[Tinyos-devel] [Tinyos-help] Collection: Loosing Connection to a Node

[Bulut ERSAVAS]

Hi all,

Roman's workaround seems to work pretty well. I've been successfully
running our application with his code since he mentioned it.

Omprakash,

I'll be away from office next week. I'll send you the logs of our
network with the latest CVS in the following week.

Thanks,
Bulut


On Mon, May 5, 2008 at 11:30 AM, Roman Lim <rlim at ee.ethz.ch> wrote:
>
> Philip Levis wrote:
>
> >
> > On Apr 29, 2008, at 3:39 AM, Bulut ERSAVAS wrote:
> >
> > > Hi David & Roman,
> > >
> > > Thanks for your comments. I would be really interested in trying Roman's
> workaround. Roman, would you mind sharing it with us?
> > >
> > > Phil & Omprakash,
> > >
> > > Considering these comments, what kind of action plan do you have in mind
> to make LPL & CTP work together? Will net2 wg own the case? We (a team of 3
> EE & CS at Bosphorus University, Istanbul) are ready to provide support in
> testing and any other aspect you see appropriate.
> > >
> > > Looking forward to hearing from you.
> > >
> >
> > The CC2420 stack lives in the core WG, while CTP, etc., live in net2.
> >
> > The best thing you can do is generate CTP traces using LPL that let us see
> what happens. We can also generate traces for testbeds we know well, but the
> more diverse the set of environments that we test protocols in, the more
> problematic edge cases we can find and deal with. The important thing is to
> be sure to use the most current code in CVS; Om just checked in the changes
> that reduce the control traffic rate.
> >
> > Roman, I'd love to hear about what changes you made that improved LPL on
> the CC2420.
> >
> > Phil
> >
>
>  the improvement is, that nodes start sending their lpl packet streams just
> at the time, when the destination node wakes up. in order to achieve this,
> nodes add their wakeup pattern information to every packet they send (lpl
> interval and wakeup offset). nodes that do not have timing information for
> their packet destination, send normal lpl packet streams. this reduces the
> time that a packet occupies the channel, and therefore also tho possibility
> of collisions.
>
>  Roman
>
>