Three separate applications of estimation problems in satellite navigation and attitude determination and in ionospheric scintillation monitoring have been solved numerically. The first application is magnetometer-based autonomous orbit determination, the second application is attitude determination using a GPS antenna on a turntable, and the third application is ionospheric scintillation monitoring using semi-codeless dual-frequency GPS techniques. As the first application, a magnetometer-based orbit determination batch filter has been improved and tested with real flight data to determine the performance of a low-cost autonomous orbit determination system. The spacecraft?s orbit, magnetometer biases, and correction terms to the Earth?s magnetic field are estimated by this filter. The maximum position error of this filter for a 24-hour batch of ?rsted data is 59.50 km without field model corrections, but is only 2.19 km with 10th order/degree field model corrections. The second application develops the signal processing algorithm for a prototype of a new attitude sensor, one that uses a single GPS antenna mounted on a rotating turntable, and evaluates its accuracy using real data. The goal of this study is to experimentally validate this new concept. The new attitude sensor measures attitude by using the sinusoidal phase modulation of the GPS carrier signal that is caused by turntable rotation. The new sensor system has been demonstrated to work in practice, and its peak attitude error is between 0.4 and 1.9 deg for a 15 cm antenna mounting radius and a rotation rate of 1000 rpm. The third application is to develop optimal semi-codeless algorithms to track weak dual-frequency P(Y) GPS signals during strong ionospheric scintillations and to test them with real and simulated RF data. The goal of this work is to determine whether these new algorithms are less prone to loss of lock during strong scintillations than are other semi-codeless algorithms.
The tracking algorithms use extended Kalman filter (EKF) and smoother methods as part of a semi-codeless technique that performs maximum a posteriori estimation of the W encryption bits of the unknown P(Y) code, which are known to chip at approximately 480 KHz.
The algorithms have been successfully tested with real wide-band dual-frequency GPS data under normal signal strength and simulated scintillation conditions with strong amplitude and phase fluctuations. The C/N0 thresholds above which the Kalman-filter algorithm and the smoother algorithm can still track without losing lock are found based on the computed steady-state standard deviation of the L2 carrier phase error from steady-state covariance analysis. This threshold roughly agrees with the tracking results for the simulation data and can be used to predict the possibility of loss of lock.