Speaker
Description
REFIMEVE is a French national research infrastructure, officially recognized in 2021, based
on a fiber-optic network that disseminates ultra-stable time and frequency signals over long
distances. Reference signals originate at LTE (former SYRTE) at Paris Observatory and are
primarily transported over the optical fibre network backbone of RENATER [1]. It currently
connects more than 20 research laboratories across metropolitan France, with a goal to reach over 40 in the next five years. The network also extends internationally through four cross-border links.
Within the T-REFIMEVE project, we aim to establish a national-scale White Rabbit (WR)
timing network, targeting time accuracy better than 10 nanoseconds across more than 80 nodes interconnected by over 5,000 km of optical fiber. Unlike the bidirectional optical carrier used in the REFIMEVE infrastructure, this deployment leverages the unidirectional optical
amplifiers of the RENATER backbone. WR signals are transmitted as alien wavelengths using standard xWDM technology, which introduces the challenge of mitigating optical path
asymmetries inherent in unidirectional links.
A critical first step toward achieving this level of timing precision is the precise calibration of
instrumental delays in the White Rabbit Switches (WRS). This work reports on the calibration of instrumental delays of dozens of WRS units.
We performed the calibration according to the CERN procedure [2] and the EMPIR/VSL good practice guide [3], for unidirectional links, using SFP transceivers operating at a wavelength of 1560.61 nm. We defined port 1 of one of our WRS to be our golden calibrator, while choosing to leave its ingress and egress delays at their default firmware values, and then used it to calibrate port 1 of another WRS, thus constituting the usual pair of reference ports in a slightly unusual way. We then used these two ports to calibrate all other WRS ports. As expected the synchronisation was already good using the default delays, with offsets of up to approximately 300 ps, and improved to within ±20 ps after calibration. (All of the presented measurements used the same slave port, which had been previously calibrated to the golden calibrator.)
We review the sources of uncertainty in our calibration process and present a preliminary
uncertainty budget.
Finally, we briefly report on the in-field implementation of White Rabbit within REFIMEVE
and discuss the estimation of asymmetric delays introduced by various components of the
active network-such as amplifiers, attenuators, multiplexers, and fiber length differences along the transmission and reception paths of the WR link.
References:
[1] O. Lopez et al., « Frequency and time transfer for metrology and beyond using
telecommunication network fibres », Comptes Rendus Physique, vol. 16, no 5, p. 531-539, juin
2015, doi: 10.1016/j.crhy.2015.04.005.
[2] G. Daniluk, “White Rabbit calibration procedure,” 2015 https://white-
rabbit.web.cern.ch/documents/WR_Calibration-v1.1-20151109.pdf
[3] Dierickx, Erik and X. Yan, “WR Good practice guide,” May 2019
https://gitlab.com/ohwr/project/white-
rabbit/wikis/uploads/7df19b6a4d0e90bf6d7b8ae32b3b32c4/WR_Good_Practice_Guide.pdf
