Deep Neural Network Approach to GNSS Signal Acquisition

This paper investigates 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. Acknowledging that the latter are optimally designed and extensively tested, it is also agreed that when the nominal models do not hold the performance of the receiver might degrade. Particularly, we investigate the use of data-driven models in the signal acquisition stage of the receiver by addressing a classification problem from Cross Ambiguity Function (CAF) delay/Doppler maps. A discussion on the training of such models and future perspectives is provided. The detection results in nominal situations are then compared to the theoretical bound in the receiver operating characteristic (ROC) plots.