Improving Smartphone GNSS Positioning Accuracy Using Contextual Information

With the widespread adoption of smartphones, location-based services have become increasingly important. Consequently, accurate and reliable global satellite navigation system positioning on smartphones has become essential. However, achieving accurate positioning in urban areas remains challenging because of the inherent limitations of smartphones and severe multipath effects. To address this issue, this study proposes two methods to improve positioning accuracy using contextual information. First, an environmental context indicator was used to refine the C/N0-based observation covariance model. Second, normalized C/N0 and code-pseudorange residuals were used to detect non-line-of-sight satellites and adjust the observation covariance. Experiments were conducted in both open and urban areas, and performance was evaluated using circular error probable (CEP) and distance root mean square (DRMS). The experimental results showed that, in open areas, the proposed method achieved submeter to decimeter-level horizontal accuracy and precision. In semi-urban areas, CEP95, CEP50, and DRMS decreased by approximately 8, 2, and 4 m, respectively. In urban canyons, CEP95, CEP50, and DRMS decreased by approximately 15, 2, and 5 m, respectively.