[Tinyos-devel] Fwd: Request for Comments: TEP 123 - Collection Tree Protocol

[Philip Levis]

On Dec 11, 2008, at 8:41 AM, Omprakash Gnawali wrote:

> Yes, we thought about this scenario. You see a few of these proposals
> in the literature and TinyDB also did something similar - I am not
> sure how well it worked. Our focus is on a practical protocol so we
> could not say if hop-count based duty-cycling is feasible for the
> protocols that can change parents at a given rate. If there is
> evidence otherwise, we are open to putting the hop-count in the
> routing frame.

Let me give a bit more background.

One of the things we wrestled with in this discussion was the fact  
that CTP doesn't promise any stability in the routing tree. The  
current implementation in T2 is highly dynamic; it's quite possible to  
change parents every 5 data packets. This means it's easy for a child  
to have a different view of what a parent's position in the tree is.  
So even if your parent is stable, your grandparent might not be, and  
so it's hard to make strong claims about your "hopcount," especially  
when the implementation is going through transient loops. This is, to  
some degree, why packets have THL: it actually measures how many hops  
a packet took.

Phil