Evaluation of Neural Network-Based Multipath Mitigation Approach for the GNSS Receivers

This article describes a neural network-based multipath estimation algorithm and the compensation approach for the GNSS receivers. A functional diagram of a typical analog front-end, as well as the digital signal processing stage is presented. Multipath error for GNSS receivers is described, presenting the influence of the phenomenon on the correlation peak. Existing mitigation approaches are evaluated, including state-of-the-art parametric algorithms and hardware, highlighting the existing advantages and drawbacks. A new algorithm is presented, targeted at the 5-point complex correlators, implemented in the existing BBP2 high-precision SDR-oriented GNSS ASIC. Pre-processing correction step and compensation are described. Modelling result and theoretical pseudorange multipath error envelope is provided.