hi Sharmistha,

TPSN calcualtes is the instantaneous error between the two nodes i.e.
the error between the two nodes at the time of running the algorithm. This
error can be the result of drift, skew etc. So to answer your first
question, no you do not need to skew synchronize the nodes before running
TPSN.

Coming to your second part, the difference between TPSN and other
algorithms such as FTSP and RATS is that they also allow you to estimate
the clock model, which you can then use to drift, skew etc. This can
potetnially save you from periodically resynchronizing the nodes.

TPSN does not provide you the estimation of skew or drift. So with
TPSN, you would have to run the algorithm periodically at some chosen
period to recalcualte the clock error between the two nodes
continuosly. So to answer your second question, no, TPSN won't take care
of it automatically. You would have to run the algorithm periodically,
although the code in CVS do provide an API for it. You just need to use it
in the proper manner.

bye,
saurabh

On Wed, 15 Mar 2006, Sharmistha Maitra wrote:

>
> Dear All and Saurabh,
>
> Can you please help me with this ?
>
> I want to implement TPSN on two nodes. AS far as I know TPSN assumes that
> the nodes are disciplined, i.e. they do not have any clock skew. TPSN
> only attempts to synchronize the nodes by calculating the offset between
> them.
>
> I want to use the source codes that is there in CVS repository for TPSN.
> My question is, do I have to skew synchronize my nodes (because in real
> life situation there will ofcourse be some skew) before applying this
> algorithm, or the algorithm(as it is available in CVS) has a method of
> dealing with that on its own.
>
> Thanks very much,
>
> Smaitra.
>
>
>
_______________________________________________
Tinyos-help mailing list
[email protected]
https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to