[sim-wg] Application for membership

[Chad Metcalf]

Tal Rusak is applying for membership in the wg. Below is his
application. Please respond to me by Wednesday of next week with a -1,
0, or +1.

Tal has already submitted TOSSIM fixes and is working with Phil on
RSSI modeling for the 2.1 release.

Thank you,
Chad

---------- Forwarded message ----------
From: Tal  Rusak <tr76 at cornell.edu>
Date: Sep 21, 2007 9:09 AM
Subject: Re: TOSSIM and the 2.1 Release
To: Chad Metcalf <metcalfc at gmail.com>
Cc: simwg <sim-wg at millennium.berkeley.edu>, Tal Rusak
<tr76 at cornell.edu>, "Razvan Musaloiu-E." <razvanm at cs.jhu.edu>,
cliang4 at cs.jhu.edu, pal at cs.stanford.edu


     Note: This message will probably bounce the simwg mailing list. Can
someone on the list please forward it to the list?

Hi,
     There have been some questions regarding what I plan to do to improve
TOSSIM. Currently, I am looking to improve the simulation of packet
reception by making the physical model that underlies the simulation more
accurate.
     Recall that the physical model is based on an SNR calculation as
described in "Improving Wireless Simulation Through Noise Modeling" by
HyungJune Lee, Alberto Cerpa, and Philip Levis. The TOSSIM implementation
of this model had/has several issues, most notably:

1) Quantization of noise values by 5, required to make state space of
TOSSIM reasonable on personal computers, was detrimental to effective
simulation of SNR when the reading was in the steep portion of the curve.

2) There is no long term or short term RSSI model in TOSSIM; gain is
assumed to be constant. I am looking to implement a physical model for
RSSI as the "Improving" paper did for noise, taking into account long term
as well as short term variations.

3) The value of K = NOISE_HISTORY (in sim_noise.h) is noise-trace and
quantization dependent, according to my experiments with TOSSIM. If set to
the non-optimal value, it will greatly reduce the effectiveness of the
simulation. An automatic optimization of K would make the TOSSIM packet
reception model much more useful in general.

     My solution to (1) has been checked into CVS; anyone using the new
version of sim_noise.c and sim_noise.h is encouraged to send me
feedback/bug report, etc. Please note that the optimal value of K is
different than it was before this change was made; for example, for the
meyer-heavy trace, an optimal value of K is 30 under the new
implementation. I am currently working on (2).

     I am interested in joining the Simulation Working Group; hopefully
the information provided is sufficient. If anyone has any more
questions, please  feel free to reply to this message.

Thanks,
Tal


> So things have been quiet and we all know that the 2.1 release is going
to sneak up faster then one might like. So given our wide
> dispersion across the time zones we should take a first cut by email.
>
> I'd like to see the wg get a handful of things into TOSSIM for the 2.1
release. Below is a list I've seen discussed and thing can reasonably be
done before the end of the year and who I think "might be the right person
for the job".
>
> SF for TOSSIM - I've got a version of this working in my directory in
contrib. The question is, is the simple implementation enough, did I
choose the right way of doing things, what else is needed, etc.  I've seen
requests for this on the help list so I think its inclusion would be
helpful. I'll start a separate thread for this later.
>
> mica2 support - IE make mica2 sim. As discussed in prior threads. This
work lives in Venkatesh's work.
>
> Improved RSSI model - Stemming from the latest round of updates.
Currently no one in particular is slated for this work. Tal Rusak, has
considered joining, this might be something he's interested in.
>
> If you have other ideas please bring them up. With the understanding of
course that this is for a 2.1 release, meaning the idea needs to easily
executed in the time remaining. For "large" ideas this means a large part
of the code must already exist because it has to be tested and ready for
the release.
>
> Cheers
> Chad
> --
> Chad @ Home
>








-- 
Chad @ Home