Detection and Classification of GNSS Signal Distortions Based on Quadratic Discriminant Analysis

Abstract

Satellite signal distortions, or so called “evil waveforms” (EWFs), may cause severe distortions of the cross-correlation function in receivers. Undetected EWFs could result in large range errors and threaten the integrity of Global Navigation Satellite System (GNSS). Analog distortion and digital distortion are two classical types of signal distortions. With the advent of modernized GNSS signals, more failure types are considered to cover potential EWFs of Binary offset carrier (BOC) modulated signals, such as code-only distortion and subcarrier-only distortion. Different failure types affect the correlation functions and the tracking precision in different ways. Therefore, it is useful to identify the failure type of a detected distortion. Conventional multi-correlator method can detect signal distortions; however, it is infeasible to identify the failure types. We developed a novel multi-correlator method based on Quadratic Discriminant Analysis (QDA). The QDA-based method also uses correlation-domain detection metrics but performs distortion type classification by supervised learning algorithm. Experimentally measured results on Beidou B1C data signals are presented, which show the effectiveness and robustness of proposed method. Compared with conventional multi-correlator method, the QDA-based signal quality monitoring (SQM) method shows better performance on detecting EWFs and provides an extra capability to identify the failure types accurately.