I have just a question . It is written in the tep 123 that  "Together, the
origin, seqno and collect_id fields denote a unique **origin packet.**" .
However if we want to analyse the CollectionDebugMessage , it is seen that
after every 255 messages sent or received , the sequence number restarts. So
e.g If I want to analyse a node 1 , if it sends data more than 255 packets
then , it is bound to have two or more messages with same origin and
sequence number. So , analysis of the CollectionDebugMessage is bit out of
hand here for me. It is such that , the msg_origin in the CtpDebugMsg.h is a
16 bit , while the originseqno of ctp data frame which writes to it is 8 bit
. If I make the ctp data frame to 16 bit as well then I can analyse the
CollectionDebugMessage . I was wondering if I change the bit of  the data
frame affects the performance of ctp.
Another question is , if I am to avoid that , how can I go about analysing
the CollectionDebugMessage?

thanks ,

Nicole

On Mon, Nov 22, 2010 at 4:11 AM, Omprakash Gnawali
<[email protected]>wrote:

> On Sun, Nov 21, 2010 at 8:03 AM, Nicole Chiesi <[email protected]>
> wrote:
> > hi,
> > I am trying to do the similar experiment as done as mentioned here
> > http://sing.stanford.edu/gnawali/ctp/ctp-sensys-data.html . Basically I
> want
> > to get and analyse the figures given in index.html file of the
> experiment.
> >
> >
> > It feels that the experiment data given there is just some excerpts of
> the
> > experiment .
> > How are these calculations made?
> >
> > #node total_sent uniq_rcv success_rate total_rcv repeated_rcv (frac)
> > repeated_rcvthl (frac) minseq maxseq
> >
> >
> > from the experimental data given there.
> >
> > I was able to set up the experiment and run the experiment but I have no
> > idea how I can come to these results.
>
> First you need to get the log files from your testbed experiments. Did
> you get the logs? The assumption is you have one log file per node.
>
> Then you need to run a series of Perl scripts on the logs. You need a
> few custom Perl libraries and neato to produce the result you see.
> Because of these dependencies, I haven't posted all those Perl scripts
> because it most likely won't work on other systems. But here is how
> you can proceed:
>
> 1. Run the parse-ctplogs.pl script on the log files.
> 2. Do statistics on the output from step 1.
>
> For example, if you want to compute delivery ratio for a specific
> node, you can keep track of sequence numbers received from that node
> and compute the delivery ratio. If you want to compute churn, you can
> count the number of times beacons was sent. All these log messages are
> parsed with parse-ctplogs.pl.
>
> Feel free to ask questions as you get going.
>
> - om_p
>
_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to