Deep Learning Multipath Error Estimation for 3DMA-Based Positioning Algorithm in High Dynamics Environments

GNSS positioning in high dynamics multipath environments faces two major challenges. The first challenge is coming from the high dynamics conditions because of the difficulty of the tracking modules to cope with the sudden and fast changes in the receiver’s speed and acceleration. This takes the tracking modules out of their steady state and can lead them to lose lock on the signals. The second challenge is the multipath problem, which remains the main threat to the application of the GNSS technology in civilian urban environments. This is because multipath signals are responsible for high positioning errors and they are complex to detect and mitigate. Each challenge is a problem by itself but combining the two challenges has a more detrimental effect on the operation of a GNSS receiver. This paper addresses the two aforementioned challenges when combined together. Three algorithms are proposed. The first algorithm aims at enhancing the stability of the tracking modules by developing a novel reacquisition algorithm for high dynamics multipath conditions, which is called ReAcq-HD-MP. The second algorithm aims at estimating the errors in the code delays due to multipath using a novel deep neural network algorithm, which is called DNN-MP. The input of the DNN-MP is taken from the output of the tracking module after applying the ReAcq-HD-MP algorithm. The third algorithm integrates the DNN-MP algorithm with a 3D mapping aided (3DMA)-based algorithm, called Optimized Position Estimation (OPE). The integrated algorithm is called OPE-DNN. The proposed algorithms are tested in high dynamics multipath environment. The results show that the ReAcq-HD-MP and the OPE-DNN algorithms can enhance the positioning accuracy up to 66% and 86% compared to a conventional algorithm, respectively. In addition, the OPE-DNN algorithm is able to provide a consistent and a high accuracy position estimation.