JP and WG members,
This may be considered link reliability.
But in our experience, we found the following information very helpful
in improving the stability of network and also (and you could say) in
avoiding loops as well.
We use what we call 'W' (weight).
When we forward data packet, we require per-hop ACK.
When we receive ACK, we adjust the new W to be old W -1 (W_new =
W_old - 1).
When we don't receive ACK, we adjust the new W to be old W + 1
(W_old =
W_old +1).
When we detect a loop, we adjust the W value to MAX.
We have experimented MAX value based on real system, MAX value of 10
seems to be a good value to start with.
Now this can be combined with other metric values you are all
discussing. We have experimented quite a lot with combing several
metrics with the above weight in determining the path, which so far
seems to work in our system.
For instance, the other metric values you are discussing (such as link
quality, RSSI, and/or everything else - please note this can be as
simple as hop count, RSSI, or whatever the WG determines) can be used
with along with W value to determine.
I am interested in what the WG thinks of using this 'W' as part of
metric for RPL.
