[Iccrg] A Wireless Channel Model Based Rate Control (WMRC) Scheme (WAS Re: present a draft)

[Lachlan Andrew]

Greetings Dong,

Thanks for the update.  My concern wasn't that the draft lacks
implementation details.  It is that I think this type of result is
better published in a journal or presented at a conference.  It is a
solution that works well for one particular model of a wireless
channel, but wireless channels are very hard to model accurately.

If you want this protocol standardized by the IETF, I would recommend
publishing several papers evaluating it under a wide range of wireless
settings (WiFi, 3G, LTE, WiMax, satellite, Zigbee) and showing that
the algorithm is robust.  Then summarize the results in a
recommendation to the IETF.

Cheers,
Lachlan

On 22 November 2010 21:27:11 UTC+11, Y Dong <dongyn at njupt.edu.cn> wrote:
>
> It seems the main concern is its being “too theoretical”. So we added a
> section in the modified draft (see attachment): Section 6 Packet Processing
> Protocol to give more details of the sender and receiver behavior for
> simulation /experimental test.
>
> As for the "fairness" issue, we have tested the inter-flow fairness of the
> proposed algorithm by simulation. And we are trying to test its performance
> in real networks with wireless access, e.g. The IEEE 802.11 based Wlan or 3G
> networks.
>
> We would be grateful to all especially Ingemar and Lachlan for any further
> comments.
>
> ---------- Forwarded message ----------
> From: Lachlan Andrew <lachlan.andrew at gmail.com>
>
> Although I'm a theorist at heart, on a cursory glance I also feel that
> this may be too theoretical for an IETF standard.  There are many
> possible channel models (for many possible channels), and
> standardizing transmission based on one should be done with caution.
>
> IETF documents are mainly useful for issues that need agreement to be
> implemented.  The main reason congestion control algorithms (as
> distinct from protocols) need agreement is for "fairness".  Fairness
> in systems with different channels is a very complicated issue, which
> doesn't seem to be addressed in this draft.
>
> $0.02,
> Lachlan


-- 
Lachlan Andrew  Centre for Advanced Internet Architectures (CAIA)
Swinburne University of Technology, Melbourne, Australia
<http://caia.swin.edu.au/cv/landrew>
Ph +61 3 9214 4837