A sniff scheduling policy for power saving in Bluetooth piconet

After deeply analyzing sniff mode which is a low power operation mode of Bluetooth, a learning function is used to approximate the distribution of the incoming traffic at a master-slave pair. Based the inter-arrival times of data packets obtained from the learning function, the mean of these inter-arrival times is the possible sniff interval; according to the backlog packets in the buffer space and forecast next burst data traffic, a cost model is used to approximate the slot occupancy assigned to a slave. Consequently, calculate sniff attempt slots, and go into sniff mode if conditions are satisfied. Finally, computer simulation results validate that the proposed scheduling policy can save about 38.6% power consumption compared to the always active mode.