One Robust Time Synchronization Based on Time-Average and Maximum-Likelihood Estimation for WSN

Time synchronization provides an essential service for wireless sensor networks. In this paper, we present a new synchronization algorithm which is based on average time synchronization in two-way message exchanges. Firstly, the proposed algorithm does not rely on any reference (or root) nodes, therefore, is totally distributed, asynchronous, robust to node failure or changes of network topology. Furthermore, in order to meet the energy efficiency and convergence requirements, and a diffusion mechanism based on Receiver-Only Synchronization is integrated into it to reduce its energy consumptions and shorten its cycles. Lastly, the protocol allows neighbouring nodes to relatively synchronize with each other using Joint Maximum Likelihood Estimators (JMLEs) for both offsets and skews under Gaussian delays models. Simulation results further demonstrate that our proposed algorithms have advantages in terms of energy consumption, cycle, accuracy, etc.