[Tinyos-devel] Re: from longer CTP/LQI experiments

[Omprakash Gnawali]

On Dec 24, 2007 1:24 PM, Philip Levis <pal at cs.stanford.edu> wrote:
>
> On Dec 24, 2007, at 10:39 AM, Omprakash Gnawali wrote:
>
> >
> >
> > I plotted the loop events over time. The graphs are called
> > cumloopevents(t) and
> > they are at the bottom of these two pages:
> > http://enl.usc.edu/~om_p/net2/ctp4bitle/mirage-12212007/figures.html
> > http://enl.usc.edu/~om_p/net2/ctp4bitle/figures.html
> >
> > On Tutornet, larger alphas lead to fewer loops but on Mirage a5 has
> > the fewest
> > loops while a9 has the most loops. But again, maybe not enough
> > data. One trend
> > that is clear on both the sets: fewer loop incidents over time. On
> > Tutornet one can
> > just follow the a9 line to appreciate this. On Mirage, each experiment
> > ran for about
> > 7200 s. If the line is truncated at t<7200, that means, there were
> > no more loops
> > after the truncation point: a7, for example, truncates shortly after
> > 1000s, that means,
> > the curve would be flat from approx. 1100 to 7200.
>
> This seems like perfectly reasonable behavior to me.
>
> Phil

The "bad" metrics tend to quiet down over time - that's a good thing.
We can see this happen on Mirage within an hour or two but have to
wait for several hours on Tutornet.

I still don't understand why we are seeing a relatively small max
degree on Tutornet, first, compard to Mirage, second, compared to the
expectation that the degree should be unconstrained.

- om_p