Satellite Orbit Determination Using Payload-Collected Observation Data

The algorithm developed in this research provides a novel way of preparing the space industry for the future. The rapid rise in small satellite deployment and miniaturization of communication technology will require a cheaper, leaner and more efficient way of tracking those satellites. Using the already-collected timing data from the payload observations means that no additional on-board equipment or processing will be required and that it could even be applied to existing missions, as well. This thesis provides a solid foundation and development analysis to support this new way of using satellite payload data. It shows how combining even the most basic form of observed data (the time-of-access and location) can provide deeper and more insightful knowledge. Each cost function used and explored combines this data in a different manner and, therefore, provides a different kind of insight pertaining to a different aspect of the satellite behaviour. This, combined with the power of machine learning, has proven to be an effective way of determining the position and velocity of the satellite with strong potential for future development in real-world Earth-observation or, perhaps even, interplanetary missions.