Objectives Under the influence of satellite signals reflected by buildings and multipath effect, global navigation satellite system (GNSS) pseudo-distance single point positioning error can be up to tens of meters in complex urban area, in order to improve GNSS positioning accuracy in urban area, a shadow matching algorithm is proposed. Shadow matching algorithm uses 3D city model for satellite visibility prediction and location matching, which can reduce the pseudo-distance single point positioning error and improve the positioning accuracy of the smartphone. However, the basic algorithm does not fully consider the influence of different observation environments, and there are still some problems such as fixed threshold scoring, low satellite utilization rate, and unreasonable selection of high-score candidate locations, and there is still a large space for improvement of its positioning accuracy.

Methods We propose a shadow matching localization method considering local optimality. First, the unimproved shadow matching algorithm is used to determine the local candidate regions in the global candidate regions, then the continuous signal-to-noise raito scoring formula is used to score the local candidate regions. Finally, the scores of the local candidate locations are normalized, and the high score candidate locations are selected for weighted average processing to obtain the global and local optimal positioning results.

Results Experimental results show that in the complex urban area, compared with the pseudo-distance single point positioning and the basic shadow matching algorithm, the localization accuracy of the locally optimal shadow matching algorithm in the static experiment is improved by about 75% and 50%, respectively, and the absolute positioning accuracy reaches about 5 m. In the dynamic experiment, the cross-street positioning accuracy of the shadow matching algorithm considering the local optimal reaches 1.36 m.

Conclusions The shadow matching algorithm considering the local optimal can significantly improve the GNSS positioning accuracy of mobile phones in complex urban environments, providing a reference for vehicle and pedestrian navigation services.