Optimal Cycle-Slip Detection Algorithm for GPS/GNSS Preprocessing Using Three Linear Combinations of Moderate-to-Low-Noise Data

Abstract

Cycle slips are discontinuity events in a receiver’s phase lock on a Global Navigation Satellite System (GNSS) signal. If cycle-slip events are not detected and repaired, the quality of positioning using carrier phase processing suffers. In this paper, we present an algorithm for flagging cycle slips during data preprocessing. This new algorithm is part of a new GNSS software that replaces a legacy software at the National Geodetic Survey (NGS). It uses the Hatch–Melbourne–Wübbena (HMW), ionosphere-free and geometry-free linear combinations to detect cycle slips. The use of the three observables in parallel helps improve sensitivity to cycle slips and overcome the challenge of detecting cycle slips of equal size and same sign on both/multiple frequencies. Different detection threshold values were evaluated in a well-controlled environment using GNSS data acquired from continuously operating reference stations (CORS) networks and artificially simulated cycle slips that were introduced into the datasets. The goal of evaluation was to determine the optimal values for cycle-slip detection parameters. In the absence of formal confidence limits, the results of the study show that, in a worst-case scenario involving small slips of magnitudes ranging from 1 to 3 cycles, there is a range of parameter values that resulted in at least 97% success rate without any false detections. Based on these and other findings, we provide a set of optimal values from which the results are an improvement of at least 34% in cycle-slip detection, compared to results from the legacy algorithm’s settings. The results of the study will support the National Oceanic and Atmospheric Administration’s (NOAA) GNSS services and tools for the public, such as the Online Positioning User Service (OPUS). The algorithm described in this paper is part of a new GNSS software supporting the National Spatial Reference System (NSRS) modernization efforts at the US National Geodetic Survey (NGS). It replaces an existing algorithm that uses default settings for cycle-slip detection parameters. The NSRS modernization has wide implications including for GPS/GNSS tools and services such as the Online Positioning User Service (OPUS), which is widely used by surveyors, engineers, and geospatial professionals. The results in this paper show that the new algorithm’s parameters provide better sensitivity to cycle slips for the type of data used in the study. The performance is based on the assumption that the GPS/GNSS data are collected using survey-grade and/or geodetic-grade receiver types such as those used in professional geodetic and land surveys, and CORS networks. Thus, the data quality is of moderate to low noise characteristics, and results show an improvement of at least 34% in the rate of cycle-slip detection. The results are based on a 30-s sampling rate, with the general knowledge that, typically, the lower the data sampling rate the harder the cycle-slip detections. (The 30-s sampling rate is the lower sampling rate compared to 15-, 5-, and 1-s sampling rates.)