Adaptive Time Synchronization for Wireless Sensor Networks with Self-Calibration

Time synchronization is important for wireless sensor networks because it facilitates cooperation among nodes and helps raise power efficiency. Time synchronization protocols like TPSN, RBS and FTSP have provided great schemes to fulfill fast synchronization with efficiency. In some applications, nodes might hope to sleep for a long time without timestamp exchanges with other nodes. In that case, accurate time drift prediction is quite necessary. For that purpose, firstly, we propose a time synchronization scheme, which fully utilizes the broadcast nature. The scheme achieves time synchronization with fewer timestamps compared with RBS and TPSN. Secondly, we introduce a method to find relative time drift rate on the fly. Thirdly, we introduce a scheme to predict time drift rates of the next few hours. We also analyze a few factors that deteriorate frequency drift or time drift rate. The diurnal periodical environment trend, instead of mathematical extrapolation, is used for time drift rates prediction of the next few hours.