Notes of Micro-mobility Routing sub-group at IETF-52.

The meeting was held during the second hour of the IRTF's Routing Research Group meeting at IETF-52.

The team leader, John Loughney (john.loughney@nokia.com), could not be present due to a timetable clash with NSIS WG, so the meeting was chaired by Jari Malinen (jmalinen@iprg.nokia.com) and Carl Williams (carlw@docomolabs-usa.com).

Presentations were made by:

Jari Malinen - who explained the background to the creation of this sub-group. Seamoby WG's initial charter had included micro-mobility, but later it had been decided that work was out of scope of the IETF because host routing solutions were not mature enough. Therefore this is a good topic for research within the IRTF.

Phil Roberts - who outlined the "Local Subnet Mobility Problem Statement" I-D http://www-nrc.nokia.com/sua/irtf-mm-rr/draft-proberts-local-subnet-mobility-problem-02.txt. In contrast to Mobile IP, these solutions would expose end system mobility to the routing infrastructure. 

Jari Malinen - who briefly discussed the "Micromobility Taxonomy" I-D: http://www.ietf.org/internet-drafts/draft-irtf-mm-taxonomy-00.txt. Points of interest were the different types of mobility identifier and various issues associated with aspects such as scalability and architecture. 

Carl Williams - who discussed the problems of mobile IP, and that IRTF WG could be about new IP routing at the 'edge' that would not involve entities deep in the network. He suggested possible deliverables as: a taxonomy document, including problem characterisation and definition; an analysis of the problem space; and an identification of those approaches that have not been explored yet. 

Alper Yegin - who drew a contrast between macromobilty, where there is minimal mobile specific state, and micromobility.

Michael Ramalho - who noted that it is often asserted "IP host routes are not scalable" - but what hidden assumptions does this make? Are host routes OK if micro-mobility is over a 'reasonably small' domain, and what is 'reasonably small'? This was suggested as a good research problem for the group, eg looking at bandwidth and CPU trade-offs as a function of the micromobility domain's size and scope. 

The presenters' slides should be consulted for a fuller account of their presentations.

Comments made by various people were:

What is the relationship of the micromobility to the MANET problem space?
The work should not attempt to duplicate that of mobile IP 
Host routes are feasible - it depends how much customers are willing to pay
How big is 'reasonably small' - the answer is technology dependent
If there is multi-homing in the IP layer, is there any need for host routing?
A fundamental question, and a good one for the group, was posed in the last presentation.