Time-Frequency Analysis and Neural Network-Based Multipath Mitigation Method for Precise Positioning

Multipath errors caused by signal reflection are challenging to correct when compared to other errors in radio navigation. Specifically, multipath error for indoor positioning is a major error because various materials that surround the environment can cause errors through reflected signals. This study focuses on mitigating multipath in single-transmitter-based indoor positioning, which employs a transmitter emitting multi-channel signals and uses carrier-phase measurements for precise positioning with low noise. Owing to the difficulty in accurately modelling reflected signals mathematically, this study proposes a neural-network-based technique to detect and categorize multipath-distorted signals using short-time Fourier transform (STFT) to extract features in the time-frequency domain. The STFT magnitudes in the spectrogram are used to train and validate the neural networks that detect and isolate multipath-distorted measurements to minimize positioning errors. This paper also presents a novel method that employs a trained neural network to classify the most contaminated signals by considering a single-transmitter-based positioning system with an antenna array. Experiments conducted under challenging indoor conditions demonstrate an approximately 30% reduction in the positioning error after applying the proposed multipath mitigation methods.