•  
  •  
 

Corresponding Author

Elhabiby, Mohamed

Subject Area

Civil and Environmental Engineering

Article Type

Original Study

Abstract

Global Navigation Satellite Systems (GNSSs) are used in many navigation and positioning applications. Unfortunately, a GNSS signal may suffer from some errors, such as cycle slips, which deteriorate the positioning solution. A cycle slip is defined as a sudden jump by an integer number of cycles in the GNSS carrier phase observations. Signal blockage or/and high troposphere activities are the most common causes for GNSSs’ cycle slips. Therefore, cycle slips should be detected and corrected to determine reliable positioning estimations. A new approach for cycle slip detection and repair is proposed based on a master-rover phase-difference with a deep Long Short-Term Memory (LSTM) neural network model; our SlipNet model can classify defective data where a cycle slip has occurred and then predict the exact epoch where the cycle slip(s) occurred. The proposed SlipNet network would be the first end-to-end learning framework to solve the integer ambiguity problem in GNSS measurements with high performance results, %99.7 detection and localization accuracy, and 0.045 MAE for slip estimation and recovery. These results are on par with the latest classical cycle slip detection methods of cycle slip detection and correction.

Keywords

GNSS; Cycle Slip; Slip Net; LSTM; Autoencoder

Download

Share

COinS