[Tinyos-devel] timestamps in seriallisten/sflisten etc.

[Eric Decker]

Seems to me there should be a service at the SF juncture that would providetime
stamping.  The SF is the interface between the wireless sensor cloud and
the internet.  It is also the first place where you can count on
significantly better
resources.

So it seems to be that added a time stamp facility, format definition, etc.
to the
sf protocol would be a useful thing.

eric


On Wed, Jun 25, 2008 at 6:28 AM, Mustafa Yuecel <yuecel at tik.ee.ethz.ch>
wrote:

> Timestamping on the listener side is not always sufficient. We are using
> here a Linux embedded system with outside communication links like GSM
> or satellite phone. There we can expect high delays between SF and
> Listener, especially if we plan to transmit a block of packets. About
> time synchronization I am not afraid. We can use a NTP services like
> chrony or ntpd with a precision under a few milliseconds.
>
> I would like to have an option for the SerialForwarder to turn on the
> timestamping and the listener should automatically handle the timestamp...
>
> I also send the patches applied to our system (against tinyos-1.x CVS
> HEAD):
>
> - filewriter-and-timestamping.patch: contains two changes, first logs
> all packets to a file (maybe a candidate for the unimplemented -log
> option), second does timestamping and appends the unixtime (8 bytes, LSB
> first) to the packet payload
>
> - proper-sf-shutdown.patch: if using the console, the SF will not close
> cleanly the local server socket with Control-C, which makes restarting
> buggy
>
> - remove-env.patch: since System.getenv() method is no more deprecated
> in Java SE 5.0, we can remove the native java class net.tinyos.util.Env
>
> We also adapted the Message classes by hand to the appended timestamp.
> This is surely not the best solution, but short-termed we had no other
> choice, as long as the TinyOS frontend will not support timestamping on
> the host side...
>
> just my 2 cents,
> Musti
>
>
> Philip Levis wrote:
> > On Jun 24, 2008, at 5:57 AM, Andreas Koepke wrote:
> >
> >> Good idea.
> >>
> >> Can this be communicated "down the chain"? (sf -t --> prettylisten) We
> >> could use the version handshake for this, and we should agree on
> >> endianess.
> >>
> >> Best, Andreas
> >
> > I think that the timestamp should be on the output printer, not
> > intermediate points. Otherwise, you could get into a situation where
> > serial forwarders running on different machines apply timestamps based
> > on their local clocks, and prettylisten windows then report an
> > ordering different from what actually arrived at the pretty printer.
> > Changing the printer applications sounds like a great idea, changing
> > the serial forwarder protocol doesn't.
> >
> > Phil
> > _______________________________________________
> > Tinyos-devel mailing list
> > Tinyos-devel at millennium.berkeley.edu
> >
> https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-devel
>
>
> _______________________________________________
> Tinyos-devel mailing list
> Tinyos-devel at millennium.berkeley.edu
> https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-devel
>
>


-- 
Eric B. Decker
Senior (over 50 :-) Researcher
Autonomous Systems Lab
Jack Baskin School of Engineering
UCSC
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://www.millennium.berkeley.edu/pipermail/tinyos-devel/attachments/20080625/98f796b8/attachment.htm