this later in the emails. Qiaobing also
reminded this discussion is closely related to the model
discussion that took place at the meeting last week. Benjamin
reminded that the MIH model discussed at the ad-hoc was not
agreed yet by the whole WG.
* Stefano presented 3 scenarios to trigger discussion for
L3 requirements.
* Yoshi pointed #1 and #2 are similar. Another scenario
was proposed (and numbered as #4): no L3 protocol is used
between the MIHF in the terminal and the MIHF in the PoA, L2
is used instead, but then from MIHF in PoA and MIHF in the
network a L3 solution is used. UE----L2--->sPoa---L3--->MIS
* Ajoy mentioned another scenario where
UE----L3--->sPoa----L3.---->cPoa or
UE----L3--->sPoa----L3---->MIH, Stefano replied it is a
subset of the current third scenario (but it will be
described explicitly)
* Ajay indicated that we still need to clarify to IETF
what we mean exactly by PoA, since it impacts this discussion
and may be confusing to IETF. Stefano suggested that a way
forward is to present to IETF example of PoAs, without
necessarily providing a comprehensive and exhaustive definition.
* Stefano indicates we need to consider two kinds of MIS
interface since requirements may be different and should be
at first looked separately (we can merge requirements if they
are the same)
* i) MIHF in UE to MIHF in network
* ii) MIHF in network to MIHF in network
* Qiaobing mentioned discovery should not be part of
transport requirements. It was emphasized that the discussion
is not just for plain transport (in IETF sense of the term)
but "L3 and above" requirements for MIIS. It was agreed this
needs ot be made very clear in slideset.
* Also, Qiaobing suggested that we separate the
requirements that relate only to transport from those that
relate to architectural/protocol aspects
* Hong-Yon asked why we are considering also protocol
requirements. Stefano indicated we should try to list all the
requirements we can come up with, then choose which one we
think are relevant for the discussion in IETF.
* Stefano will send out new slideset for discussion on
mailing list.
* It was agreed to send contributions to requirements at
least 4 (four) hours before the next tele-conference so that
the input can be consolidated
* WG is encouraged to discuss and send scenarios and L3
requirements by next conf meeting on Thursday 9 PM EST.