Clock Synchronization Using Fast Matrix Completion-Based Maximum Likelihood Estimation in TSN

Abstract

In time-sensitive networking (TSN), precise clock synchronization is an important guarantee for the normal operation of the network. The collection of timestamps plays a crucial role in common centralized clock synchronization, directly affecting synchronization accuracy. However, the exchange of timestamps is susceptible to the impact of packet loss. To address this issue, this letter proposes a matrix completion scheme based on random matrix approximation to recover the lost timestamp information. First, we formulate the clock synchronization problem based on incomplete timestamp information as a low-rank matrix completion problem. By utilizing the principle of random matrix approximation, we can efficiently recover complete timestamp information using partial timestamp information. Furthermore, the recovered timestamp information is then used to estimate the clock skew and clock offset with a maximum likelihood estimator (MLE). Numerical results confirm the effectiveness and accuracy of the proposed approach.