An Enhanced Local Optimization Algorithm for GNSS Shadow Matching in Mobile Phones

In the context of mobile phones, the local optimal global navigation satellite systems (GNSS) shadow matching algorithm, which is based on the urban three-dimensional model, can effectively reduce the error of GNSS pseudo-range single-point positioning. However, the positioning accuracy of this algorithm is susceptible to noise, and its continuous signal-to-noise ratio (SNR) scoring method does not fully exploit the probability density and probability distribution information contained in the SNR. Therefore, this paper proposes two improvements for the local optimal shadow matching algorithm: (1) utilizing low-pass filtering to filter SNR, thereby reducing the influence of noise on the algorithm and (2) introducing a probability-based SNR scoring method to fully leverage the probability density and probability distribution information of SNR. In dynamic single-point positioning, the improved algorithm attains an absolute positioning accuracy of up to 3 m, representing a decimeter-level enhancement over the original algorithm. Experiments confirm that using the SNR statistical information of non-line of sight (NLOS) and line-of-sight (LOS) as prior information results in better positioning accuracy compared to when this information is distorted by multipath effects. Additionally, to address the issue of high time complexity in the shadow matching algorithm, especially when considering local optima, this paper presents a scheme to simplify the algorithm’s flow, reducing its time complexity by approximately 75%.