Time synchronization accuracy in real-time wireless sensor networks

Wireless sensor networks are generally used to sense and capture information about physical events, from which correlation patterns are then extracted collaboratively. To this end, the sensors must be time synchronized to project the relative chronological order of occurrences in an event. Such time synchronization is also required for energy efficient radio scheduling and distributed coordination within the network. This paper determines the factors affecting the accuracy of time synchronization under a broadcast flooding protocol. The sources of delays and uncertainties in the sender-receiver path are examined and minimized with add-on MAC layer timestamps. The side-effect of periodic synchronization beacons on energy consumption is compensated by clock drift estimation. Also, the effective regression data set for different synchronization periods is determined. The complete framework, named ¿-Synch, is implemented in a multi-hop network on Sensinode Nano motes which employ CC2431 transceiver and FreeRTOS operating system. The ¿-Synch achieves synchronization accuracy higher than that of originally proposed flooding time synchronization protocol (FTSP). At the end, a novel scheme for collision avoidance of broadcast synchronization beacons is proposed.