Non-Line-Of-Sight (NLOS) is the most significant error affecting positioning accuracy in urban canyon areas. The traditional approach of excluding NLOS or reducing the weight of NLOS may result in an insufficient number of satellites and introduce distortions in satellite distribution. The latest strategy is to explore the application of NLOS as information in positioning, which eliminates errors, increases effective observations, and improves positioning accuracy and stability. In current NLOS applications, NLOS satellite signals are utilized as virtual Line-Of-Sight signals for positioning after the reflection delay correction, which is only effective if the prior position of the antenna is known and accurate. However, these methods fail once there is an error in the prior position. In this study, we demonstrated that both the reflection delay and design matrix must be rectified if the prior position is inaccurate. Additionally, the Modified Design Matrix approach was proposed. The results of static and dynamic mobile phone observation experiments demonstrate that the new method significantly improves the accuracy in the cross-street direction compared with the conventional weight reduction model and the method in which only the reflection delay is corrected. In the case of a poor satellite distribution or more NLOS satellites, this method also has the advantage of along-street and vertical accuracy. This method applies to scenarios with inaccurate priori positions while maintaining high computational efficiency. Based on this new method, the positioning accuracy in urban canyon areas can be increased, which is especially helpful to improve road-level positioning to lane-level positioning.
Non-Line-Of-Sight (NLOS) is the most significant error affecting positioning accuracy in urban canyon areas. The traditional approach of excluding NLOS or reducing the weight of NLOS may result in an insufficient number of satellites and introduce distortions in satellite distribution. The latest strategy is to explore the application of NLOS as information in positioning, which eliminates errors, increases effective observations, and improves positioning accuracy and stability. In current NLOS applications, NLOS satellite signals are utilized as virtual Line-Of-Sight signals for positioning after the reflection delay correction, which is only effective if the prior position of the antenna is known and accurate. However, these methods fail once there is an error in the prior position. In this study, we demonstrated that both the reflection delay and design matrix must be rectified if the prior position is inaccurate. Additionally, the Modified Design Matrix approach was proposed. The results of static and dynamic mobile phone observation experiments demonstrate that the new method significantly improves the accuracy in the cross-street direction compared with the conventional weight reduction model and the method in which only the reflection delay is corrected. In the case of a poor satellite distribution or more NLOS satellites, this method also has the advantage of along-street and vertical accuracy. This method applies to scenarios with inaccurate priori positions while maintaining high computational efficiency. Based on this new method, the positioning accuracy in urban canyon areas can be increased, which is especially helpful to improve road-level positioning to lane-level positioning.