Time Synchronization in Packet Switching Networks

Abstract

Time synchronization holds an essential role in computer science, particularly in methods that offer precise time alignment down to nanosecond levels across packet-switched networks. Fields such as robotics, advanced communication networks like 5G/6G, and high-fidelity localization systems heavily rely on precise time synchronization. Among the technologies facilitating accurate timestamping, this research primarily focuses on the widely utilized Precision Time Protocol and ultra-wideband technology, both encountering similar challenges regarding synchronization and synthonization. Addressing clock state estimation within packet-switched networks, optimal estimation deals with the inherently stochastic nature of packet delay variation, typically requiring a priori specification.

This study introduces two methods for achieving adaptive estimation by online estimation of the packet delay variation: employing an expectation-maximization approach for ultra-wideband frequency synchronization, ensuring convergence to an optimal solution, and utilizing the Adaptive Kalman filter in IEEE 1588 systems, offering a computationally efficient approximation to the optimal solution. While optimizing estimation yields superior results, utilizing enhanced measurement models can further boost performance. This work shows that leveraging a higher effective data rate in ultra-wideband communication can increase the number of measurements available to the estimator. Lastly, ultra-wideband technology is brought together with the Precision Time Protocol through specific use cases, necessitating an implementation method to achieve synchronization of standard PTP clocks via ultra-wideband communication, including resolving the issue of chip-level clock domain alteration. All proposed solutions in this study underwent evaluation and validation using real and simulated measurements.