LAUSR

Global navigation satellite system (GNSS) positioning in urban areas does not currently provide accurate and stable performance because surrounding buildings can block and reflect satellite signals. However, if we can determine the environment in which the receiver is located, appropriate positioning can be applied. For example, GNSS real-time kinematic and 3D-mapping-aided GNSS (3DMA GNSS) are used for positioning in open sky and urban areas, respectively. Thus, the context awareness of the GNSS receiver is important. In fact, an urban canyon can be further categorized into different levels based on sky visibility. We propose an innovative algorithm based on this categorization, which can provide information on surrounding buildings and give an estimation of sky visibility from raw GNSS measurements. This idea was inspired by the use of low-orbit satellite data for remote sensing applications. The recent development of multi-GNSS has led to a notable increase in the number of navigation satellites. Crucially, the visibility of satellites and the blockage of line-of-sight satellite signals are representative of the surrounding environment. The visibility of satellites can be classified by machine learning techniques, and an accurate classification can afford an estimation that is close to the real-sky visibility, as derived from a 3D building model and ground truth location. To assess the sensitivity of our proposed sky visibility estimation algorithm, we simulate different classification accuracies to investigate their effect on the performance of the algorithm.