Hi Carsten,
I'm supportive of this effort and have been working on similar
improvements to the HC draft based on feedback from the last WG
meeting. Support for different context IDs makes setup and teardown of
context clean, as mentioned in your I.D. RAs are a natural place to
advertise context information. Let me put some thoughts out on the
encoding.
HC would benefit greatly by expanding the encoding to two bytes. For
each address, 2 bits are already used to indicate the compressed
address mode (0-bit, 16-bit, 64-bit, and full 128-bit). We can augment
the encoding by adding another 2 bits for each address to carry the
context ID. This approach is similar to yours, but differs in that it
explicitly separates the context ID space from the address compression
mode. In my mind, the compression prefix and address compression mode
are orthogonal concepts. A single context ID can be used to support
different address compression modes. For example, a single /64 prefix
could be used along with 64-bit, 16-bit, or 0-bit compressed
addresses. We could even reserve context ID 0 to be the link-local
prefix. Using the same encoding for link-local and global addresses
reduces encoding/decoding requirements.
Here's a proposed encoding format that incorporates the context ID
mechanism, as well as a couple other improvements. Specifically, I've
added a bit to HLIM based on past discussions within the WG. I've also
used separate bits for Traffic Class and Flow Label. There's 3 extra
bits, so there's still room to play with it a bit.
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
| T |VF |NH | HLIM | rsv | SC | SCtxt | DC | DCtxt |
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
--
Jonathan Hui
On Jul 7, 2008, at 3:31 PM, Carsten Bormann wrote:
Lowpanners,
I submitted a short draft on a simple header compression scheme.
Filename: draft-bormann-6lowpan-cbhc-00.txt
Title: Context-based Header Compression for 6lowpan
Abstract:
6lowpan (RFC 4944) so far has only defined stateless forms of header
compression. This document specifies how a node and a router can
agree on state that will allow much better header compression of
global addresses.
I think this could combine well with draft-hui-6lowpan-hc-00.txt,
but I haven't really done the integration work.
This draft is still in a very early stage. Looking forward to
comments.
Gruesse, Carsten
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
