Robust GNSS Position Estimation Using Graph Optimization Based Vector Tracking

Abstract

Vector tracking (VT) is proposed and demonstrated as a superior method to obtain more robust navigation solutions. In VT, instead of individually tracking the signals, VT accomplishes signal tracking and navigation solutions estimation through a central navigation filter, mutual aiding between the channels is realized in this manner. Commonly, a Kalman Filter (KF) is employed as the center navigation filter to estimate the navigation solutions, the estimated navigation solutions are then fed back to calculate the signal tracking parameters. However, KF works in a recursive manner, relationships between the current state and all the past states are ignored, which might degrade the estimation of the navigation solutions. In this paper, we proposed a Graph Optimization (GO) based on VT. GO optimized the state estimation utilizing all the past information instead of KF, the state transformation, and measurement model were all added to the GO as the constraints to optimize the state estimation. An experiment was carried out for assessing the proposed GO-VT, statistical analysis of the navigation solutions and the corresponding comparisons demonstrated the superiority of the proposed GO-VT method.