A Dynamic Time Synchronization Scheme for Multihop Sensor Networks

Time synchronization plays a crucial role in many sensor network applications. Existing time synchronization methods proposed in the literature mainly focus on reducing single-hop (pair-wise) uncertainties to improve synchronization accuracy. Based on this, network-wide synchronization is usually achieved via a predetermined spanning tree. In this paper, we employ the Bayesian inference method to analyze the synchronization noises introduced by all upstream nodes in a multi-hop scenario. In addition, we propose a novel network-wide synchronization scheme in which one node can dynamically select the optimal upstream node to synchronize with based on its local decision. The proposed method is distributed, robust, and can adapt to the high dynamics of the network.