Location-Assisted Graph-Based User Scheduling in Multi-User MIMO LEO NTN Systems

This paper addresses user clustering and scheduling for multi-user MIMO low Earth orbit nonterrestrial network systems in full frequency reuse. Since the number of on-ground user terminals is usually much higher than the number of on-board LEO satellite antennas, user scheduling becomes a fundamental task. We accomplish user scheduling by grouping users into clusters. Users within the same cluster are served by the satellite at the same time by means of space division multiplexing via location-based feed space digital beamforming. Each cluster is then assigned to a distinct time slot and served by means of time division multiplexing. Given the full frequency reuse nature of the system, we design user scheduling algorithms with the goal of maximizing the average per-user throughput while minimizing the co-channel interference and preserving fairness among users. To this aim, we propose in this paper (a) a distance-based iterative graph-based scheduler based on the maximum clique approach and (b) a distance-based implementation of the multiple antenna downlink orthogonal user clustering algorithm. For both these schedulers, the great circle distance between the users is employed as a dissimilarity metric to compute the user adjacency matrix, avoiding the need for the transmission of downlink pilots for channel state information estimation. To further validate our analysis, the proposed approaches are compared with (a) channel state information-based graph and maximum clique approach and (b) original multiple antenna downlink orthogonal user clustering algorithm. Extensive simulations assess the achievable per-user throughput and signal-to-noise plus interference ratio achieved by the proposed schedulers, highlighting the impact that the distance-based metric has on the system performance. This study provides valuable insights into the effective use of user scheduling algorithms.