심층 강화 학습과 협상 해법을 이용한 인공위성 빔 호핑 대역폭 할당 기법

초록

This thesis presents a method that improves the data throughput and fairness of a beam hopping satellite system. The goal is to adjustthe bandwidths allocated to each chosen beam while using beam hopping technique in satellite systems with limited wireless resources. To achieve this goal, this thesis proposes using Deep Q-Network(DQN ), which is a type of deep reinforcement learning, and IteratedKalai-Smorodinsky-Nash Compromise(IKSNC) bargaining solution, which is a type of bargaining solution in game theory. The test resultsshow that the method proposed in this thesis provides higher throughput and better fairness than other methods.

키워드