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

[Philip Levis]

On Apr 6, 2008, at 5:56 AM, Bulut ERSAVAS wrote:
> Hi Omprakash,
>
> I can't really say that it is quite deterministic. In networks where  
> we collect measurements every 15 minutes, we lost path to some nodes  
> within a few hours to half a day. Our networks are generally small  
> (i.e. 3 to 5 nodes per network). There are cases where we lost the  
> connection in a single-hop-to-root setting. As we are experimenting,  
> the nodes are not moved around without our knowledge.
>
> This issue happens specifically at two locations. One is a  
> greenhouse with barely any obstacles. The other is a closed  
> environment where there are reflective metal walls. For the latter  
> case, the motes are placed very close to each other (about 10-15  
> meters).
>
> Would rebooting the motes periodically (e.g. every few hours) help  
> ease this problem? Is it possible for CTP to get stuck in a state  
> where it can no longer discover a path to root?

As long as there is a path to the root, CTP should be able to find it.  
That being said, there could be a bug in CTP or a bug in your  
application. Given how heavily CTP is tested, I'm leery of assuming  
it's a bug there.

Stationary nodes does not mean the topology doesn't change. You can  
easily see significant signal strength shifts due to temperature and  
humidity. One study on ATPC showed RSSI vary by as up to 10dB for  
nodes sitting in on a lawn, over 24 hours.

Phil