User Scheduling with Hybrid Beamforming Architectures for SDM-based LEO Non-Terrestrial Networks
Abstract
This paper addresses user scheduling and hybrid beamforming design for Space Division Multiplexing (SDM) Low Earth Orbit (LEO) satellite systems. A location-based radio resource management framework is proposed, where user grouping relies on geographical separation rather than instantaneous channel state information, through a graph-based scheduler. The latter is evaluated jointly with four beamforming architectures: full digital, fully-connected hybrid, partially-connected hybrid, and matched filtering. Results show that the beamforming architecture is the dominant performance factor. Fully-connected hybrid beamforming achieves the highest Signal-to-Interference-plus-Noise Ratio (SINR), owing to its smaller average cluster size, while full digital beamforming leads in throughput and sum-rate capacity thanks to more efficient resource utilization. Partially-connected hybrid beamforming offers a favorable cost-performance trade-off, and matched filtering consistently yields the lowest performance. The proposed framework confirms the benefit of jointly designing location-based scheduling and beamforming to enhance spectral efficiency in LEO satellite communications.