On Wed, 18 Oct 2000, Thomas Eklund wrote:
> Hi some commente regarding the movement detection algorithm.
> I agree that it might not be a good idea to tie your movement detection to
> when you are seeing an advertisement from a router with a different prefix.
>
> First of all - the statemachine for the movementdetection can be at two
> places - either in the 1) mobile node (which is the case today) or 2) it can
> sit in the network and then you could have network controlled handoff.
>
> To improve handoff you must make increase mobile IP's link-layer awarness.
> This is usually achived by listening on the field strengh or something
> similar of the radio link. The solution is different for every different
> radio link you have.
[cut]
I don't doubt that there are many ways the link layer could provide
excellent movement detection. I think the issue here is how to provide
some movement detection for MIPv6 without any special signalling from
lower layers.
I think you did touch on an important point - movement detection is really
two problems - (1) detecting movement on to a link, and (2) detecting
movement off of a link. If the mobile node is eager to move to new
networks, then it is only critial that (1) take place quickly for smooth
handoffs. Its somewhat less clear to me when (2) must take place, but it
should happen eventually.
With this in mind, I'd like to suggest this as an idea for movement
detection:
- Require that a home agent's router advertisements always contain all
of the prefixes on the network that are valid for home or CO addresses.
- Then, whenever a MN receives an RA with the H bit set, it will know
with certainty whether that RA came from the current home or foreign
network, or from some new network, based on whether the prefix of the MN's
home or current CO address is advertised in the RA. This way, movement
onto a new link is detected as soon as any home agent on that link sends
an advertisement.
- Whenever an RA is received indicating movement onto a new link,
the MN transitions neigbor cache entry for the old default router(s) to
stale, to accelerate unreachability detection.
Of course, the danger here is that the network could be misconfigured so
that home agents are not advertising all prefixes. This would be
sub-optimal (mobility would be uses when it is not needed), but wouldn't
prevent the mobile node from communicating.
I'm curious what people think of this.
-------------------------------------------------------------------
John Leser UNH InterOperability Lab
IP/Routing Consortium Engineer Morse Hall, 39 College Rd, Room 332
Tel: (603) 862-0090 (8-5 EST) Durham, NH 03824-3525
Fax: (603) 862-1761 Web: www.iol.unh.edu
-------------------------------------------------------------------
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
