[manet] Draft update

[Christophe Jelger <jelger@dpt-info.u-strasbg.fr>]

Hi all,

I submitted an updated version of our MANET ADDRESS AUTOCONF draft. It 
is available at the following address :

http://clarinet.u-strasbg.fr/~jelger/draft-jelger-manet-gateway-autoconf-v6-01.txt

We tried to reflect some of the comments we got from MANET members.

Below is a short list of modifications from the previous version. I also 
give some answers to comments we got from MANET members.

Regards
Christophe Jelger

-----------------

Main changes :

- the exchange of GW_INFO messages is not any longer based on a 
DSDV-like protocol
- the LNK 3 bits field has been suppressed
- periodical sending of GW_INFO messages is only done by gateways - a 
node simply forward an updated version of the message it has decided to 
select to build its IPv6 global address --> this method leads to prefix 
continuity
- different algorithms are proposed in order to update the selection of 
the upstream neighbor (the node from which GW_INFO is taken)

Replies to comments of MANET members :

- maintaining prefix continuity is very simple, it is inherent to the 
protocol operation itself
- if a node decides to always choose the closest gateway (the F_DISTANCE 
algorithm), prefix continuity is still valid. The main drawback is that 
a node may have to frequently change its global address
- direct "ad hoc" routing between nodes with different prefixes is 
allowed --> with proactive protocols, there would be a /128 entry for 
each ad hoc node ; for reactive protocols, a path accumulation paradigm 
as proposed in the new AODV draft would allow direct routing
- prefix continuity avoids the need of a routing header
- prefix stability (F_DISTANCE_STABILITY algorithm) does NOT lead to 
more frequent prefix changes : the goal is to minimize these changes !
- with proactive protocols and F_DISTANCE_STABILITY algorithm, routing 
may not always follow shortest-path to gateway (if shortest-path 
includes nodes with different prefixes)
- with proactive protocols and F_DISTANCE algorithm, routing always 
follows a shortest-path which may be different than shortest-path 
computed by the routing protocol





_______________________________________________
manet mailing list
manet@ietf.org
https://www1.ietf.org/mailman/listinfo/manet