I was trying to figure out the time synchronisation mechanism using beacons and the standard seems to indicate that one needs to maintain a time synchronisation of about 4 micro-secs with respect to the beacon.

I thought that such kind of accuracy was achievable only at the hardware layer however while I was looking through the madwifi code and it seems to me that this is being done by the driver in net80211/ieee80211_input.

I have a couple of questions and would be very obliged if someone could point me in the rigth direction / point out that my assumptions may are wrong...

1. How is one able to chieve such fine time synchronisation at the driver level?

2. If I insert printk statements before and after the time stamp is being copied from a received beacon, will I be able to work out for a pair of node by how much the clocks are going out of sync?

3.  If I can repeat this process over more than one hop is it technically possible to  synchrnise a  node  two hops down with a node  upstream by minor changes to the madwif code?

Thanks and Regards


"When we long for life without difficulties, remind us that oaks grow strong in contrary winds and diamonds are made under pressure"
-- -- -- Peter Marshall