[sim-wg] Meetings

[Philip Levis]

On May 11, 2007, at 10:04 AM, Junzhao Du wrote:

> Hi All,
>
>  Last conference time, I just had a short talk with Philip. I hope to
> build a wiki for simulation working group. Philip has run two working
> groups, core and net2, and hope me to run this one.

I think Junzhao means "has run the wiki for two working groups." :)


>  To make things easy, I hope to use Philip's wiki engine and open
> another page for this working group.

This sounds like a great idea.

>
>  To make some progress, we should list more detail work and others
> can give more comments about how to make it done.
>  I hope to use the following style to list the works:
>
> ----------------------------------------------------
> (1) What's the detail goal?
>       In this part, we should list the detail goal.
>
> (2) What's the state-of-art of tossim?
>       In this part, we should list the current features of tossim and
> why we need improve this or integrate new features.
>
> (3) What's the state-of-art of other simulators?
>      In this part, we can "borrow" some features from other
> simulators, such as from NS-2.
>
> (4)How to achieve this goal?
>      Team members can give comments about how to achieve this.
>      Just like Philip has given some comments for venkatesh's work.
>
> (5)The current working status?
>      List the current status of the works. Will update this according
> to the current work.
>
> (6)Which part need get some helps from others?
>      Maybe need some hands from others due to tight time or short of
> hands. should list there to seek help.
>
> ---------------------------------------------------------------------- 
> -----
> In current stage, many team members are working on different parts, we
> should update the status and get comments from others.
>
> Currently, my group just are porting our works from tinyos 1.1 to  
> tinyos 2.0.
> And after that, I hope to work on the following features:
>
> 1, dynamically change the sending power, so that it will change the
> link quality of neighbors and increase the interference.  how to model
> this?
>
> 2, in most cases the sensor nodes are stationary, but maybe the sink
> nodes will move to collect data from sensors. How to model the moving
> sinks?
>
> 3, maybe on the future, the sensor radio can dynamically switch to
> different channels. maybe neighbors will operate with different
> channels.  How to model this?
>
> This is just the initial ideas, any comments are welcome!


Overall, this sounds like a great idea. It means that we can all know  
what everyone else is doing. Right now, there are basically three  
projects: PowerTOSSIM, hybrid simulation, and wireless models. Having  
a wiki page for this would keep everyone -- including the larger  
community -- informed of progress.

For my part, I've realized that one issue with TOSSIM's radio model  
is that I haven't had a set of simple test cases to determine if it's  
operating as intended. For example, it turns out in 2.0.1 that  
there's a missed edge case in packet collisions, such that the first  
packet received will always be received, even overlapping packets  
lower the SNR below threshold. Also, acks are 100% accurate (no false  
positives and no false negatives), something I need to fix for 2.0.2.  
With the test cases I've started writing, checking these properties  
becomes a lot easier: right now, all that's left are acks.

One question is how we can get broader participation. There are a few  
projects that would be perfect for an enterprising student or  
individual to do which could benefit the community a lot. E.g., a  
chip-level CC2420 simulation, simulation of the MSP430 timer stack,  
and simulation of storage chips.

Phil