Media%20Independent%20Pre-authentication%20Framework%20for%20Inter-domain%20Handover%20Optimization%20draft-irtf-mobopts-mpa-framework-01.txt

1
Media Independent Pre-authentication Framework
for Inter-domain Handover Optimizationdraft-irtf-
mobopts-mpa-framework-01.txt

• Ashutosh Dutta (Telcordia)
• Victor Fajardo (Toshiba)
• Yoshihiro Ohba (Toshiba)
• Kenichi Taniuchi (Toshiba)
• Henning Schulzrinne (Columbia University)

2
Changes from previous presentation

• Based on feedback from IETF 69 at Chicago,
  comments were incorporated and new draft was
  submitted as a RG item
• Solicited comments from the RG members and expert
  comments from Christian Vogt and Marco Liebsch
• Incorporated these comments and submitted the new
  version of the draft

3
Overall summary of comments

• Strength of this framework
• Based on security relationships between a mobile
  host and different networks
• Does not depend on trust or security
  relationships between administratively separate
  networks.
• Provides advantage in terms of deployment
  scalability, in particular in a more and more
  heterogeneous Internet
• The draft is of good editorial quality
• Describes related work and clearly explains how
  it differs from that
• Very good work!
• Lot of things have been considered
• Different means for address acquisition, etc.
• Individual protocol components and needs have
  been evaluated thoroughly
• Some modifications to existing components, such
  as address configuration, MM, local routing table
  management on MNs are needed for MPA to work in a
  particular system

4
Detailed Comments from Christian (1/2)

• Comment 1 Why the proactive handover tunnel
  connects the MN's old PoA with the nAR, and not
  the MN's new PoA with the oAR?
• In case of Inter-domain handoff, SA between
  network nodes in different domains is a hard
  assumption
• Proactive BU is possible with the tunnel to the
  nAR
• Works with any mobility management protocol
• Comment 2 How does MPA react in case of quickly
  fading signal
• Description has been added in Section 8.1

5
Detailed Comments from Christian (2/2)

• Comment 3 What is the relationship between
  mobility and transmission Delay
• Added description to make clear that one-way
  transmission delay for the in-handoff packets is
  dependent upon the buffering modules in MPA
• Comment 4 Description of jitter is not proper
• Sentence has been rephrased as suggested
• Comment 5 2 packet loss as an average value is
  may not be valid for handoff
• In-handoff packet loss can be 100
• This sentence has been rephrased
• Comment 6Optimal condition for the mobile to
  switch to a specific network
• Recommendation on specific metrics (e.g., SNR,
  Location, Cost) that should be used to decide the
  handover has been discussed

6
Detailed Comments from Marco (1/2)

• Comment 1 Description about on-link router
  advertisement in the runnel to trigger binding
  update
• description about the method for receipt of
  on-link router advertisement added in Section 7.6
• Comment 2 DAD should be part of address
  configuration not address resolution
• Description on DAD has been moved to Section 7.3
• Added new description for stateless
  auto-configuration
• Comment 3 Why is the pre-authentication through
  the previous interface is required
• Added few valid scenarios for use of multiple
  interfaces Battery power saving, Non-overlapping
  access, carriers may not allow simultaneous
  activation of both the interfaces

7
Detailed comments from Marco (2/2)

• Comment 4 How can one do authentication to CDMA
  network via WiFi network
• Clarified few steps regarding MPA applicability
• Depending on the access type, some or all parts
  of MPA steps would be applied
• e.g., in WiFi-CDMA operation PPP context can be
  set up ahead of time
• Comment 5 What is the delay effect on packets
  due to proactive tunnel
• In-handoff packets over the transient tunnel
  would be subjected to added delay if the
  neighboring access networks are far apart in
  terms of IP routing path
• Text has been added to describe the tradeoff
  between longer packet delay vs. packet loss