Location-based User Scheduling through Graph Coloring for Cell-Free MIMO NTN Systems

Non-Terrestrial Networks (NTN) have recently grown in popularity, with Low Earth Orbit (LEO) mega-constellations delivering broadband services to households world-wide. Under the cell-free MIMO paradigm in full frequency reuse systems, meeting the increasing traffic demand requires complex scheduling and digital beamforming algorithms to minimize the excessive co-channel interference. In this paper, we present a user scheduling algorithm for LEO-based B5G NTNs with reduced computational complexity that does not require knowledge of downlink (DL) Channel State Information (CSI). In our proposed method, we perform user grouping by first solving the Minimum Clique Cover (MCC) problem on an inter-user distance adjacency matrix, avoiding the need for DL pilots for CSI estimation. We approach MCC as a graph coloring task on the complementary graph, using the DSatur algorithm to minimize the computational complexity of the scheduler. Users within the same group are served via space-division multiplexing by means of feed space digital beamforming, inferring the users' channel vectors from their position. System-level analysis show that the proposed algorithm, named Distance-based MCC DSatur (D-MCC-DSatur), achieves a per-cluster sum-rate capacity gain of up to 1.8% with respect to the distance-based Multiple Antenna Downlink User Scheduling (D-MADOC) algorithm.