Last Modified: 2005-07-10
Monday, August 1 at 0900-1000
CHAIRS: James Kempf <firstname.lastname@example.org>
Phil Roberts <Phil.Roberts@motorola.com>
SCRIBES: Spencer Dawkins <email@example.com>
Total Time: 1
- Agenda discussion, co-chairs
Agenda was abused for being unreadable, but not bashed.
- Problem statement, James Kempf, DoCoMo Labs USA
Why not just use global mobilty management whenever you move? Drop packets if your anchor point is a long way away, a lot of signaling to come up on a new subnet, changes in care-of addresses tell eavesdroppers a lot about where mobile nodes actually are at any given point.
<>What has changed in last year? IETF has been working on HMIP, FMIP, LLMIPv4 (all experimental, but FMIP has been moved to standards track). Micromobility hasn't really progressed in IETF. HIP and MOBIKE are popping up, and WLAN switches have taken over (which doesn't require IP address changes within a local enterprise network, doesn't require MN protocol stack changes, and which customers really like).>
Problems with experimental IETF protocols - changes to host stack, designed for MIPv6, requires security associations, virus/malware can expose local care-of address for attack.
Problems with WLAN protocols - proprietary, scaling capability unclear.
- 10 min. - Requirements and Gap Analysis, Gerardo Giaretta, Telcom Italia Lab
ID has two objectives - requirements and gap analysis.
Solutions identified so far - MIPv6 with local HA, HMIPv6, Micromobility protocols.
Must improve handover performance, reduce signaling volume, ensure location privacy, use wireless resources efficiently, reduce signaling within wired network, no extra security between MN and network, support heterogenous wireless, support unmodified hosts, support for both IPv4 and IPv6.
Lots of gaps for MIPv6 with local HA (only meets #5, #7). HMIP has same gaps. Micromobility protocols meet most requirements but have scalability and deployability problems.
Conclusion - new work is required.
- Localized Mobility Management for 3GPP All IP Networks (AIPN) with
New Access Technology, Katsutoshi Nisida,DoCoMo YRP
3GPP now looking at "evolved UTRAN" - higher data rates, lower latencies - with target deployments in 2010s. Expecting standardization work completed by mid-2007.
Expecting all-IP core network in same timeframe as evolved UTRAN.
Goal is "seamless mobility".
Expect mobility management for "all-IP core network" will be used in ITU-T NGN as well.
Expecting netlmm to meet needs of both all-IP core network and NGN.
- Localized Mobility Management and 802.16, Parvez Yehani
Mobile WiMax is IEE 802.16-2004 (amended by 802.16e), targeting higher bandwidth rates, mixed real-time and non-real-time, 200 Km/hour-plus mobility, low power modes.
Mobile WiMax has knowledge of adjacent cells and can do predictive handoff (prepared in advance).
Mobile IP isn't adequate for next-gen multiaccess wide-area networks. Some flavor of localized mobility management is needed.
- Potential WG charter presentation, Phil Roberts
Proposed activities - develop a protocol, develop supporting documentation, demonstrate protocol through simulation and implementation.
v6-only? Start with v6, but could be used for both, must work for v4 traffic.
zero/minimal host configuration.
Focus on tunneling, not on host routes for now.
Inter-Access Router interface? After the basic protocol works.
Control-bearer separation? Not out of scope, but defer until we understand the basic signaling.
Design team? After working group is formed.
- Charter discussion, co-chairs and all
Charlie - why not in MOBOPTS? Performance mentioned many times, but MIPv6 performance enhancements were dropped from the documents and could be added back if that was the right answer. Security management isn't magic - if you have security, how would a MIPv6 solution work?
Hesham - no host involvement? too constraining when we're getting started. Isn't this a product requirement? Analysis very subjective. All of a sudden there are different LMM requirements with no justification for the differences?
John - Interesting problem space. Not sold on analysis and requirements, schedule is way agressive (especially on design team).
Steven - 3GPP work is still in progress until mid-next year. Don't think 3GPP knows that we need LMM solution yet. Still being studied.
Erik - Tunneling-based solution that doesn't involve the host - how does this interact with TRILL? Orthogonal.
Eliot - Schedule has cart before the horse, need requirements work, have a scoping problem that needs to be addressed.
Raj - Strong requirement for no host involvement. Need to quantify why HMIP, FMIP aren't being deployed now.
Henrik - a number of requirements make sense. Worried about developing Yet Another Tunneling Protocol - or is this just another tunnel setup protocol?
Parvez - if client is not involved, it will be effective to have a network-based solution.
Pat - no host stack is what happened with wireless
Hesham - no host involvement? No post IP stack involvement? Several possibilities here. Even with routing protocols there is some host involvement for some link layers.
- Five questions - Phil Roberts
Will use the protocol? 15-20
Will work on the spec? 20-25
Will implement the spec? 10
Believes the working group should be formed roughly according to the charter? 50
- note that MOBOPTS is an IRTF RG, shouldn't be doing standards
- need good scoping (is this really large networks? etc.)
- why not in MIPSHOP, which is rechartering? Want to work on network-based solutions, MIPSHOP is usually host-based.
- Gopal - one protocol for WLAN and 3GPP may not be the right goal, relying on access system and that may be better in specific SDO for that access system.
Believes the working group should not be formed? 30-40
Margaret's question - do we still need to do the work, whether we form netlmm or not? about 20
Thomas - please say on the mailing list what needs to change in the charter