Predicting White Rabbit Time Synchronization Error on DC-QNet Using Statistical Machine Learning Methods

Quantum networks leverage the laws of quantum mechanics to transport quantum states over geographic distances by using the quantum teleportation and entanglement swapping protocols. These protocols necessitate successful Bell state measurements which require the distribution of high precision timing between the participating nodes. White Rabbit (WR) is a precision timing protocol developed at the European Organization for Nuclear Research (CERN) which provides the means to obtain sub-nanosecond time resolution between nodes in a terrestrial quantum optical network using optical Ethernet switches. The purpose of this research is to characterize the error in distributed high precision timing when using the White Rabbit protocol wave division multiplexed in single-mode optical fiber as a function of time of day, length of fiber, percentage of aerial fiber, and environmental conditions (e.g., temperature, humidity, barometric pressure, wind speed, cloud cover). We evaluate and compare several statistical machine learning models in terms of their ability to accurately predict distributed timing error as a function of several geographic and environmental variables. The long-term goal of this research effort is to support decision making when implementing a quantum network using existing low-cost conventional optical telecommunications infrastructure technologies.