Dynamic resource allocations based on Q-learning for D2D communication in cellular networks

In order to solve the problem of spectrum and power allocation for D2D communication in cellular networks when the prior knowledge is not available, a method based on machine learning is proposed in this paper. Q-learning, which is one of the most important algorithms in machine learning, is proposed to solve the radio resource management in underlay mode to find the optimal strategy in time series. In this mode, Q-learning is used to finish the channel assignment and the power allocation at the same time. And the simulation results show that greater system capacity can be achieved through the method proposed in this paper.