Deep Learning of GNSS Signal Detection

Global Navigation Satellite Systems (GNSS) is the de facto technology for Position, Navigation, and Timing (PNT) applications when it is available. GNSS relies on one or more satellite constellations that transmit ranging signals, which a receiver can use to self-localize. Signal acquisition is a crucial step in GNSS receivers, which is typically solved by maximizing the so-called Cross Ambiguity Function (CAF) resulting from a hypothesis testing problem. The CAF is a two-dimensional function that is related to the correlation between the received signal and a local code replica for every possible delay/Doppler pair, which is then maximized for signal detection and coarse synchronization. The outcome of this statistical process decides whether the signal from a particular satellite is present or absent in the received signal, as well as provides a rough estimate of its associated code delay and Doppler frequency, if present. Critical infrastructures and safety-critical applications, such as in the context of connected and autonomous vehicles (CAV heavily rely on GNSS for PNT purposes and as a consequence, these services are vulnerable to degradation and deliberate attacks on GNSS, for which this thesis proposes data-driven approaches to address those vulnerabilities. With the increased popularity of artificial intelligence, machine learning, and deep learning begin to play an important role in adapting traditional algorithms in a variety of disciplines, especially when it comes to estimation and classification tasks where remarkable results have been demonstrated. The contributions of this work include (i) the use of data-driven models, popular in the machine learning literature, as an alternative to well-engineered signal processing blocks used in state-of-the-art GNSS receivers to perform signal acquisition. Particularly, with and without spoofing attacks. The CAFs are fed to a data-driven classifier that outputs binary class posteriors, which are used in a Bayesian hypothesis test to statistically decide the presence or absence of a legitimate GNSS signal or a spoofing signal; (ii) to increase the detection accuracy with less computational complexity by leveraging data-driven methods. The versatility and computational affordability of the proposed methods are addressed by splitting the CAF into smaller overlapping sections, which are fed to a bank of parallel classifiers whose probabilistic results are optimally fused to provide a so-called probability ratio map from which signal detection is inferred; (iii) the research shows how non-coherent integration schemes are enabled through optimal data fusion, to increase the resulting classifier accuracy and (iv) to estimate the number and parameters of both the legitimate and spoofing signals using machine learning clustering algorithms such as Gaussian Mixture Model (GMM). This thesis provides simulation results showing that the proposed data-driven method outperforms current CAF maximization strategies, enabling enhanced acquisition at lower carrier-to-noise density ratios.