|M.Sc Student||Leiter Noam|
|Subject||Optimizing Cluster Flight for Space-based Geolocation|
|Department||Department of Aerospace Engineering||Supervisor||Professor Pinchas Gurfil|
|Full Thesis text|
Space based geolocation with a cluster of satellites performing time difference of arrival measurements (TDOAs) can provide accurate tracking of a Mars rover, a redundant navigation system in a jammed GNNS environment, or a cost-effective system for autonomously locating distress signals. Numerous estimation schemes have been used for geolocation with TDOAs. In this study, it is shown that former methods are highly sensitive to the initial estimate due to the nonlinear nature of the hyperbolic TDOAs. This initial ambiguity is solved in the current research by first applying algebraic methods, borrowed from computer graphics, to find all possible initial positions.
These positions are the real intersection points of an oblate sphere, modeling Earth's surface and two hyperboloid surfaces which are realizations of two TDOA measurements. We adapt former methods and derive the effective quadratic intersection (EQI) method for obtaining initial estimates from noisy measurement with quadratic realizations.
Then, additional measurements are used to perform multiple model estimation techniques to statistically distinguish between the true target and fictitious initial positions. The same procedure developed here can be used in general nonlinear estimation problems with initial ambiguity. In addition, tight estimation bounds are derived for the multiple model estimator. These estimation bounds are then used to optimize the cluster trajectories resulting in optimal positioning performance in a simulation scenario.