CorTiS: Correlation-Based Time Synchronization in Internet of Things

Efficient and accurate time synchronization is critical to various applications in the Internet of Things and has been extensively studied. The emergence of large-scale, duty-cycled sensor networks ruins the previous assumptions used for protocol design, leading to the necessity of developing new synchronization protocols to achieve better energy efficiency and reliability. In this paper, the correlation-based time synchronization protocol (CorTiS), which exploits the spatiotemporal correlation between clocks, is developed. To the best of our knowledge, this is the first effort that exploits spatiotemporal correlation for time synchronization. A theoretical spatiotemporal model is presented to capture the collective behaviors of clocks. With this model, a probabilistic prediction scheme to infer synchronization errors based on correlation is developed without additional message exchanges. The CorTiS protocol is then developed incorporating community detection technique to cluster clocks that are tightly correlated as synchronization communities, such that different communities can be adaptively synchronized. Simulation results show that with CorTiS, participating nodes in each synchronization round can be significantly reduced in dynamic environments.