RE: [802.21] [DNA] Prefix information for link identification in DNA
IMHO, any "make before break" mechanism would result in relatively
better performance than hard handover but that is not due to having L2
MIH services. Any L2 signaling (including beacon) is sufficient to
determine that a new link is detected, there is no need for MIH or BER
threshold here. It is a separate thing that we have MIH capability bit
in the beacon.
For a UE that wants to handover, it can actively request for a specifc
QoS (including BER) based on its application needs using the native L2
specific signaling. This is also true in WLAN where 802.11k is working
on the radio measurments aspects.
________________________________
From: ext Peretz Feder [mailto:pfeder@lucent.com]
Sent: Friday, September 30, 2005 12:03 PM
To: Sreemanthula Srinivas (Nokia-NRC/Dallas)
Cc: STDS-802-21@listserv.ieee.org
Subject: Re: [802.21] [DNA] Prefix information for link identification
in DNA
On 9/30/2005 12:46 PM, Srinivas.Sreemanthula@nokia.com wrote:
Think MIP. It gets a trigger from MIH to switch
over into an
alternate ethnology which then assigns the IP
through MIP
interaction. By saying MIH is available after IP
is assigned
you killed MIP as a user.
We should be careful not to mix link and MIH triggers on
the local stack
with remote triggers between two nodes. The transport
under discussion
is for "remote CS/ES".
Target BS MIH sends BER uplink threshold crossing event to MIH
in UE.
Think MIP again. It detects movement when FA
advertises. It
can take minutes no msecs.
This is exactly why I said you must have a specific
handover procedure
in mind. The handover procedure itself is slow and not
suitable for RT
services irrespective of how the MIH services are
provided (that's why
they have fast handover). One of the main latency issue
in IPv4 (basic
MIP) is due to IP address configuration. It is a little
better in IPv6,
you can send a Router Solicitation and do stateless
address auto
configuration.
I would sincerely like to know how MIH services at L2
can solve this
problem.
Make before break to start the slow MIP Reg/Reply process ahead
of time as a result of the alternate and preferred technology BER
threshold crossing.
802.21 MIH aids in handover, but we should remember that
the
overall handover performance depends on the L3 handover
procedure
itself.
-----Original Message-----
From: ext Peretz Feder
[mailto:pfeder@lucent.com]
Sent: Friday, September 30, 2005 11:04 AM
To: Sreemanthula Srinivas (Nokia-NRC/Dallas)
Cc: STDS-802-21@listserv.ieee.org
Subject: Re: [802.21] [DNA] Prefix information
for link
identification in DNA
On 9/30/2005 11:49 AM,
Srinivas.Sreemanthula@nokia.com wrote:
OK, good we cleared that.
We are doing inter-tech (or
iinter-subnet) handovers of IP
services, so
the UE already has the IP connectivity
both to current and
target (via
current).
Think MIP. It gets a trigger from MIH to switch
over into an
alternate ethnology which then assigns the IP
through MIP
interaction. By saying MIH is available after IP
is assigned
you killed MIP as a user.
Eliminating MIP as an MIH user is a fatal
mistake for 802.21,
as MIP is the numero uno MM user IMO.
On the issue of performance, It all
depends on the specific handover
scenario and procedure you are referring
to. Another related
issue is
the relative performance gains between
implementing L2 MIH vs L3 MIH.
If the IP services handover delay
constraints can be met with
L3 MIH, I
don't need the additional 10ms
improvement with the L2 MIH.
Think MIP again. It detects movement when FA
advertises. It
can take minutes no msecs.
________________________________
From: ext Peretz Feder
[mailto:pfeder@LUCENT.COM]
Sent: Friday, September 30, 2005
10:07 AM
To:
STDS-802-21@listserv.ieee.org
Subject: Re: [802.21] [DNA]
Prefix information for link
identification
in DNA
"I do not understand how any one
would conclude that
the MIH services
are only between UE and the AP/BS."
The discussion is PoA and not
services. The 1st PoA
could be L2 for IS
and CS. With no PoA at L2, the poor UE
will have no MIH
services until
IP is established. The performance will
be very different, not to
mention a UE with a bridging only
attributes, such as 802.16 terminal
with only Ethernet CS (no IP CS).
Nobody is saying MIH services
are strictly between UE and BS.
Performance will be better when PoA L2
MIH is established.
Peretz
On 9/30/2005 10:50 AM,
Srinivas.Sreemanthula@nokia.com wrote:
The MIIS is provisioned
between MIH in UE to a
network counter part
any
where in the network.
This network node can
either act as a proxy
info
server or an info
server. We also identified
MIIS requires L3 and
hence
the WG went through the
exercise of identifying
all the UL
requirements
and establish
coordination with IETF. However,
in that discussion,
there
was no reference to
whether the AP/BS was MIH
or non-MIH capable.
Even if we leave out the
info services from the
discussion, I do not
understand how any one
would conclude that the
MIH services are only
between UE and the
AP/BS.
-----Original
Message-----
From: ext Peretz
Feder
[mailto:pfeder@lucent.com]
Sent: Friday,
September 30, 2005 9:39 AM
To: Sreemanthula
Srinivas (Nokia-NRC/Dallas)
Cc:
STDS-802-21@listserv.ieee.org
Subject: Re:
[802.21] [DNA] Prefix
information for link
identification
in DNA
Are you
indicating attaching to a non
MIH enabled AP/BS and
receiving MIH IS
over R4 from a remote
MIH info server?
On 9/30/2005
10:27 AM,
Srinivas.Sreemanthula@nokia.com wrote:
Did we
miss the whole
discussion of MIH information services?
________________________________
From: ext Peretz Feder
[mailto:pfeder@LUCENT.COM]
Sent: Friday, September 30, 2005
9:16 AM
To:
STDS-802-21@LISTSERV.IEEE.ORG
Subject: Re: [802.21] [DNA]
Prefix information for link
identification
in DNA
"you have first to be very clear
about what you're attaching"
I would think that in 802.21, we
first attach the UE's
MIH to a BS/AP
that
supports MIH capability.
On 9/30/2005 8:55 AM, Stefano M.
Faccin wrote:
Mike, well said!
Stefano
________________________________
From: ext Mike Moreton
[mailto:mm2006@MAILSNARE.NET]
Sent: Fri 9/30/2005 3:09
AM
To:
STDS-802-21@listserv.ieee.org
Subject: Re: [802.21]
[DNA] Prefix information for link
identification in DNA
To extend (I think!)
Stefano's point, before
determining what
the PoA
is, you
have first to be very
clear about what you're
attaching. Just
saying
"the terminal" makes no sense,
because different layers in the
terminal's protocol stack attach to
different places in the network.
For example, the PHY
layer attaches to the AP,
but the TCP
layer
attaches
to the destination host.
Mike.
-----Original
Message-----
From: Stefano M.
Faccin
[mailto:stefano.faccin@NOKIA.COM]
Sent: Friday,
September 30, 2005 1:08 AM
To:
STDS-802-21@listserv.ieee.org
Subject: Re:
[802.21] [DNA] Prefix
information for
link
identification
in DNA
Yoshihiro,
I'm not sure why
should restrict the
term PoA to have
only a
L2 meaning as
you suggest below. I
think we should
distinguish
clearly between L2 PoA and L3 PoA.
For me,
the L3
PoA is where the
terminal gets IP conenctivity.
E.g. for
GPRS
the L3 PoA is
the IP link on which the
GGSN is located.
In
L2, PoA is the
point where the access-specific
L2
connection
terminates (e.g.
an AP in 802.11).
Stefano