The ability of reinforcement learning (RL)-based convolutional neural network (CNN) to mitigate multipath signals for opportunistic navigation with downlink 5G signals is assessed. The CNN uses inputs from the autocorrelation function (ACF) to learn the errors in the code phase estimates. A ray tracing algorithm is used to produce high fidelity training data that could model the dynamics between the line of sight (LOS) component and the non-line of sight (NLOS) components. Experimental results on a ground vehicle navigating with 5G signals for 902 m in a multipath-rich environment are presented, demonstrating that the proposed RL-CNN achieved a position root-mean squared error (RMSE) of 14.7 m compared to 20.6 m with a conventional delay-locked loop (DLL).
Multipath Mitigation of 5G Signals via Reinforcement Learning for Navigation in Urban Environments
2022-06-01
622291 byte
Conference paper
Electronic Resource
English
Performance Assessment of Navigation Signals in Realistic Multipath Environments
German Aerospace Center (DLR) | 2008
|Markovian Channel Modeling for Multipath Mitigation in Navigation Receivers
German Aerospace Center (DLR) | 2008
|