Robust Estimation of the Offset Between UTC and SBAS Network Time

The main purpose of the Satellite Based Augmentation Systems (SBAS), such as EGNOS or WAAS, is to provide the civil aviation user community with reliable navigation services for different flight phases. To achieve its missions, any SBAS implementation intrinsically defines its own time scale with respect to which the core GNSS satellites clocks are corrected; this scale is named SBAS Network Time (SNT). When the users determine their position with the corrections broadcast by the SBAS GEO satellites, they obtain both their location and the synchronization bias of their receiver clocks with respect to the SNT. The information broadcast by an SBAS can include the offset between UTC and SNT. Consequently, the application of this offset to the synchronization bias estimated for the receiver clock can be used to further synchronize it to UTC. With this process, any inexpensive GNSS receiver having the SBAS capability can be used as a source of UTC time with an accuracy of a few tens of nanoseconds. Currently, EGNOS broadcast such information by referring the realization of the UTC located at Observatory of Paris (UTC(OP)). This broadcasted information is continuously observed and their performance evaluated thanks to the CNES Navigation and Time Monitoring Facility (NTMF). Performances are quite good (on the long term ENT – UTC behaves very similarly to ENT – UTC(OP)), unfortunately some degradations can be observed in the offset extrapolation area when the data from UTC(OP) are not available at EGNOS Central Processing Facilities (CPF) input (e.g. upon EGNOS RIMS failure at UTC(OP) or communication network outage). This study has taken place in the frame of the experimentation test plan called “TENOR” and based on the SPEED Test Bed (Support Platform for EGNOS Evolution and Demonstrations). The SPEED Test Bed has been developed by ESA and CNES with Thales Alenia Space France as prime contractor, to support EGNOS evolutions, and to test new navigation concepts. This platform is fully representative of EGNOS (algorithms are identical) and is easily configurable at parameters level and modifiable at modules level, providing huge capabilities for engineering, prototyping and validation activities. Now, CNES has contracted GMV, in charge in the EGNOS Industrial consortium of the EGNOS CPF processing set algorithms and developments to analyze NTMF observations on the extrapolation performance of the offset between UTC and SNT as provided by EGNOS and to consider new algorithms that could be implemented to further improve the performance of this offset estimation including its extrapolation. Therefore, the objectives of the new algorithms are to enhance the extrapolation of the offset when the system is deprived of the UTC input data that comes from the specific reference station located at the Paris Observatory for long periods of time. This is possible because of the excellent stability of UTC(OP). Furthermore, the authors have investigated the possibility to back the synchronization of UTC to SNT with cesium clocks available to the SBAS system. In this way, it is not only possible to propagate the UTC-SNT offset but also to follow the behavior of the SNT time scale that could be originated, for instance, on particular events in CPF selection that serves a particular GEO PRN. The algorithms proposed are based on two-state multi-clock Kalman filters with specific barriers designed to detect instabilities in the different clocks considered, particularly SNT or any of the cesium clocks used to back UTC(OP) upon station data gaps. In order to isolate the potential irregularities of the SNT scale as much as possible, the states considered are the offsets of UTC to SNT, and the cesium clocks to UTC. The filter is updated with measurements of the UTC and the cesium clocks with respect to SNT. This selection of filter states and measurements is optimal for the sought purpose: on the one hand, the filter internal states include the offset of cesium clocks to UTC, which should not be affected by irregularities in the SNT scale. On the other hand, the measurements chosen allow for the update of the filter even when UTC(OP) data is not available. One of the key ingredients of the new algorithms refers to the increment of the crossed components of the state covariance matrix upon detection of irregularities in the clocks. This ensures an appropriate filter response and a seamless transition whenever cesium clocks are rejected or reintroduced, a SNT jump is detected, or when UTC(OP) measurement are available again after a data gap. The design of the algorithms has been done considering the feasibility of their implementation as an incremental evolution in the operational EGNOS. For that, maximum reuse of already available internal data is proposed, the need of RAM and CPU is minimized, and the more CPU demanding modules are introduced outside the critical real time chain of the current EGNOS central processing facility and their results are then used to feed the content of the message dedicated to broadcast the offset (MT#12). The new algorithms have been calibrated with two weeks of real data using a platform that is representative of the current behavior of EGNOS, namely magicSBAS [9] with simulation of Paris reference station data gaps and SNT behavior and will then be implemented into SPEED [10] for a demonstration in real time. The experimentation include configuration of the algorithms with and without backing cesium clocks. The preliminary results achieved show that, in nominal conditions, it is possible to propagate the UTC-SNT offset that would be broadcast by a SBAS with the new algorithms for periods of several days with accuracies in the range of 10 nanoseconds. The new algorithms are also capable of following SNT behavior even in periods of absence of Paris UTC data. The paper describes in details the algorithms proposed, their calibration, the definition of the barriers implemented, and the performance achieved in different real input data conditions. Once the experimentation of the new technique is consolidated, the authors will report the main conclusions to the EGNOS Project Office with recommendations for potential implementation in future evolutions of EGNOS.