Mobility pricing based on GNSS is bound to become a significant tool for both traffic control and infrastructure funding, compared to geofencing road-side technologies. But GNSS positioning in cities remains challenging, because of the poor satellite visibility and of the occurrence of many satellite signal reflections (multipath), that in the best case combine with the direct signal, but, when the direct signal is obscured, create significant biases of several tens of meter in the pseudo-range measurements and consequently on the computed position. The method proposed here is based on a simplified 3D model of buildings at both sides of streets, named 'Urban Trench' for that reason, which can benefit in the process of satellite visibility check and GNSS positioning. Homogeneous street sections are described with three parameters: width, building height and transversal position of the vehicle. These parameters allow to identify the satellites in direct visibility and to compute the estimated additional range brought by the reflections. It has been proved that, in some environments such as the Grands Boulevards in Paris, the position computed by a least squares method can be improved by a factor of 70% after application of our visibility check and multipath bias correction, without degrading the availability.