currently olsrd is pretty chatty and especially in large network clouds a lot of TC messages need to get processed. The following is a incomplete list of all the doftware change that we plan to make in order to get olsrd wake up less frequently and hence save more power.
This is mainly meant as a way to correct the deficits (read routing loops) from fisheye. The basic idea is that only significant changes get distributed throughout the network. Total outcome is that the airtime consumes by the routing protocol gets reduced to the bare minimum. Each node itself can control its update threshold,
- Complete Sequence Number (CSN) Protocol Extensions
This is a cheap way of syncing the link-state databases, in order to avoid routing loops. The idea was inspired from IS-IS another fixed-line line-state protocol, that aims to simplify link-state database synchronization.
Today olsrd wakes up every 50ms in order to walk the timer wheel, and check if any timer is due. One could sort the timers by expiration date and sleep as long as the first timer one the time line does expire.
