Non-Uniform User Distribution in Non-Terrestrial Networks with Application to User Scheduling

Non-Terrestrial Networks (NTNs) have recently seen a surge in interest from both the industry and consumers, attracted by the promise of a global coverage with drastically increased performance. Such coverage is achieved through the use of satellite mega-constellations which field of view can span from a few tens to hundreds of kilometers within each beam. Usually, the performance of NTN systems is assessed assuming uniformly distributed users. This approach, which has been extremely effective in the past when the satellite user densities were limited, neglects the human tendency to settle in clusters. Thus, a strong approximation on the real distribution of users is introduced when the subscriber density starts approaching the terrestrial one. This leads to results that do not reflect the actual system performance and, hence, might lead to suboptimal designs. In this paper, we therefore propose a Cluster-Process-based nonuniform distribution that better reflects the actual user density on ground, thus providing an improved tool for the design of NTN systems. To this aim, we propose a general procedure to extract statistics from population count datasets and tailor the user distribution model to the coverage area. To show the effectiveness and importance of considering non-uniform distributions, we empirically analyze the user scheduling performance in a B5G Low Earth Orbit High Throughput Satellite system. The per-cluster sum-rate capacity and user throughput achieved under non-uniform and uniform distribution are compared, highlighting the inaccuracies that the uniform assumption can introduce.