Net2WG/Notes/20080425
Contents
Meeting Notes
Agenda
* Administrative updates * CTP PULL bit
Participants
* Phil, Stanford * Razvan, JHU * Om, USC
Action items
* Om will send emails to tinyos-devel about charter and AM ID allocation * Om will update the website with the new charter * Phil will make a pass through the Collection TEP * Phil will get in touch with Romain about TYMO problems
Discussion Notes
Administrative
Om: Sent to David C. the updated net2 charter. No update.
Phil: I think we should assume that the charter has been approved.
Om: Will send an email with the updated charter and update the website.
Om: Sent an email to David about AM ID allocation as well.
Phil: No word, so lets just reserve.
Om: Will send an email about this as well.
Om: What is the status of Collection TEP?
Phil: Will process some of the comments and make a pass.
Om: Andre, the official maintainer of net2 ZigBee has left the project. I have been talking to Mario from IPP. Might trigger gargabe collection. Will keep you guys posted.
Om: Romain has written two nice documents describing TYMO. I think they should be put on the wiki as a tutorial. The problem is the current collection/dissemination tutorial is called "Networking tutorial".
Phil: We need to change their name to collection/dissemination tutorial.
Phil: We did some experiment with TYMO and turns out it has some problems. I will get in touch with Romain about this.
CTP PULL bit
Om: The problem is the neighbor table gets decimated but the node does not send a pull bit. We are talking about the problem discussed with Bulut on tinyos-help and Phil and I have been trying to understand what is going on.
Phil: CTP is not designed for mobility.
Om: But this can, in theory, happen in static networks too.
Phil: True.
Om: So, we need a policy on when to set the PULL bit.
Phil: When do we do that now?
Om: Looking at the code, we do this only when we don't have a parent.
Phil: Maybe the thing to do is, when the link metric goes beyond 6 ETX, we should set the PULL bit.
Om: Will look in the code in more detail and send an email.
Om: Will also commit the recent changes that reduces the number of beacons.
Misc Notes
Razvan: Why do we have a separate Trickle implementation inside CtpRoutingEngine? I was expecting to see Trickle interface so it was confusing.
Razvan - Should we use default trickle?
Phil - What is your concern? People not understanding implementation? Code reuse?
Om - How about putting a comment that says why CTP uses its own version of Trickle.?
Om - Code duplication is the main concern?
Phil - We should definitely document our implementation of trickle. if the issue is software enginering, the answer right now is not clear.
Razvan - Why not pull it out
Phil - What is the benefit?
Razvan - Makes it easier to understand.
Phil - Lets revisit the software engineering concerns once we put the finishing touches on CTP.
