NLOS Multipath Classification of GNSS Signal Correlation Output Using Machine Learning.

Taro Suzuki,Yoshiharu Amano

doi:10.3390/s21072503

Abstract

This paper proposes a method for detecting non-line-of-sight (NLOS) multipath, which causes large positioning errors in a global navigation satellite system (GNSS). We use GNSS signal correlation output, which is the most primitive GNSS signal processing output, to detect NLOS multipath based on machine learning. The shape of the multi-correlator outputs is distorted due to the NLOS multipath. The features of the shape of the multi-correlator are used to discriminate the NLOS multipath. We implement two supervised learning methods, a support vector machine (SVM) and a neural network (NN), and compare their performance. In addition, we also propose an automated method of collecting training data for LOS and NLOS signals of machine learning. The evaluation of the proposed NLOS detection method in an urban environment confirmed that NN was better than SVM, and 97.7% of NLOS signals were correctly discriminated.

Highlights

IntroductionDespite the increase in the number of positioning satellites, large positioning errors occur suddenly in urban environments
We proposed a novel NLOS multipath detection method that can employ machine learning techniques to enhance the global navigation satellite system (GNSS) positioning performance in urban environments, where NLOS multipath signals lead to major positioning errors
When the signal-to-noise ratio (SNR) of the input signal is lower than the individual threshold, it is judged as an NLOS signal

Summary

Introduction

Despite the increase in the number of positioning satellites, large positioning errors occur suddenly in urban environments. This positioning error is caused by multipath signals, where GNSS signals are reflected or diffracted by buildings and other objects [1]. GNSS antennas receive direct signals and reflected or diffracted signals simultaneously, which is called LOS multipath. The reflected or diffracted signals affect the correlation process of the direct signal, arriving at the antenna first and deteriorate the ranging accuracy of GNSS. This ranging error is called LOS multipath error. Owing to an innovation of the signal correlation method, the LOS multipath error within the GNSS pseudorange is only a few meters at most, and the maximum error has an upper limit

Methods

Results

Discussion

Conclusion