David,

Since motes A and B that receive the RadioCountToLeds message with counter
value i at the same time, ideally, both A and B should return the same
global timestamps of the message reception. You can, for every counter value
i, take the difference of global timestamps by A and by B. You can use
simple statistics (mean, variance) of these differences for a series of
counter values to characterize the time synchronization error between motes
A and B.

Janos

On Mon, Nov 3, 2008 at 11:59 AM, Yi-Tao Wang <[EMAIL PROTECTED]> wrote:

> Thanks for your help Janos.
>
> If I want the clock error between the root and another mote, would
> that be the offset? Also, what's is are the timestamps in
> milliseconds?
>
> David
>
> On Mon, Nov 3, 2008 at 8:22 AM, Janos Sallai <[EMAIL PROTECTED]>
> wrote:
> > David,
> >
> > The local timestamp is the time of reception of a RadioCounToLeds message
> > according to the motes local clock, whereas the global timestamp is the
> time
> > of reception of a RadioCounToLeds message according to the global time in
> > the network. When a mote is synchronized, they are typically different,
> so
> > what you're seeing is correct.
> >
> > Regarding the skew being zero: FTSP maintains two local state variables,
> > skew and offset, which are used to translate local time to global time in
> > the following way:
> >
> > global_time = (1 + skew) * local_time + offset;
> >
> > The test app print out skew*1000000, which, per the above definition,
> should
> > be pretty close to zero.The 32kHz crystal on the mote has a +-50ppm skew
> > within the whole operating temperature range. If all the motes are
> operating
> > at the same temperature, the skew should be much less than that. I'm
> seeing
> > mostly zeros, and -1 here and there. As long as the global times for the
> > same counter value agree when the network is synchronized, seeing zeros
> > should not be a problem.
> >
> > Please note that, according to the readme file in tests/TestFtsp, a mote
> > flag is synced if the synced flag is zero.The skew value is only
> meaningful
> > when the mote is synced (i.e. the synced flag is zero).
> >
> > Janos
> >
> > On Mon, Nov 3, 2008 at 9:55 AM, Yi-Tao Wang <[EMAIL PROTECTED]> wrote:
> >>
> >> Hi,
> >>
> >> I'm having some trouble interpreting the results of the testftsp
> >> application on tinyos 2.x. The skew value is always 0 even when the
> >> synched flag is false and global timestamp doesn't equal local
> >> timestamp. Does anyone know why?
> >>
> >> Thanks,
> >>
> >> David
> >> _______________________________________________
> >> Tinyos-help mailing list
> >> [email protected]
> >>
> https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
> >
> >
>
_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to