Quality control of the kinematic positioning into GNSS networks

GNSS (Global Navigation Satellite System) positioning is nowadays a common practice, not only in academia but also in the professional world, thanks to the development of several types of instruments and the decreasing costs of the receivers and antennae. Today, considering also the new constellations that are emerging, it is possible to affirm that GNSS positioning can be assessed almost everywhere: therefore, it is no longer necessary to ask whether satellite positioning is possible but rather how precise it can be. The main focus and goal of this thesis is to suggest new and innovative methodologies to control the quality of the positioning, focusing attention on the network positioning both in real time and post processing. Different types of instruments will be considered, in terms of both cost and performance: in fact, geodetic, GIS and mass-market receivers will be considered when used for Network Real-Time Kinematic (NRTK) or differential positioning. This thesis consists of five enlightening chapters. Chapter 1 includes a brief introduction to the subject as well as encompassing the research objectives and this outline. Chapter 2 outlines the quality control of geodetic and GIS receivers in a network of permanent stations, considering different types of networks, as a function of inter-station distances. Chapter 3 focuses attention on the quality of ambiguity fixing, considering an innovative method based on the creation of a tool that is able to predict a wrong geodetic receiver fix when it is used in a CORSs (Continuously Operating Reference Stations) network. Chapter 4 describes the quality control of the positioning when mass-market instruments (receivers and antennae) were used both in a network of permanent stations and as a network itself. Chapter 5 gives the conclusions and recommendations for further work. Two appendices conclude this work: the first concerns the Kalman filter in geodesy and its importance for real-time positioning, considering both the classical configuration and other versions (e.g. the extended Kalman filter and unscented Kalman filter); the second analyses more deeply the neural network concepts that are mentioned briefly in Chapter 3.