Neural Networks algorithms for Improving Delay Lock Loops Performances for Navigation Satellite Systems in urban environment

1 Abstract In this paper a novel code tracking algorithm is proposed which can be used in GNSS receivers. Classical code tracking approaches consider AWGN channels as an hypothesis for designing theoretically optional trackers. However, this hypothesis does not hold in multipath channels. The method proposed , based on neural networks, provides results that outperform classical DLL trackers. One of the most important performance metrics of a navigation system is localization accuracy. Within GPS-like systems, location can be estimated on the bases of hyperbolic trilateration (i.e. by calculating position on the intersection of 3 circles). The accuracy of location estimation is related with the precision by which time of arrival (TOA) is estimated of incoming satellite signals. Synchronization modules are devoted to perform TOA estimation within GPS receivers. The synchronization process is done by match the incoming signal with a locally generated signal replica. This process is divided into two steps: coarse alignment (first acquisition) and fine synchronization (tracking) [1]. The result of the first step is to obtain an estimation of TOA within limited difference between the incoming signal and the local replica. The second step performs and maintains in time a more accurate alignment between the incoming coded signal and the code replica at the receiver. A fine synchronization is essential because an error in TOA estimation causes a proportional error in the estimation of the user's localization. Unfortunately the electromagnetic wave can be reflected by the presence of obstacles, such as mountains, buildings, objects or atmosphere discontinuities. This fact, better known as multipath fading, creates replicas that can be summed up to the direct LOS (Line of Sight) path. Multipath causes distortions in the received signal. In fact, each path is characterized by a different length, so the spectral components of the received