[Tinyos-devel] Re: CTP vs LQI

[Omprakash Gnawali]

I just committed a bug fix to 4bitle including an updated alpha of 9.
So, you should use
the CVS head (tinyos repo) when you do any experiment. Make sure your Makefile
points to 4bitle, not le. I use 1pkt/8s but this depends on your network.

- om_p

On Dec 20, 2007 4:48 PM, Rodrigo Fonseca <rfonseca at cs.berkeley.edu> wrote:
> Hi Om,
>
> I hadn't looked at all of your graphs, but Phil and I just went
> through them. I need to run some experiments on Mirage to see what the
> behavior is there.
> Where is the last version of the code you used so we can get comparable results?
>
> Thanks,
> Rodrigo
>
>
>
> On Dec 20, 2007 3:37 PM, Philip Levis <pal at cs.stanford.edu> wrote:
> >
> >
> > On Dec 20, 2007, at 3:22 PM, Omprakash Gnawali wrote:
> >
> > > I added four sets of graph to see if goodcost is roughly the same
> > > over time:
> > > http://enl.usc.edu/~om_p/net2/post-hotnets/figures.html
> > >
> > > The first set is called "cumpktcount(t) with goodcost(pkt)> 20 (ctp)
> > > goodcost(pkt) > 8 (lqi)".
> > > This graph plots the cumulative pkt count with "high" cost over time.
> > > There are three
> > > groups of links - the groups correspond to the experiments with
> > > different durations. The first
> > > six experiments (a1..lqi2) were about 1300 seconds long, the next set
> > > (lqi3..ctpa5) were
> > > about 3700 seconds long, and the last set (aa8..ca8) were about 4050
> > > seconds long. If
> > > the high cost pkts mostly occur in the beginning, we expect the
> > > distribution to start out
> > > steep and flatten later. If these occurrences are uniform, we expect
> > > the distributions to be
> > > linear. Two groups of the distributions are linear suggesting that the
> > > high cost pkts are random
> > > events. The middle group (in the graph) which actually correspond to
> > > the last set of data that
> > > goes to 4050 shows relatively more high cost pkts in the beginning
> > > compared to later
> > > in the experiment: 50% of the high cost pkts in the first 1/4 of
> > > the experiment.
> > >
> > > We can look at the data in more detail using the next three sets of
> > > data that are titled
> > > goodcost(t), cumgoodcost(seq), and  cumgoodcost(t), which are at the
> > > bottom of the page.
> > > These three sets plot goodcost for all (high as well as low cost)
> > > pkts. High cost pkts
> > > are seen jumping out of the mass of colors randomly throughout the
> > > experiment in goodcost(t).
> > > The cumgoodcost(seq) seems mostly linear - which says that there is
> > > no systemic
> > > improvement or degradation of tree quality over time.
> >
> > I don't think cumulative cost is a good way to look at the data: it's
> > hard to see slight slope changes. I'm interested in cumpktcount(t)
> > with goodcost(pkt) > 20 (ctp), goodcost(pkt) > 8 (lqi): if you loop
> > at aa9, then 40% of its high cost packets are in the first 500
> > seconds, and 80% are in the first half of the experiment. If you were
> > to take the derivative of this -- the rate at which high cost packets
> > appear -- then it is decreasing and approaching zero. Of course,
> > chances are it will not reach zero, but...
> >
> > Why are there so many experiments with alpha=8 but not 9? It seems
> > like 9 outperforms significantly.
> >
> > To tease apart the routing efficiency and recovery costs, it might be
> > useful to look at things like 50th percentile costs. It LQI is just
> > dropping the packets that are hard to deliver, then it does have
> > lower goodcost (we need a better name) but as you've noted the wasted
> > transmissions before drops lead its cost to be equivalent. The Mirage
> > results suggest 4b can find better routes in stable networks.
> >
> > Phil
> >
>