--- Begin Message ---
Hi all,
I think that we cannot speak about CoA or HoA before discovering if we are in a home network or in a foreign network. I think that it will be made as follows:
At bootstrap step (this means that the MN has neither HoA, nor HA-address...) the MN connects to a network, discovers a link-prefix and configures an address (MN-address). At this point it does not know if this address is a HoA or a CoA. The MN has to discover a HA and home-prefixes (prefixes for which the HA acts as Home Agent):
If the link-prefix (= MN-prefix) belongs to the home-prefixes then the MN is in a home network and the HoA = MN-address (MN-home-prefix = MN-prefix).
If the link-prefix does not belong to the home-prefixes then the MN is in a foreign network and the CoA = MN-address. In this case, the MN asks for a HoA from the HA in the first BU.
At a handover step (this means that the MN was connected to a network and moves to another network) the MN has already a HoA, a HA address, may be a CoA.... When the MN connects to a network and discovers a link-prefix it compares this prefix to the MN-home-prefix:
If the same then the MN is in its home network and sends a BU to HA (if necessary).
If not then the MN is in a foreign network. It will configure a CoA and send a BU to HA.
Kassi
________________________________
De : mext-bounces at ietf.org [mailto:mext-bounces at ietf.org] De la part de Behcet Sarikaya
Envoyé : mardi 25 mars 2008 22:42
À : Vijay Devarapalli; Julien Laganier
Cc : mext at ietf.org
Objet : Re: [MEXT] Home Link Detection
Julien, your scenario is not according to SDO links. I think that the home link operation work item should concentrate on p2t SDO home link operation and should complement previous bootstrapping work.
Please see my comment inline below.
Behcet
----- Original Message ----
From: Vijay Devarapalli <vijay.devarapalli at azairenet.com>
To: Julien Laganier <julien.IETF at laposte.net>
Cc: mext at ietf.org
Sent: Wednesday, March 19, 2008 7:41:24 PM
Subject: Re: [MEXT] Home Link Detection
Julien Laganier wrote:
> Vijay,
>
> Still <chair hat off>:
>
> Vijay Devarapalli wrote:
>> Julien Laganier wrote:
>>> Vijay Devarapalli wrote:
>>>> In RFC 3775 and 3963, it is assumed that the mobile node is
>>>> pre-configure with the home prefix. So when it sees a router
>>>> advertisement with the same prefix, it knows it is on the home
>>>> link.
>>> When RFC 5026 is used, the MN has no pre-configured home prefix but
>>> get it dynamically in the MIP6_HOME_PREFIX attribute obtained via
>>> the IKEv2 exchange of the bootstrapping mechanism. Then it simply
>>> behave as usual, i.e., quoting you "when it sees a router
>>> advertisement with the same prefix, it knows it is on the home
>>> link."
>> The sequence of steps when the MN boots up on the home link would be
>> the other way around. The MN would first boot up, setup the link and
>> then run IKEv2 with the home agent. Not run IKEv2 with the home agent
>> and then send a router solicitation on the home link to trigger a
>> router advertisement.
>
> This isn't what I had in mind. What I had in mind is:
>
> 1) MN would boot up
> 2) Setup the link:
> - send an RS on the link
> - receives an RA with PIO from the link
> - configures a CoA for the link
> 3) Run IKEv2:
> - receives MIP6_HOME_PREFIX
> 4) Compare MIP6_HOME_PREFIX and PIO
> - same: conclude it's at home
> - different: conclude it's not at home
Ok.
[behcet] Not OK. According to http://tools.ietf.org/id/draft-ietf-dime-mip6-split-07.txt
IKEv2 happens with EAP in authentication phase which in SDOs happens when immediately MN enters the network.
So home prefix assignment occurs after CFG_REQUEST is received. http://tools.ietf.org/id/draft-ietf-dime-mip6-split-07.txt lacks any discussion on how the home prefix is assigned
>>> I don't see what requires additional specification.
[behcet] Sorry, I do think it requires additional specification.
>> There is nothing that specifies that the mobile node should first
>> setup a the p2p link that connects it to the home agent, then run
>> IKEv2 from the home link and obtain the home prefix, then compare
>> the prefix obtained from IKEv2 with the prefix obtained from the p2p
>> link setup and finally conclude it is on the home link. This is not
>> obvious. The MIP6_HOME_PREFIX attribute was designed for home address
>> auto-configuration. Not for home link detection.
>
> I don't see what isn't obvious,
:) Well, if it is obvious for everyone, we are done.
> and it doesn't matter what the
> MIP6_HOME_PREFIX was "designed" for.
It does. Implementing the MIP6_HOME_PREFIX attribute is optional,
since it is supposed to be used for home address auto-configuration.
If you are not supporting home address auto-configuration in your
system (for various reasons), then you don't even have to implement
this attribute.
Vijay
> The prefix received in the RA's PIO is an on-link prefix. If the
> MIP6_HOME_PREFIX from which the MN configures a home address is the
> same as the RA's PIO, then the home address belongs to the RA's PIO,
> therefore the home address is on-link, therefore the MN is at home.
>
> --julien
_______________________________________________
MEXT mailing list
MEXT at ietf.org
https://www.ietf.org/mailman/listinfo/mext
_______________________________________________
MEXT mailing list
MEXT at ietf.org
https://www.ietf.org/mailman/listinfo/mext
--- End Message ---
