Edge intelligence assisted resource management for satellite communication

Satellite communication has been seen as a vital part of the sixth generation communication, which greatly extends network coverage. In satellite communication, resource management is a key problem attracting many research interests. However, previous study mainly focuses on throughput improvement via power allocation and spectrum assignment and the proposed approaches are mostly model-based and dedicated to specific problem structures. Fortunately, with the trend of edge intelligence, complex resource management problems can be efficiently resolved in a model-free manner. In this paper, a joint beam activation, user-beam association and time resource allocation approach is proposed. The core idea is using stochastic learning at the ground station to identify active user-link beams to meet user rate demand. In addition, the convergence, optimality and complexity of our proposal are rigorously discussed. By simulation, it is shown that the rate goal of most of the users can be met and meanwhile satellite energy is saved owing to much less active beams.