Adaptive Kalman Filtering in Offset Estimation for Precision Time Protocol

The synchronization of digital clocks driven by crystal oscillators through packet-based protocols is widely employed across various applications. The IEEE 1588-2019 protocol facilitates the accurate synchronization of follower devices with leader clocks. Nevertheless, the algorithms for clock state estimation face challenges due to the continuous fluctuations in packet delay variations, leading to degradation in the quality of the state estimation. Although Kalman filtering has been introduced for IEEE 1588 to enhance time estimation accuracy, the selection of the measurement noise covariance remains a persistent issue. This article suggests an approach based on adaptive Kalman filtering to estimate measurement noise variance, with a particular focus on maintaining low computational complexity. This aims to establish lower state estimation variance and bias by introducing a novel measurement model with time-invariant measurement noise variance applicable in adaptive Kalman filtering. The proposed method exhibits superior performance compared to state-of-the-art estimation algorithms designed for IEEE 1588 state estimation, as demonstrated through both simulation and real measurements.