Objectives Mobile phone positioning is a widely used approach for navigation, which has broad application prospects. The global navigation satellite system (GNSS) is widely used in smartphone positioning, but its performance can be degraded in urban canyons because of signal reflections or blockages. Shadow matching (SM) based on the three-dimensional (3D) city model can effectively improve positioning accuracy in cross-street direction. However, variation of signal-to-noise ratio (SNR) is large using smartphone for GNSS signal reception while the conventional method fails to distinguish neighboring streets, hence, greater cross-street errors.
Methods This paper proposes an improved SM method together with SNR smoothing implemented in smartphones to improve the positioning accuracy in urban canyons. Firstly, a SNR smoothing method based on low-pass filter is proposed to mitigate the variation, and further to improve the correctness and stability of the visibility classification based on observations. On this basis, an improved SM, namely cluster shadow matching (Cluster-SM), is proposed, in which, the effective candidate points are clustered related to their locations.
Results Experiment results showed that SNR smoothing reduces error rate of the SNR classification from 5% -30% to 0% -20%, while the implementation of optimization Cluster-SM based on SNR filtering significantly improve the GNSS positioning accuracy from 19.4 m to 2.1 m in dynamic test, compared to conventional method.
Conclusions This shows the effectiveness of the novel approach in increasing positioning accuracy with the ability to distinguish neighboring streets, which provides opportunities to implement the smartphones in location-based services applications, pedestrian positioning or vehicle navigation which requires a higher positioning accuracy.
DownLoad: