[H.323 Mobility:] questions on MTD-016

[Kumar, Vineet]

Radhika,

I wasn't aware of any issues raised on my contribution. In the last
conference call I just summarized my contribution; we never had time to go
over it. But if you see any issues, please bring them up on this reflector.
Did anyone else bring up issues on my contribution that I am not aware of ?

vineet

-----Original Message-----
From: Roy, Radhika R, ALARC [mailto:rrroy at att.com]
Sent: Friday, April 14, 2000 9:25 AM
To: Kumar, Vineet
Cc: 'ITU-SG16 at MAILBAG.INTEL.COM'
Subject: RE: [H.323 Mobility:] questions on MTD-016


Hi, Vineet:

Yes, you are right that multicast should NOT be the only way to discover the
GK. That is why, I appreciate your approach to use DNS in combination with
GRQ as proposed last time in your contribution. I hope that you will
finalize your contribution resolving the remaining issues pointed out in
your contribution.

You are right that MGA's approach is also like mobile IP.

Best regards,
Radhika

-----Original Message-----
From: Kumar, Vineet [mailto:vineet.kumar at intel.com]
Sent: Thursday, April 13, 2000 7:33 PM
To: Roy, Radhika R, ALARC
Cc: 'ITU-SG16 at MAILBAG.INTEL.COM'
Subject: RE: [H.323 Mobility:] questions on MTD-016


Radhika,

We cannot assume the existence of multicast, so multicast cannot be the only
mechanism for discovering the gk. The terminal will have to try various
methods to discover the gk. These methods are: use of DNS, use of
pre-configured or previously cached IP address of the gk, and finally the
use of multicast.

Regarding gk discovery through multicast, in H.323 the terminal initiates
the process of finding its gk by sending GRQ on the multicast address. You
are saying that this method will cause a lot of traffic if 100s or 1000s of
terminals try to simultaneously discover their gks through this method. Your
solution is to have the gks periodically send the MGA message which is used
by terminals to obtain the addresses (plus other information)of the gks.
This type of approach is, I believe, also used in Mobile IP. Seems like a
reasonable approach assuming the MGA messages are properly scoped, and the
discovery/registration messages between the user and the gk are
authenticated and their integrity checked.

Regards,
vineet
-----Original Message-----
From: Roy, Radhika R, ALARC [mailto:rrroy at ATT.COM]
Sent: Wednesday, April 12, 2000 3:09 PM
To: ITU-SG16 at MAILBAG.INTEL.COM
Subject: Re: [H.323 Mobility:] questions on MTD-016


Vineet:

I am just pointing out about only one point of yours: "H.323 already has
mechanisms for discovering the gatekeeper. "

Yes, it is true: GRQ is the mechanism that is used today in H.323.

The time it was decided to use GRQ for discovering the GK, it was not
envisioned for the highly mobile communications environments like cellular
(wireless) network or a combination of cellular (wireless), wireless LAN,
and/or wire-line network.

The fundamental question is: Does the same GRQ mechanism make sense to use
for the H.323 mobility architecture that we are considering now because of
the heavy traffic generated (and associated problems) by the 100s, 1000s of
mobile users that will access to a cell?

AT&T's contributions MD-017 and MD-018 have addressed this question:
Listening to the MGA (Mobility GK Advertise) message (GRQ message can only
be sent if the MGA message is not received within the certain time
interval).

Like you, I also have the same questions 1, 2, and 3 to Steve.

Best regards,
Radhika

> -----Original Message-----
> From: Kumar, Vineet [SMTP:vineet.kumar at INTEL.COM]
> Sent: Wednesday, April 12, 2000 5:17 PM
> To:   ITU-SG16 at MAILBAG.INTEL.COM
> Subject:      [H.323 Mobility:] questions on MTD-016
>
> Stephen,
>
> I have a couple of questions on your contribution MTD-016. These are:
>
> 1. H.323 already has mechanisms for discovering the gatekeeper. Are you
> suggesting in your contribution that the terminal should discover the VLF
> instead of the visiting gatekeeper ? Or, are you assuming that the VLF is
> integrated in the visiting gatekeeper ?
>
> 2. In H.323, authentication of the terminal and the gatekeeper is done at
> the time of discovery. In fact, in H.323 all messages between the terminal
> and the gatekeeper can be authenticated and the message integrity
> preserved. In your contribution, authentication is done at the time of
> registration. Why is this preferable to what is already in H.323 ?
>
> 3. What is the reason for the information flow from the HLF to the home
> gatekeeper, and from the home gatekeeperr to the HLF ? I don't think we
> can assume that there is only one home gatekeeper that the terminal may be
> using. In fact, the home gatekeeper may not have any information about the
> user.
>
> Regards,
> vineet