The synchronization of signals in a Global Navigation Satellite System (GNSS) is an essential part of the system architecture. It is typically achieved by using long-term stable clocks on all satellites and estimating their offsets jointly with a large number of other parameters. The Kepler system uses a different approach. Each satellite carries a cavity-stabilized laser, which is characterized by a very high short-time frequency stability of about 10-15 [s/s] (Allan deviation) over averaging times time intervals up to 10 seconds. References on different satellites are synchronized by time- and frequency-transfer techniques via two-way laser links. The latter are at least one order of magnitude more stable, so that the time stability is dominated by the performance of the cavities.
The cavity-stabilized lasers and the laser links are the basis for achieving near-perfect synchronization across the whole constellation. However, for time distribution purposes the timing subsystem needs to be complemented with a few clocks, e.g. optical clocks based on iodine-stabilized lasers and/or active hydrogen masers or Cs atomic clocks.
One of the approaches investigated for the generation of a Kepler system time is to use a composite clock processing for synchronizing all references in the constellation. A composite clock is based on the time offset of each satellite’s reference relayed via the optical links, which are then combined to compute a stable overall time reference, called the Implicit Ensemble Mean (IEM). Each reference in the system may then be referenced to this IEM. In order to fully exploit the stability of the cavity-stabilized lasers, readings are performed every second (with 10 seconds the target upper bound). The estimated corrections for realizing the IEM are then applied to the oscillators timing the navigation signal generation, but not to the cavity-stabilized lasers themselves. The navigation messages are thus broadcast with near-perfect synchronization.
The generation of an accurate IEM requires accurate models for predicting the behavior of the reference employed: this is one of the aspects investigated at present stage. Different clock behaviors are modeled by using two-state clock models, which are extended by Markov processes in order to take into account the non-classical behavior of cavity- and iodine-stabilized references.
An example of simulated IEM generated is shown in the figure: 30 cavity-stabilized lasers (24 on MEO satellites and 6 on LEO satellites), 6 Iodine clocks (one on each LEO satellite) and one Active Hydrogen Maser co-located with one ground station are used to obtain an IEM showing a fractional stability below 6 · 10-16 [s/s] (Allan deviation) over averaging times from 0.1 s to 100000s.
All current GNSSs also support time distribution to Earth users. This capability is offered within the Kepler system by monitoring the evolution of the generated system time with respect to some long-term stable time reference on ground. A single-ground station is sufficient to align a terrestrial time scale (e.g. UTC) with the Kepler system time.
The designed time-generation algorithms are being tested on a set of laboratory clocks, based on optical systems, developed by the DLR Institute of Space System, and classical systems, both operated by the Time Lab at the DLR Institute of Communications and Navigation. These include a cavity-stabilized laser, an optical iodine clock, active hydrogen masers and high performance cesium standards.