Transcript PPT Version

Problem Statement for Common
Interface Support in Localized
Mobility Management
draft-corujo-ps-common-interfaces-lmm-00
Problem Space
..the document aims at identifying a common
problem space for the design of a MN-AR multipurpose interface covering fast link change
detection, proactive and reactive handovers in
LMD environments (such as Proxy MIPv6
solutions).
…starting from existing solutions for fast link
detection the document aims at discussing
possibilities for a layer 2.5 interface between MN
and AR
• Several solutions make use of link events:
– DNA/FRD and its 802.21 Integration
– Proxy Mobile IPv6 and reactive HO
– Fast Handovers for Proxy Mobile IPv6
• However so far not much is available for
MN/AR interface
– MN-AR interface based on ND
Where the common interface applies:
– Bootstrapping (e.g. power on of a MN)
– Proactive handover
– Reactive handover
Providing interfaces with 802.21
An example
ACCESS ROUTER
..................
|
........
|
|
| LMP
|
|
|
|ENGINE|
|
|
`''''''
|
|
/ \
| MIH
LINK EVENT
/
|
\
|
|EVENT
| |
|
|
| |
|
____________ +----------+
FROM MN
\|
|
|
OR PoA ____________/| FUNCTION |
|
|
MIH
|
+----------+
|
..................

LMP Engine as an MIH user?
PoAs co-located with AR?
Requirements
•
•
•
•
•
•
•
This framework should accommodate with future instances of the NetLMM
protocol, and be flexible enough to allow and support possible optimizations
of the NetLMM procedures.
This framework does not aim at replacing L3 procedures rather to improve
them by facilitating the information exchange between the host and the AR
even prior to full network configuration.
While several deployment scenarios of the MIHF and its relation to the
upper layers are possible (IP stack, LMP engine), a more clear
communication model needs to be agreed and specified.
Need to align different identifiers.
Link layer events have to be carefully registered by whoever can use them
to trigger procedures. A single event can be forwarded to more than one
high level entity, inducing parallel behavior that might not be desirable.
The host must be able to store the previous network configuration
information, both for detecting subnet changes upon attachment, and also
to report it to the nAR upon a reactive handover.
Sufficient authentication of devices supplying link-layer events has to be
considered. For example, reachability and attachment notifications may be
falsely asserted by an attacker
Views…ideas…comments?