Satellite user grouping and multi-beam patterning for generalized rate-splitting multiple access

Abstract

With an exponential increase of data traffic, satellite communication systems are being required to make an efficient use of limited frequency resources, and several multiple access schemes have been studied to achieve a higher sum-rate than an orthogonal multiple access (OMA) on a broadcasting (BC) channel such as satellite downlink. Encompassing spatial division multiple access (SDMA) and non-orthogonal multiple access (NOMA) on the whole, generalized rate-splitting multiple access (RSMA) suffers from high complexity of optimizing the number of common streams corresponding to all user combinations and a brute-force procedure to find the best decoding order. To tackle this problem, in this paper, we preemptively optimize the location of the satellite beams, convert the precoding matrix design of generalized RSMA into a low complex power allocation problem. In addition, a heuristic decoding order is set instead of the exhaustive search by using the intensity of the pre-manipulated beams. We then propose a user grouping method that reduces the number of common streams by using a sequential search algorithm based on the position of users and a solution of leakage-based beam patterning from non-convex programming regarding the location of each beam. Finally, simulation results show that the novel framework of generalized RSMA in the satellite BC channel has a multiplexing gain up to 48 percent compared to NOMA and 1-layer RSMA with fixed cells.