Change Detection Techniques for GNSS Signal-Level Integrity

The provision of accurate positioning is becoming essential to our modern society. One of the main reasons is the great success and ease of use of Global Navigation Satellite Systems (GNSSs which has led to an unprecedented amount of GNSS-based applications. In particular, the current trend shows that a new era of GNSS-based applications and services is emerging. These applications are the so-called critical applications, in which the physical safety of users may be in danger due to a miss-performance of the positioning system. These applications have very stringent requirements in terms of integrity. Integrity is a measure of reliability and trust that can be placed on the information provided by the system. Integrity algorithms were originally designed for civil aviation in the 1980s. Unfortunately, GNSS-based critical applications are often associated with terrestrial environments and original integrity algorithms usually fail. The main impairments are due to local effects such as interference, multipath or spoofing, which are assumed to be controlled in civil aviation but they are not in terrestrial environments. Thus, a new methodology for integrity is necessary in order to detect local effects and provide the additional level of integrity needed for GNSS-based critical applications; the so-called signal-level integrity. This thesis investigates novel detection algorithms with the aim of providing a new generation of integrity techniques in GNSS. For this purpose, the framework of Statistical Change Detection (SCD) is considered. This framework is of particular interest because its optimal criterion target the temporal dimension. This is an indispensable requirement for critical applications, in which a prompt detection is necessary. This approach is in contrast to classical detection schemes in which the probability of missing a detection is minimized, thus disregarding the temporal dimension. Thereby, the first part of this dissertation deals with the study of the field of SCD, including both Quickest Change Detection (QCD) and Transient Change Detection (TCD). Novel contributions are provided in the field of TCD, including the finite moving average solution and its statistical characterization. Numerical results show the superiority of our contributions. Finally, to conclude our study of SCD we compare it with classical detection schemes under the same mathematical framework. This comparison shows the appropriateness of SCD when dealing with timely detections. The main contribution of this thesis is provided in the second part by applying the framework of SCD to threat detection and integrity in GNSS. To this end, we first investigate several properties of the received GNSS signal that may be useful for local threat detection. This leads us to move a step forward in the field of threat detection by proposing a novel QCD-based framework to interference and multipath detection in GNSS. QCD aims at detecting a threat as soon as possible, which is of interest to monitor whether a threat is present or not so that a prompt alert may be raised. Nonetheless, for integrity purposes a bounded delay is desirable, and it is here where TCD is of interest. For this reason, a novel TCD-based framework is considered for both multipath detection and integrity algorithms in GNSS, thus leading to the provision of signal-level integrity. Finally, the second part of this dissertation also investigates the practical implementation of signal-level and current integrity algorithms. A notable improvement is shown by the proposed TCD-based solutions considered in this thesis with respect to current solutions. In the last part of the thesis, the goal is to validate the proposed threat detectors and signal-level integrity algorithm using real GNSS signals. For this purpose, a MATLAB-based software is developed to process gathered GNSS signals with the aim of detecting the presence of interference and multipath. Real signal gathered in the context of an EC-funded research project is processed to show the results of the implemented detectors. Two data collection campaigns are processed to validate the behavior of the proposed detectors in real working conditions. The work is completed with the validation of the navigation accuracy of the proposed signal-level integrity algorithm. The results obtained in a realistic scenario show the improvement of the accuracy and integrity by using the proposed solution for signal-level integrity, with respect to current integrity algorithms. Furthermore, the proposed solution is shown to have real-time processing capabilities, thus being very attractive to improve current integrity algorithms and easily implementable in mass-market receivers.