Simulating Urban Satellite-Based Vehicular Networks: Focus on Maps, Building Heights, or Vehicle Density?

Low-Earth Orbit constellations are one of the recent advancements that could provide a fresh perspective on Internet connectivity. They are appealing because they provide global coverage, they offer low round-trip-time, and they can grow to tens of thousands satellites to increase the offered capacity. However, simulating large urban areas is challenging due to the scalability limitations of the simulators. We present the results of a simulation campaign in which we improve the current state-of-the-art vehicular network simulator, Veins, in its satellite-based variant (Space Veins). We introduce a precise three-dimensional analysis that allows us to use detailed city maps, which in turn allows us to use realistic traffic patterns. We show that the parameters that most impact the results of the simulators are two: the assumptions on the building height and the positioning of the vehicles in the map. Results are extremely sensitive to these factors, underscoring the need for more open data and standardized, repeatable scenarios that can facilitate better research evaluation in this field.