Golden Section Search-Based GNSS Multipath Mitigation for Enhanced Urban Positioning

Multipath has long been recognized as a major source of error in Global Navigation Satellite System (GNSS) urban positioning. Since no external assistance is required, baseband signal processing is a preferred solution for multipath mitigation. Although many multipath mitigation techniques at the baseband signal processing level have been proposed, the case where the power of the reflected signals is greater than that of the direct-path (DP) signal has received little attention. Moreover, the mitigation for short-delay multipath remains a challenge. To address this gap, this paper analyzes the characteristics of the composite autocorrelation function (ACF) with multiple reflected signals and considers the case where the power of the reflected signals is greater than the power of the DP signal. Based on this analysis, a code discrimination technique is proposed that utilizes multiple rounds of golden section (GS) search to directly estimate the code phase of the DP signal. Simulation results show that the proposed algorithm can effectively mitigate the multipath effect even in cases where the power of the reflected signals is greater than that of the DP signal. On Spirent short-delay multipath datasets, the proposed algorithm reduces 3D positioning root mean square error (RMSE) by over 22% compared to the narrow correlator, confirming its superior capability for mitigating short-delay multipath. For real urban datasets, the proposed algorithm outperforms both the narrow correlator and the classical multipath estimating delay-lock loop (MEDLL), achieving average reductions of 17.8% and 36.8% in 3D positioning RMSE respectively, which validates its robustness and practical utility in complex urban environments.