Learning-based Distributed Multi-channel Dynamic Access for Cellular Spectrum Sharing of Multiple Operators

Spectrum sharing is one of the ways to overcome the inefficient use of wireless resources. For successful spectrum sharing, dynamic spectrum access based on reinforcement learning has been actively studied. The environment in which mobile network operators (MNOs) share spectrum has not been considered. This paper considers a cellular spectrum sharing scenario where MNOs share the single band in distributive manner. For successful spectrum sharing in the scenario, we propose a learning-based distributed multi-channel dynamic access scheme. The proposed scheme enables multi-channel access dynamically while minimizing interference among MNOs. To evaluate our proposed scheme, we conduct system level simulations in simple network environments. The simulation results indicate that the proposed scheme not only improve the throughput of the network, but also guarantee the fairness among operators.