Abstract:

In a satellite communication system deployed with massive MIMO, when the number of users far exceeds the number of beams of the satellite, it is still difficult to satisfy the delay requirements of the satellite communication system even if the number of RF chains is increased, i.e., the number of concurrently serving users is increased.To address the above problems, it proposed a dynamic beam hopping (BH) scheme based on user grouping for communication service coverage.The proposed method utilized the sparsity of the angular domain channel distribution to adaptively divide user groups accorded to the beam direction where the user energy is maximum.The number of co-channel user groups was constrained by power and interference constraints, limiting the reduction of communication traffic delay, as power limitations on the satellite make the number of RF chains limited and increased the number of simultaneously served user groups increased the interference between user groups.So the BH scheme with beam cluster grouping was further proposed, when the beam clusters reduced the user delay by increased the number of users served at the same moment.Since the above problem was non-convex, it adopt a data-driven determinantal point process (DPP) learning method to simplify the beam scheduling in BH scheme.Simulation results showed that the dynamic BH algorithm based on user grouping could effectively reduce the average user delay by more than 52% compared with the single-beam scheme, also verifies the versatility of the DPP learning.

Key words: LEO satellite, beam-hopping technology, resource allocation, determinantal point process learning