Weak GNSS Signal Navigation for Lunar Exploration Missions
Publikation aus Digital
Proceedings of the 28th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2015), Tampa, Florida,, 2015
Weak-signal GNSS navigation could potentially increase the robustness, flexibility and autonomy of the navigation architectures for future lunar exploration missions. GNSS reception during the various phases of such a mission suffers from very low signal levels, partial visibility of the GNSS sources and unfavourable geometry, actually making use of the spill over of the beams around the Earth mask. The objective of a recent ESA study was to evaluate the challenges of such a navigation solution using GPS and future Galileo signals with carrier to noise density ratio levels as low as 10 to 15 dBHz. The investigated mission phases included transfer orbit, low lunar orbits, lunar ascent and descent as well as surface operation and navigation at the Lagrange points L1 and L2. The suggested GNSS receiver architecture is a software based snapshot receiver with limited ground station aiding to help with or even to substitute the information from the navigation data messages. Such architecture has been proven to provide a navigation solution during all phases of a lunar exploration mission. Integration with external sensors is required during descent/ascent and low lunar orbits: a loose coupling to an INS/radar altimeter is considered and evaluated for these phases. All of these solutions strongly build on the on-board navigation propagator, including a kinematic model to provide navigation data during partial signal outages. The availability of GNSS signals, in terms of geometrical line-of-sight, signal strength, and Doppler shift, has been computed for the different mission scenarios taking into account GPS and Galileo constellations. Interestingly, such a computation revealed that the mission date has a slight effect on the performance. In fact, the inclination of the lunar orbit with respect to the Earth's equator (to which GNSS sources orbits are anchored) showed a non-negligible variation in the performance. As far as they concern the user spacecraft bound to the Moon, different antenna accommodations were analysed and a single, steerable high gain antenna was considered as the baseline for the study. In fact, as the distance from the Earth increases and the signals become weaker, even the relatively narrow beam opening of a 13 dB gain antenna efficiently includes all the sources in its field of view. During the frame of the ESA study we developed a snapshot receiver simulation in Matlab to implement the receiver acquisition stage. Starting with the GPS L1 C/A signal, we then concentrated on the data-less pilot signals of the Galileo E1 C and Galileo E5 a-Q/b-Q services in order to achieve longer coherent integration times. The suggested receiver concept relies on aiding from either ground station TT&C channels to provide
coarse position aiding or utilises the position updates from the INS navigation part that follows the signal acquisition stage and pseudoranges computation. The INS part delivers updates of the spacecraft dynamic state, with update rates well above 1 Hz, while the GNSS acquisition performs initialisation and position updates from the GNSS. In the signal acquisition block we use block averaging preprocessing for coherent signal integration followed by non-coherent combining. The signal acquisition performance is tested using synthetic input signals with random content navigation messages. These simulations suggest that a receiver sensitivity down to about 10 dBHz can be reached when using the Galileo pilot signals and a coherent integration
time of 500 ms. The choice of the snapshot architecture allows for the use of very high integration time during the acquisition while avoiding problems with stability and continuity of the usual tracking stage at a very low carrier to noise density ratio. At the same time, the chosen architecture leads to some constraints on clock stabilityand the availability of an initial coarse position. The accuracy of the position/velocity/time solution was tested with the simulation of ground station aiding. It is assumed that in a future mission an internal representation of the spacecraft's trajectory and GNSS constellation could further alleviate the amount of data required from ground station support. The propagator simulation implements the fusion of the sensor data coming from the PVT solution and from aerospace grade sensors (INS, radar Doppler altimeter 'RDA') through the use of an Extended Kalman Filter (EKF). The other main task of the propagator is to provide a valid position/velocity solution during upcoming periods of GNSS signal outage (at the Lagrange points and behind the Moon, where the latter is the worst scenario in terms of time elapsed from last GNSS signal update to its re-acquisition). The propagator further reduces the uncertainty associated with the PVT solution, introducing internal knowledge of the trajectory. The performance of the EKF is evaluated by comparing the uncertainties (sigmas) associated to GNSS PVT with those evaluated at the exit of the propagator stage. The software receiver and propagator combination was chosen for a first evaluation approach to an aided space based navigation solution. With the addition of hardware sensors like IMU, Doppler radar altimeter and other sensors, we expect that a high degree of autonomy of navigation during lunar exploration missions can be achieved. With the exploitation of weak signals from existing and future GNSS, the robustness of the navigation architecture of future lunar exploration missions could be improved for all mission phases. Above all, GNSS-based navigation helps, by means of the improved autonomy, to strongly limit the cost of the lunar mission tracking. This aspect will become especially important in view of the emerging low-thrust, electricallypropelled missions with mission durations in the order of several months.