Robust Onboard Timescale Generation for Next-Generation GNSS

Abstract

Next-generation Global Navigation Satellite Systems (GNSSs) are expected to take advantage of inter-satellite links to increase service performance and robustness. Radio-frequency intersatellite links are planned (Galileo 2nd Generation) or already in use (BeiDou). Coherent optical inter-satellite links (OISLs) have been proposed but not yet implemented in GNSSs, while they are in operation for communications in the European Data Relay System (EDRS). OISLs enable time transfer at picosecond level and inter-satellite range measurements at millimeter level. Further, modulating data on the ranging signals allows for the in-space generation of a system timescale via a distributed clock ensemble, and for the autonomous synchronization of satellites clocks. This represents a paradigm shift in the generation of the system time from current GNSSs, which rely on complex ground processing schemes to jointly estimate satellites orbits and clock offsets, to the onboard generation of system time. The local realization can be actively used to steer an onboard time representation, and to autonomously synchronize the satellites at an unprecedented level.

Satellite clocks are subject to failures and outages, therefore the proposed onboard synchronization scheme must be robust against non-nominal scenarios. Onboard algorithms promptly detect malfunctioning locks to minimize the effects of clock faults. On the other hand, the design of the synchronization scheme can provide an inherent level of robustness and fault isolation. In this work, we compare two onboard hardware architectures to achieve the mentioned goal of robust satellite synchronization. In a first version the onboard clocks are directly corrected by autonomously computed clock corrections: the resulting signals are part of the ensemble and thus contribute to the system timescale. In a second scheme, the ensemble uses the free running clocks, with an additional steered oscillator generating the local copy of system time on every satellite. This involves two Kalman filters, one for the computation of system time and one for steering its local realization.

The two schemes are compared using simulations and theoretical analyses to assess synchronization performance (statistical distribution of the differences between satellites clocks) and robustness. In addition, the ability to detect, isolate, and possibly correct faulty clocks is analyzed. The results show how the second scheme can isolate the system timescale generation from faults occurring in the steering chain, at the cost of an increase in hardware and computational complexity.